Hey Common Folks!

Last edition, we walked through pre-training — the general-education phase where an AI model reads the internet and slowly learns how the world is described in text. We ended with a promise: come back and show you what the day-to-day classroom actually looks like. This is that edition.

Because pre-training raises a question that sits right underneath it: how does a data scientist know whether the model has genuinely learned something, or just memorized the examples it was shown?

The answer comes down to a simple but crucial idea: you need to test a model on data it has never seen before. And to do that correctly, you have to split your data into three very specific groups from the start.

That’s what training, validation, and test sets are all about.

The Exam Analogy

Think back to how you prepared for an important exam in school.

Your textbook came with practice problems at the end of each chapter. You worked through them, checked your answers, and kept reviewing the material you got wrong. These are your training set — examples you learn from, with feedback.

Then your teacher handed out a mock exam. You hadn’t seen these specific questions before, but you used them to figure out where you still had gaps. Based on how you did, you went back and studied those weak areas. These are your validation set — a checkpoint to help you tune your preparation.

Finally: the real exam. Brand new questions, no second chances, no adjustments afterward. This is the only true measure of whether you actually learned. This is your test set.

Here’s the critical rule: you cannot use the same data for all three purposes. If you train on the same examples you test on, you’re not measuring learning — you’re measuring memorization. It’s like scoring a student on the exact practice problems they studied. A perfect score tells you nothing.

The Three Sets Explained

Training Set

This is the largest portion of your data — usually around 70-80%. The model learns from this data. It sees thousands or millions of examples, makes predictions, gets corrected, and adjusts its parameters. This is where all the actual learning happens.

The model sees the training data over and over again across multiple training runs. It’s allowed to learn from its mistakes here.

Validation Set

This is typically 10-15% of your data. The model never trains on this data — it only uses it to check its progress.

During training, you periodically pause and run the model on the validation set. If the model’s accuracy on training data is climbing but its accuracy on the validation set is flat or dropping, you’re watching overfitting happen in real time. The model is memorizing training examples instead of learning generalizable patterns.

The validation set also helps you make decisions: Should you train for more epochs? Should you adjust the learning rate? Should you use a simpler model? Every tuning decision gets made based on how the model performs here — not on the training set.

Test Set

This is the smallest portion — maybe 10-15% — and it gets used exactly once: at the very end, after all training and tuning is complete.

The test set is your honest final grade. The model has never seen these examples. You haven’t made any decisions based on them. This is the only clean measurement of how the model will perform on truly new data.

If you use the test set during tuning — making adjustments based on test performance — you’ve contaminated it. You’ve effectively turned it into a validation set, and now you have no honest evaluation left.

What Happens When You Get This Wrong

Skipping or contaminating the test set is more common than you’d think, and the consequences can be serious.

In the late 2010s and early 2020s, AI medical-imaging papers regularly landed in major journals reporting accuracy figures in the 90s — sometimes claiming they could spot tumors, predict COVID, or flag heart disease nearly perfectly. A 2021 review published in Nature Machine Intelligence looked at over 2,000 such papers on COVID detection alone and found that not a single one was clinically useful. The most common reason? Data leakage. The “test set” had quietly seen the same patients, the same scanners, or the same preprocessing as the training set. The exam was rigged.

When hospitals tried to deploy these tools on truly new patients, accuracy collapsed.

This isn’t just a technical problem. When AI is making recommendations about someone’s cancer diagnosis, loan application, or parole hearing, the gap between “performed well in testing” and “performs well on real people” has real consequences.

By 2026, the response has started to bite. The FDA now expects AI medical-device submissions to demonstrate performance on prospective test data — patients enrolled after the model was already locked in — because retrospective test sets had proven too easy to contaminate, even by accident. A properly held-out test set is the closest thing we have to a clean, honest audit before deployment, and regulators have stopped trusting anything weaker.

An Important Twist: Cross-Validation

Sometimes your dataset is small enough that you can’t afford to lock away 15% as a test set — you need every example to train on.

In those cases, data scientists use a technique called cross-validation. Instead of one fixed split, you divide the data into five or ten equal “folds.” You train on nine of them and validate on the tenth. Then you rotate — train on a different nine, validate on the remaining one. Repeat until every fold has been the validation set exactly once.

This gives you a more reliable estimate of how the model will perform on new data without sacrificing training examples. It’s slower, but it’s smart when data is scarce.

Where You See This in Practice

Every responsible AI system uses this three-set approach, even if the terminology differs.

Google testing a new search ranking algorithm trains on historical search data, validates on recent data to tune the approach, and evaluates on a final held-out period before pushing anything to production.

Netflix’s recommendation models train on older viewing history, validate on more recent history to catch drift and tune parameters, and measure final performance on the most recent week of data they haven’t touched.

Even the foundation models behind ChatGPT, Claude, and Gemini get evaluated on held-out benchmarks the model has never seen during pre-training — though, as we’ll see in a future edition, even those held-out exams are getting harder to keep clean as models scale.

The principle is always the same: learn on one thing, tune on another, and only trust what you measure on something completely separate from both.

The Takeaway

The training set teaches the model. The validation set helps you tune it. The test set honestly evaluates it.

Confuse these roles — or skip any of them — and you lose the ability to know whether your AI actually works.

This is why, when you hear “AI achieved 95% accuracy,” the right question isn’t “how high is the number?” It’s “what data was it tested on, and was that data truly held out?” The number only means something if the exam was fair.

Every AI tool you trust — from medical imaging to fraud detection to the spam filter keeping your inbox clean — depends on data scientists getting this split right before it ever reaches you.

Coming Up

We’ve now seen the clean version of how AI learns — labeled examples, a tidy split, an honest final exam. But that whole picture assumes someone already labeled the data for you. In the real world, most data shows up messy and unlabeled. Nobody has sat down to tag every tweet, every photo, or every customer record.

So what happens when you take the answer key away? Can AI still learn anything useful from a pile of raw, unsorted data? Next edition: how AI learns without a teacher.

Was this helpful? Did you know this was how AI gets tested? Reply and let us know what you want us to explain next.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

Yesterday we covered tokens — how AI breaks your prompt into the small pieces it can actually read. Now we answer the question that opens up right underneath that one: where did AI learn what those tokens actually mean in the first place?

The answer is Pre-training. It is the “P” in GPT (Generative Pre-trained Transformer). It is the phase where a Foundation Model learns 99% of everything it knows, and it is the reason today’s AI is so shockingly capable, and occasionally so shockingly confident about things it shouldn’t be.

What is Pre-training?

Think of it as General Education. Before you become a doctor, an accountant, or a coder, you first have to learn the alphabet, how to read, how to do basic math, and how the world works. You don’t start kindergarten by performing brain surgery.

Pre-training is the AI version of that long, broad foundation phase. The model spends an enormous amount of time learning general knowledge before anyone ever asks it to do a specific job.

The Problem: Learning from Scratch is Hard

Why do we even need pre-training? Why can’t we just teach an AI to answer customer support emails directly?

Because deep learning models are data hungry. If you want to train a model from scratch to recognize a cat, you need thousands of photos of cats. But not just photos — you need humans to manually label them: “This is a cat,” “This is a dog.” That labeling process is slow, expensive, and tedious.

And if you tried to teach a computer to understand English by only showing it customer support emails, it would fail. It wouldn’t know what a verb is, what “angry” sounds like, or even how to structure a sentence.

You can’t shortcut the foundation. You have to build it.

The Solution: Predict What Comes Next

Pre-training flips the problem on its head. Instead of teaching the AI a specific job immediately, we let it loose on a massive amount of data — the internet, books, papers, code — with one beautifully simple goal: predict what comes next.

The AI reads billions of sentences and tries to guess the next word over and over. After “Hello,” the next word is often “World” or “There.” After “Once upon a,” it is almost always “time.” Through trillions of these tiny prediction games, the model gradually picks up grammar, language patterns, factual knowledge, and slang — the broad shape of how the world is described in text.

The beautiful part: nobody has to label anything. The next word in the sentence is itself the answer.

It learns the general features of the world first.

• In images: it learns what edges, circles, and shapes look like before it learns what a “face” is.

• In text: it learns how language works before it learns how to write a poem about your dog.

The whole philosophy comes from a simple idea in transfer learning: don’t reinvent the wheel. If a model already exists that understands the basics, build on top of that knowledge instead of starting from zero.

The Analogy: The Medical Student

To understand pre-training versus what comes after it, think of a medical student.

1. Pre-training (Medical School): The student spends years reading textbooks and learning anatomy, biology, and chemistry. They aren’t treating patients yet. They are just building a massive foundation of general knowledge. They know a little bit about everything.

2. Specialty Training (Residency): Now that student goes to a hospital to specialize, maybe in cardiology, maybe in surgery. They take all that general knowledge and focus it on one specific task.

A pre-trained model like GPT-5, Claude, or Gemini is the medical school graduate. It has read the library. It is smart and broad, but it hasn’t specialized in your company’s data, your customer’s tone, or your industry’s jargon yet. That comes later.

Why is This a Game Changer?

Before pre-training became the standard, if you wanted to build an AI to translate languages, you needed a massive labeled dataset of English-to-French sentences. If you didn’t have that data, you were stuck. Every new task meant starting from zero.

With pre-training:

1. You need less data later. Because the model already knows English, you only need to show it a small number of examples of what you specifically want it to do for it to catch on.

2. You skip the hardest part. A pre-trained foundation already exists. You just point it at the specific job you care about.

A note on scale, because this matters for understanding why pre-training is such a big deal in 2026: back when this technique first took off around 2018–2020, pre-training a small language model was a days-or-weeks project on a modest cluster. Today’s frontier models — GPT-5, Claude, Gemini — take months to pre-train, run on tens of thousands of GPUs, and cost hundreds of millions of dollars in compute alone. Pre-training has gotten bigger, not smaller, as AI has scaled.

The Takeaway

Pre-training is the heavy lifting. It is the process of creating a Foundation Model — a knowledgeable generalist that can later be pointed at almost any task.

• It is the bridge between a blank computer brain and one that has read the library.

• It is the difference between teaching a baby to write a novel (impossible) and asking a college graduate to write a novel (possible).

• It is the layer underneath every modern AI you use — ChatGPT, Claude, Gemini, Copilot — built long before you ever typed a single prompt.

One honest caveat for 2026: pre-training is the foundation, but it is not the whole story anymore. Modern AI also goes through additional specialty training on top of pre-training, and that is where today’s models learn to be helpful, careful, and reasoning-capable. We’ll dig into that in the upcoming articles.

When you use ChatGPT or Claude, you are talking to a model that has already finished its general education. It has read the library. Now it is ready to work for you.

Coming Up

You now know what pre-training is — the general-education phase where AI learns how the world is described in text. But how does the AI actually do the learning? What does the day-to-day classroom look like? Next, we’ll walk through the study, practice, and test loop — the actual mechanics of how a blank-slate model goes from random guesses to meaningful answers.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

Over the last two articles, we covered what a prompt is and how to write a good one. That was about how to talk to AI. Now we look at the other side of that conversation: how AI actually reads what you typed.

If you have ever looked at the pricing page for ChatGPT or Claude, or seen an error message saying “Token limit exceeded,” you have probably scratched your head. Why don’t they just count words? Why this fancy term “Token”?

It turns out, computers don’t read the way we do. To an AI, a Token is the fundamental unit of reality.

What is a Token?

A token is a chunk of text. It can be a whole word (”apple”), part of a word (”smart” + “phones”), a piece of punctuation, or even a single character.

Think of tokens as the atoms of language. Just as a molecule of water is made up of atoms (Hydrogen and Oxygen), a sentence is made up of tokens.

The Analogy: The Lego Castle

Imagine you have a beautiful Lego castle (a sentence).

• For Humans: We look at the castle and say, “That’s a castle.” We read the whole word.

• For AI: The AI looks at the individual plastic bricks used to build it.

Sometimes, a single brick is a whole window (a whole word like “apple”). Other times, to build a long wall (a complex word like “smartphones”), you need two bricks: “smart” and “phones.”

The AI doesn’t see the castle; it sees a pile of bricks. It processes those bricks one by one to understand the structure.

Why Not Just Use Words?

A fair question: “Why break ‘smartphones’ into two tokens? Why not just treat it as one word?”

Because computers don’t understand English. They understand numbers.

To teach an AI language, every piece of text first has to be converted into a list of numbers (called token IDs), which the AI later turns into rich mathematical representations called embeddings.

• If we assigned a unique number to every single word in the English language, the list would be infinite and unmanageable. Every name, every typo, every made-up word would need its own slot.

• By breaking complex words into smaller chunks (tokens), the AI can understand words it has never seen before by recognizing their parts.

If the AI knows “smart” and it knows “phones,” it can understand “smartphones” without needing a separate definition for it.

How Does It Work? (The Tokenization Process)

Before your prompt hits the AI, a process called Tokenization happens.

1. Input: You type “I love AI.”

2. Chopping: The tokenizer chops this up. In a modern GPT-style tokenizer, it looks roughly like: ["I", " love", " AI", "."] — four tokens, including the leading spaces. (Real tokenizers are picky like that, and they bake the spacing into the tokens themselves.)

3. Numbering: The system assigns a specific ID number to each chunk. Modern tokenizers have a vocabulary of tens of thousands to a few hundred thousand possible tokens, so real IDs are usually large numbers. We’ll use small ones below for illustration: [40, 1842, 16124, 13].

4. Processing: The AI receives that list of numbers and asks itself the one question it knows how to answer — what number should come next? It picks the most likely next number based on the patterns it learned during training (we walked through that pattern-prediction in How AI Actually Learns). That predicted number is converted back into a piece of text, and then the AI does it again — one token at a time — until your full answer is built.

Want to see this live? Free tools like OpenAI’s tokenizer page or Anthropic’s token-counting API let you paste any text and watch how it gets split. It’s worth doing once — you will never look at a long prompt the same way again.

The “Currency” of AI

Why should you care about tokens? Because in the world of Generative AI,
Tokens = Money.

When companies like OpenAI or Anthropic charge you, they don’t charge per question. They charge per token.

• Input Tokens: what you type, paste, or upload into the chat.

• Output Tokens: what the AI writes back.

A useful 2026 reality check: output tokens almost always cost more than input tokens — typically 3 to 5 times more — because generating a careful answer is harder for the model than reading one. So a long, rambling AI response costs you more than a short, precise one. Brevity in your prompt and in the format you ask for is a real lever on cost.

Roughly speaking, 1,000 tokens is about 750 words. So if you ask the AI to summarize a 50-page document, you are “spending” tokens to feed that document into the model, and spending more tokens to get the summary back out.

The good news for 2026: frontier models like Claude and Gemini now support context windows in the hundreds of thousands to over a million tokens. A 500-page novel can fit in a single prompt today — something that was impossible just two years ago. Tokens still cost money, but the ceiling on how much you can hand the AI in one shot has gone way up.

The Takeaway

A Token is simply a chunk of text.

• It is the bridge between human language and machine numbers.

• It is the unit used to measure the size of the AI’s memory (the context window).

• It is the unit used to calculate the cost of using the AI.

Understanding tokens explains why your long prompt sometimes gets cut off, why running a massive analysis costs a few dollars instead of a few cents, and why brevity in your prompts is a quietly powerful skill.

Coming Up

You now know how AI breaks your prompt into tokens. But where did it learn what those tokens actually mean in the first place? Next, we’ll dig into how AI gets its basic education — the massive pre-training phase where a blank-slate model reads a huge slice of the internet and starts to make sense of the world.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

In our last article, we covered what a prompt is — the bridge between human intent and machine output. Before that, we met the Large Language Model — the engine inside ChatGPT, Claude, and Gemini — and a few articles back we built up to it through Neural Networks.

You now know what a prompt is. This article is about the part that actually changes your results: how to write a good one.

Have you ever asked ChatGPT a question, gotten a generic, unhelpful answer, then rephrased it and suddenly gotten something brilliant? That’s not luck. That’s the difference between a bad prompt and a good one.

The skill of crafting better prompts has a name: Prompt Engineering.

Why “Engineering”?

Don’t let the word scare you. This isn’t about writing code or building bridges.

Prompt Engineering is about communication skills — talking to a machine in a way it understands best. It’s the art of being specific, structured, and strategic with your instructions.

The Analogy: Back to Our New Hire

Remember the new hire we’ve been working with? The one who read the entire internet before day one?

They have all the knowledge in the world. But they are extremely literal and have zero context about your specific life or business.

Bad Manager (You):
“Write an email to the client.”

The new hire (The AI):
Panic. Which client? Good news or bad? Formal or casual? They guess and write something generic and robotic.

Prompt Engineer (You):
“Act as a senior sales manager. Write a polite but firm email to ‘Client X’ regarding their overdue payment of $500. Keep it under 100 words. Don’t use emojis.”

The new hire (The AI):
Understood. They write exactly what you need because you gave them the Role, the Context, and the Constraints.

Prompt Engineering is just being a really good manager to your AI new hire.

Why Does This Work? (The Technical Bit)

Remember how LLMs work? They predict the next word based on patterns they’ve seen.

When you write a detailed prompt, you’re filling the AI’s Context Window with specific patterns. The AI looks at those patterns and generates text that fits that specific context.

When you go a step further and include examples of the input-output pattern you want, you’re using something researchers call In-Context Learning — the AI picks up the pattern from the examples in your prompt without any retraining. We’ll see this in action in technique #3.

Vague stage = vague performance.
Specific stage = specific performance.

The Five Techniques That Actually Work

You don’t need a degree for this. Master these five approaches.

1. Role Prompting (The “Act As” Hack)

Tell the AI who it’s supposed to be.

• Instead of: “Explain quantum physics.”

• Try: “You are a kindergarten teacher. Explain quantum physics to a 5-year-old using only examples they’d understand.”

This sets the tone, complexity, and approach immediately.

2. Be Specific About Output

Don’t leave format to chance.

• Instead of: “Give me marketing ideas.”

• Try: “Give me 5 marketing ideas for a local bakery. Format as a numbered list. Each idea should be under 20 words.”

The more specific your constraints, the more useful the output.

3. Few-Shot Prompting (Show, Don’t Just Tell)

Sometimes instructions aren’t enough. Show examples.

If you want the AI to convert slang to formal English, demonstrate:

• Input: “Sup bro?” → Output: “Hello, how are you?”

• Input: “Gotta run.” → Output: “I must leave now.”

• Input: “No way!” → Output: ...

The AI sees the pattern and continues it perfectly. That’s In-Context Learning at work.

4. Chain of Thought (Let’s Think Step-by-Step)

For complex problems, add: “Let’s think step-by-step.”

This forces the AI to show its reasoning before giving the final answer. Accuracy on math and logic problems goes up because the AI can’t just guess — it has to work through the problem.

A note for 2026: this was the original trick that started the whole “reasoning model” era. Today’s reasoning-tier models from OpenAI, Anthropic, and Google already do this internally before they answer you. So the phrase matters less for the latest models, but it still helps with smaller or older ones, and it is the foundation everything else is built on.

5. Give Context and Background

The AI doesn’t know your situation. Tell it.

• Instead of: “Write a resignation letter.”

• Try: “I’ve worked at this company for 3 years. My boss has been supportive. I’m leaving for a better opportunity, not because I’m unhappy. Write a professional resignation letter that maintains the relationship.”

Context changes everything.

Want to Practice This?

A friend of mine, Don Barger, built a free tool at ripen.donbarger.com around a clean little framework he calls RIPEN:

• Role — who or what should the AI act as?

• Input — what information are you giving it?

• Process — what steps should it take to get to the answer?

• Example — show it what good output looks like.

• Notes — tone, constraints, guidelines, anything else.

It’s the same territory we just walked through, repackaged as a five-letter mnemonic that’s easy to recall the moment you’re actually typing a prompt. If you want a structured place to drill these techniques into muscle memory — whether you’re writing a one-off prompt or building a chatbot’s personality — ripen.donbarger.com is a clean starting point.

Common Mistakes to Avoid

Being too vague: “Help me with my project” tells the AI nothing.

Mashing unrelated tasks into one prompt: Back when ChatGPT first launched in 2022, even simple multi-task prompts like “write me an email, also summarize this, and list action items” would come out confused or messy. Today’s models handle that combo easily — so this advice has evolved, not disappeared. The underlying principle still holds in two specific cases: when the tasks have conflicting tones or audiences (a casual Slack message and a formal client email about the same news), and when you want to iterate on one piece of the answer without regenerating the rest. For serious work, separating prompts still gives you cleaner output and tighter control.

Not iterating: Your first prompt rarely gives perfect results. Treat it as a conversation — refine and improve.

Forgetting constraints: Without limits, AI tends toward verbose, generic responses. Add word counts, formats, and restrictions.

Simple Examples to start with:

Writing

• Before: “Write a blog post about productivity.”

• After: “Write a 500-word blog post about productivity for remote workers. Use a conversational tone. Include 3 actionable tips. Start with a relatable scenario.”

Research

• Before: “Tell me about climate change.”

• After: “Summarize the top 3 causes of climate change in simple terms a high school student would understand. Use bullet points. Keep it under 200 words.”

Code

• Before: “Write Python code.”

• After: “Write a Python function that takes a list of numbers and returns the average. Include comments explaining each step. Handle the case of an empty list.”

The Limitations (Keeping It Real)

Prompt Engineering has limits.

It can’t fix bad models: If the underlying AI is weak, no prompt will save it.

It’s not magic: Some tasks are genuinely hard for AI. Better prompts help, but they don’t make AI capable of everything.

It takes practice: You’ll write bad prompts before you write good ones. That’s normal.

The Takeaway

Prompt Engineering isn’t a “technical” skill — it’s a clarity skill.

• Vague prompts = average results.

• Specific, structured prompts with examples = excellent results.

Writing code is still a real, valuable craft. The engineers who can also articulate clearly, in plain language, with the right context, are the ones getting the most out of AI right now. Clarity is no longer a soft skill. It is a multiplier on top of every other skill you already have.

Coming Up

You now know how to ask. But before the AI can even read your prompt, it has to break it into pieces. Next, we’ll explore Tokens — why AI counts your words differently than you do, why a four-letter word can sometimes count as two tokens, and why that quietly affects what AI costs you and what it remembers.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

In our last article, we met the Large Language Model — the engine inside ChatGPT, Claude, and Gemini. Before that, we decoded GPT, and before that we introduced Foundation Models as the general-purpose AI brains powering today’s tools.

We know these AI models are incredibly powerful. But a Ferrari is useless if you don’t know how to drive it.

How do you actually talk to this super-brain? You don’t use Python code or binary zeros and ones. You use a Prompt.

And here’s the part most people miss: the same AI can give you a brilliant answer or a useless one based on nothing but how you asked. Same model. Same knowledge. Completely different output. That gap is what this article is about.

What is a Prompt?

In simple terms, a prompt is whatever you give the AI to get it to do something.

• When you type a question into ChatGPT, that text is the prompt.

• When you upload a screenshot of a spreadsheet and ask for key insights, that combination is the prompt.

• When you paste a 50-page PDF and ask for a summary, the document plus the instruction is the prompt.

• When you speak to AI through your phone, your voice is the prompt.

A prompt is the bridge between human intent and machine output. Everything you want, you have to express through it. The AI cannot read your mind. It can only work with what you give it.

The Analogy: Back to Our New Hire

Remember the new hire we met last article? The one who read the entire internet before day one?

They have all the knowledge in the world. But they don’t know you yet. They don’t know your style, your boss, your deadlines, your preferences. On day one, they need instructions for every single task, and the quality of your instructions determines the quality of their work.

• Bad Prompt: “Write an email.”

  • The new hire thinks: To whom? About what? Angry tone? Professional? Long? Short?

  • The result: A generic, useless draft.

• Good Prompt: “Write a polite email to my boss asking for two days of sick leave next week. Keep it under 50 words and don’t sound demanding.”

  • The new hire thinks: Got it. Topic, tone, recipient, length — all clear.

  • The result: A perfect, ready-to-send email.

The AI is that new hire. It has the capability to do almost anything, but it relies heavily on your instructions to know what to do, how to do it, and who it’s doing it for.

Why One Word Can Change Everything

You might have heard the term Prompt Engineering. It sounds fancy, but it just means “the art of asking correctly.”

A fair question to raise: in 2026, with ChatGPT, Claude, and Gemini so much more capable than they were a few years ago, does this skill still matter? The honest answer is yes, and the bar has moved. Early AI models needed near-magical phrasing to produce anything useful at all. Today’s models are far more forgiving — they’ll give you something even from a sloppy prompt. But the gap between something useful and exactly what you need still comes down to how you asked.

LLMs are sensitive to wording. Changing a single word in your prompt can completely change the answer you get back. Three quick examples:

One word changes the audience.

• “Explain gravity.” → a physics textbook definition.

• “Explain gravity to a 5-year-old.” → a story about falling apples.

One word changes the tone.

• “Rewrite this email.” → the AI picks a tone for you, maybe the wrong one.

• “Rewrite this email more politely.” → it keeps your meaning but softens the edges.

One word changes the format.

• “Summarize this article.” → a paragraph.

• “Summarize this article in bullet points.” → a scannable list.

Same AI. Same underlying knowledge. Completely different outputs. From one word of difference.

Why does this happen? Remember from the last article that the LLM predicts one word at a time, always choosing the most likely continuation of what came before. Your prompt is the “what came before.” When you change a word, you change the entire downstream probability of what word should come next. The AI isn’t being tricky. It’s responding exactly as designed to the pattern you gave it.

Same engine, same knowledge, completely different output. All from how you asked.

The Anatomy of a Prompt

Every strong prompt has three basic parts:

1. The Persona: Tell the AI who it is.

   • Example: “You are an expert travel guide” or “You are a Python coding tutor.”

2. The Task: Tell it what to do.

   • Example: “Plan a 3-day trip to Goa.”

3. The Constraints and Format: Tell it how you want the answer.

   • Example: “Give me the answer as a bulleted list” or “Keep it under 100 words.”

Stack all three together and the new hire knows exactly who they are, what they’re doing, and how the answer should look.

Most prompts people type skip one or two of these parts. That’s why most AI answers feel generic.

Bad Prompts vs Good Prompts: Three Real Examples

Here’s what that looks like in practice across three everyday tasks.

Writing

• Bad: “Write a blog post about productivity.”

• Good: “Write a 500-word blog post about productivity tips for remote workers. Use a conversational tone. Include three actionable tips. Start with a relatable scenario.”

Why the second works: The AI now knows the length, audience, tone, structure, and opening style. Five decisions you didn’t have to make yourself.

Research

• Bad: “Tell me about climate change.”

• Good: “Summarize the top three causes of climate change in simple terms a high school student would understand. Use bullet points. Keep it under 200 words.”

Why the second works: You’ve turned an infinite-scope question into a focused, scannable answer at a specific reading level.

Code

• Bad: “Write Python code.”

• Good: “Write a Python function that takes a list of numbers and returns the average. Include comments explaining each step. Handle the case of an empty list.”

Why the second works: The AI now has a specification — inputs, outputs, edge cases, and documentation. It produces code you can actually use.

In all three cases, the AI didn’t get smarter. The prompt did.

The Four Most Common Prompting Mistakes

Before we close, here are the four patterns that produce most of the frustrating AI experiences. Name them once, and you’ll start catching yourself doing them.

1. Being too vague. “Help me with my project” tells the AI nothing. No topic, no format, no outcome.

2. Not giving the AI the context it needs. “Fix this bug” without sharing the code, the error message, and what the code is supposed to do leaves the AI guessing. Being clear about what you want is not the same as giving the AI the raw material to do the job. Modern models are great, but they still can’t read your screen or your mind.

3. Giving no constraints. Without a word limit, audience, or format, the AI defaults to verbose and generic.

4. Expecting perfection on the first try for complex tasks. For simple asks, modern models often nail it immediately. But for anything nuanced — a layered analysis, a specific tone, a tricky coding problem — iteration is part of the skill, not a sign you’re doing it wrong.

The good news: each of these has a clean fix, and there are specific techniques that turn vague prompts into surgically precise ones. We’ll walk through all of them in the next article.

The Takeaway

A prompt is how we program modern computers using natural language instead of code.

• It is the steering wheel of the AI.

• It is the instructions you hand the new hire.

• It is the difference between a useless generic reply and a precise, personalized answer.

The better you get at prompting, the smarter the AI seems to become. Not because the model changes, but because your instructions unlock more of what was always there.

Coming Up

Now you know what a prompt is and why it matters. But knowing what a steering wheel is doesn’t make you a driver. How do you actually get good at this? What separates the people who get brilliant AI answers from the ones who give up after a vague first try? Next, we’ll unpack the five techniques that turn any prompt into a great one — from role-playing to step-by-step thinking. That’s how you graduate from knowing about AI to actually getting what you want from it.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

In our last two articles, we covered Foundation Models, the massive general-purpose AI brains, and GPT, the most famous family in that category. We talked about the Swiss Army Knives of AI and the three-letter recipe (Generative, Pre-trained, Transformer) that cracked modern language AI.

But GPT is just one example of a broader category. Claude, Gemini, Llama, DeepSeek — these are all in the same family. That family is called Large Language Models, or LLMs.

And LLMs are the specific technology powering every AI chatbot you’ve ever used.

The best way to understand one is to think of it as a new employee at your company. A very unusual one.

Meet the New Hire

Imagine your company just hired someone. Before their first day, they did something no human could do: they read every email, every Slack message, every report, every meeting note, every document your company has ever produced. Not just your company, actually. Every company. Every book. Every website. Every Wikipedia article. Every Reddit thread. Every piece of code on GitHub.

They didn’t understand all of it the way you would. They didn’t form opinions or have experiences. But they noticed patterns. They noticed that after “Dear” people usually write a name. That after “quarterly revenue increased” people usually write “by” followed by a percentage. That when someone asks “how do I” the next words are usually a task, followed by step-by-step instructions.

This new hire didn’t memorize facts like a textbook. They memorized how language flows. They can finish anyone’s sentence, in any department, on any topic, because they’ve seen millions of similar sentences before.

That’s an LLM. That’s the whole idea.

The World’s Best Sentence Finisher

At its core, an LLM does one thing: predict the next word. (Technically, it predicts the next token — a small chunk of text that’s usually a word or part of a word. We’ll cover tokens in a future article. For now, “word” is close enough.)

You actually do this too. If I say: “The capital of India is New...”

Your brain instantly completes: “Delhi.”

You didn’t look it up. You’ve seen those words together enough times that the completion is automatic.

Your phone does this too. You type “I am on my...” and your keyboard suggests “way.” Your phone learned this pattern from your text messages.

Now scale that up dramatically.

Your phone looks at the last 3 words to guess the next one. An LLM looks at the last 300,000 words. Your phone learned from your texts. An LLM learned from the entire internet.

Back to our new hire analogy: imagine asking them to finish this sentence: “Based on our Q3 projections and the current market conditions, the board recommends that we...”

Because they’ve read millions of similar corporate emails, they know what typically comes next. Not because they understand finance. Because they’ve seen this pattern thousands of times. They’re pattern-matching at a scale no human could match.

That’s how ChatGPT writes entire paragraphs. One word at a time, each chosen because it’s the most likely continuation of everything before it. Like a new hire who’s so well-read that they can finish any sentence in any department.

How the New Hire Follows Conversations

Here’s where it gets interesting. Early AI systems were terrible at long sentences. Tell them a long story and by the end, they’d forgotten the beginning. Like a new hire who nods along in a meeting but can’t connect what was said in minute one to what’s being discussed in minute thirty.

Then in 2017, researchers at Google published a breakthrough called the Transformer. (The “T” in GPT stands for Transformer. That’s how fundamental this is.)

Transformers gave LLMs a superpower called self-attention. Here’s what that means.

Consider this sentence: “The animal didn’t cross the street because it was too tired.”

What does “it” refer to? The animal or the street?

You know “it” means the animal because the animal is “tired.” Streets don’t get tired.

Before transformers, AI would struggle with this. It read words one by one, left to right, and by the time it got to “it,” the word “animal” was already fading from memory.

Self-attention changed that. Now the LLM looks at all the words in the sentence at once and draws connections between them. When it hits the word “it,” it checks: what does “it” connect to? It sees “tired” and traces back to “animal,” not “street.” It understands the relationship.

Back to our new hire: imagine they’re reading a 50-page email thread where someone says “she approved the budget.” Self-attention is how the new hire traces “she” back to the CFO mentioned 30 emails ago, not the intern mentioned 2 emails ago. They can follow references across long, messy conversations.

This is what lets LLMs understand context, answer follow-up questions, get jokes, and write code that actually makes sense across hundreds of lines. Before transformers, AI was like a new hire reading one word at a time and forgetting the beginning of the email by the end. After transformers, they can hold the entire conversation in their head at once.

Training the New Hire: Three Stages

Building an LLM like ChatGPT or Claude isn’t one step. It’s an onboarding process with three stages. Just like any new hire goes through orientation before they’re ready to talk to customers.

Stage 1: The Reading Phase (Pre-Training)

This is where the new hire reads everything. Terabytes of text. Books, websites, Wikipedia, code, academic papers.

During this phase, the LLM plays a game: we hide a word in a sentence and ask it to guess. If it guesses wrong, it adjusts its internal settings. (Remember the chai analogy? Same loop. Predict, check the error, adjust, repeat. Millions of times.)

Those “internal settings” are called parameters. Think of them as tiny dials. Modern frontier models (GPT-5, Claude 4, Gemini 2) are believed to have trillions of them, though the exact numbers are kept secret. Each dial is like one of the chai recipe settings from our previous article: a small adjustment that slightly changes the output. Together, trillions of tiny dials produce language that sounds remarkably human.

That’s what “Large” means in Large Language Model. Large = trillions of adjustable dials, trained on a massive amount of text.

After pre-training, the new hire knows grammar, facts, writing patterns, coding conventions, and the general structure of human communication. But they’re not helpful yet. Ask them a question and they’ll just keep writing, trying to complete the sentence rather than answer you. They’re like a new hire who’s read the entire company wiki but doesn’t know how to have a normal conversation.

Stage 2: Job Training (Fine-Tuning)

Now we teach the new hire how to actually do their job.

We show them thousands of examples of good conversations:

• Customer asks: “How do I reset my password?”

• Good response: “Here are the steps: go to Settings, click Security...”

• Bad response: “...and also how to reset your username and your profile picture and your billing information and...”

The new hire learns the format: when someone asks a question, give a direct, helpful answer. Don’t ramble. Don’t go off on tangents.

This is fine-tuning. Same new hire, same knowledge from the reading phase, but now they know how to channel it into a helpful conversation instead of an endless monologue.

Stage 3: Performance Reviews (Human Feedback)

The new hire is now having real conversations. But sometimes they’re rude. Sometimes they make things up. Sometimes they give dangerous advice.

So we bring in human reviewers. They chat with the LLM and rate the responses. Helpful and accurate? Thumbs up. Rude, wrong, or harmful? Thumbs down.

The model learns: “Humans prefer it when I’m clear, honest, and careful. They don’t like it when I make things up or lecture them.”

Think of it as ongoing performance reviews. The new hire adjusts their behavior based on what gets positive feedback and what gets complaints.

This last stage is why ChatGPT and Claude feel different from each other even though they’re both LLMs. Different companies hire different reviewers with different values. Same new hire, different management styles, different workplace cultures.

When the New Hire Makes Things Up

Here’s the catch with our new hire. They’re so well-read and so good at sounding confident that sometimes they make things up. And they deliver the fiction with the exact same confidence as the facts.

This is called hallucination.

Remember: the LLM predicts the next most likely word. It doesn’t have a database of facts. It doesn’t “look things up.” It generates text that sounds right based on patterns.

Imagine asking the new hire: “Who designed the Golden Gate Bridge?”

They’ve read enough about bridges and famous people that they might say: “The Golden Gate Bridge was designed by Thomas Edison in 1932.” That sentence is completely wrong. But it sounds like a fact. It has the right structure, the right confidence, the right rhythm of a true statement.

The new hire isn’t lying on purpose. They’re doing what they always do: predicting what the most likely next words would be. And sometimes the most likely-sounding answer isn’t the true answer.

This is the single most important thing to understand about LLMs: they are designed to sound right. Not to be right.

Often they are right, because patterns in language usually reflect reality. But not always. And they’ll never pause and say “Actually, I’m not sure about this.” They’ll just keep predicting the next most confident-sounding word.

Where You Already Use LLMs

You interact with this new hire more than you realize. They’ve been placed in departments all across your digital life:

• ChatGPT, Claude, Gemini: The obvious ones. Every conversation is an LLM predicting one word at a time.

• Email: Gmail’s “Help me write” and Outlook’s Copilot. The new hire is drafting your emails.

• Code: GitHub Copilot suggests code as developers type. The new hire sits next to every programmer.

• Search: Google and Bing now use LLMs to summarize search results instead of just showing links. The new hire reads all the results and writes you a summary.

• Customer service: Many companies have replaced scripted chatbots with LLM-powered support. The new hire handles your complaints now.

The New Hire’s Limitations (Keeping It Real)

Our new hire is impressive. But they have real weaknesses you should know about:

They stopped reading on a specific date. Every LLM has a knowledge cutoff. Ask about yesterday’s news and they genuinely don’t know. It’s like the new hire read everything up to their start date but hasn’t checked the news since. (Some systems work around this by connecting to the internet, but the core model itself is frozen in time.)

They don’t truly understand. They’re the world’s best pattern matcher, not a thinker. They can sound confident while being completely wrong. They don’t “know” anything the way you know your own name. They know what words usually follow other words. That’s it.

They’re expensive to keep around. Every response costs computing power. That’s why advanced AI access isn’t free. Running a trillion dials for every single word in every single response adds up fast.

Their memory has limits. They can only hold so much of the conversation at once. This is called the context window. It’s like the new hire can remember the last hour of conversation clearly but starts forgetting what was said this morning. Long conversations can feel like the AI forgot what you told them earlier, because in a real sense, it did.

The Takeaway

A Large Language Model is the engine powering the AI revolution you’re living through right now.

It’s a new hire who read the entire internet before day one. They predict the next word, one word at a time, with a confidence that makes it look like understanding. They went through reading (pre-training), job training (fine-tuning), and performance reviews (human feedback) to become the helpful assistant you chat with today.

They’re extraordinary at sounding human. They’re terrible at knowing when they’re wrong. And they’re sitting in more of your apps than you probably realized.

Under the hood, it’s the same loop you learned about in our How AI Actually Learns article. Predict, check, adjust, repeat. Just with trillions of dials instead of four chai settings.

Coming Up

Now you know what the engine is. But here’s a subtle truth: the same LLM can give you a brilliant answer or a useless one depending entirely on how you ask. That little box where you type your question? It has a name — the prompt — and the words you put in it are the steering wheel of the entire engine. Next, we’ll break down what a prompt actually is and why it matters more than most people realize.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

AI for Common Folks
Apr 2026

Hey Common Folks!

In our last article on Foundation Models, we talked about the general-purpose brains that power modern AI — the Swiss Army Knives trained to do everything from writing code to drafting emails. Before that, we explored Generative AI, the broad category of AI that creates new content.

Now let’s zoom in on the most famous Foundation Model family of them all: GPT.

You see it everywhere. GPT-4, GPT-5, ChatGPT. But what do those three letters actually stand for? Is it a robot? A company? A magic spell?

Here’s the real story: GPT is not just an acronym. It is three separate breakthroughs in AI that had never been combined at massive scale. OpenAI put them together, and that combination is why modern AI works.

Let’s unpack each one.

What is GPT?

GPT stands for Generative Pre-trained Transformer.

It is a specific type of Large Language Model (LLM) developed by OpenAI. If AI is the broad industry, GPT is a specific product line, like the “iPhone” of AI models.

But here is the part nobody tells you: each of those three words (Generative, Pre-trained, Transformer) represents a problem that AI researchers had been stuck on for decades. GPT is the name for what happened when all three got solved at the same time.

Before GPT: Three Problems AI Couldn’t Crack

To understand why GPT matters, you have to understand what AI looked like before it existed.

For most of AI’s history (roughly the 1950s through the 2010s), researchers were stuck on three problems simultaneously:

1. AI could classify, but it couldn’t create. It could tell you if an email was spam, but it couldn’t write an email.

2. AI had to be trained from scratch for every task. Want translation? Build a translation model. Want summarization? Build a summarization model. One model, one job, always starting from zero.

3. AI could only read one word at a time. The dominant technology of the day (called RNNs and LSTMs) processed text sequentially, like reading a book strictly left to right. It was slow, and by the end of a long sentence, it had often forgotten the beginning.

Every single letter in “GPT” was an answer to one of these problems. Let’s take them one by one.

1. G is for Generative: The Shift from “Classify” to “Create”

This is the easy part to say, but the hardest to appreciate.

What it means: GPT can create new content. Essays, code, poetry, emails. It generates output that didn’t exist before.

Why it’s a big deal: For decades, AI was a world of “yes/no” answers. Is this spam? Is this a cat or a dog? Does this customer churn? These are classification tasks. AI looks at something and puts it in a bucket.

Creating something new from scratch (a paragraph, a story, a working function of code) was considered nearly impossible. Language is infinite. There are more possible sentences than atoms in the universe. How would an AI pick a good one?

The Generative approach said: don’t pick the “right” sentence. Generate it word by word, always predicting the most likely next word given what came before. Do that billions of times, and coherent writing emerges.

That sounds simple. It is also the shift that took AI from “recognizing patterns in data” to “creating patterns that look human.”

2. P is for Pre-trained: The Free Labels Trick

This one is the real genius, and most explanations skip it.

What it means: Before GPT is ever asked to do anything useful, it has already read a massive amount of text. Books, Wikipedia, websites, articles, code. That’s the “pre” in pre-trained.

Why it’s a big deal: Traditional AI needed labeled data. To teach AI to spot spam, humans had to label millions of emails as “spam” or “not spam.” To teach it to tell cats from dogs, humans had to label millions of photos. Labeled data is expensive, slow, and limited.

Pre-training flipped the entire problem on its head with one insight:

If the task is “predict the next word,” the internet is already labeled. The label is just the next word.

Read “The cat sat on the ___” and the correct answer is whatever word came next in the original sentence. No humans needed. The data labels itself. And the internet has trillions of words.

Suddenly, AI had unlimited training data. GPT-3 was trained on roughly 570 GB of filtered text, pulled from an even larger 45 TB of raw internet data. Later models like GPT-4 and GPT-5 used dramatically more. That scale would have been unimaginable with human-labeled data.

Think of Pre-training as a student reading every book in the library to learn general knowledge. Later, this student can be Fine-tuned (specialized training for a specific job) to become a doctor, a coder, or a chatbot. But the broad education comes first, and it comes from the text itself.

3. T is for Transformer: Seeing All the Words at Once

What it means: The Transformer is a specific type of Neural Network architecture introduced by Google researchers in 2017, in a paper famously titled “Attention Is All You Need.”

Why it’s a big deal: Before Transformers, AI read sentences one word at a time, sequentially. This was slow, and the model often forgot the beginning of a long sentence by the time it reached the end. It also meant you couldn’t spread the work across thousands of chips in parallel, which put a hard ceiling on how big these models could get.

Transformers introduced two superpowers:

1. Parallel Processing: They look at all the words in a sentence simultaneously, rather than one by one. This makes them dramatically faster and, critically, scalable to billions of parameters. Without Transformers, no amount of compute could have produced GPT-3 or GPT-4.

2. Self-Attention: They figure out which words in a sentence relate to each other. In “The bank of the river,” the Transformer pays attention to “river” to know that “bank” means land. In “The bank approved my loan,” it pays attention to “loan” to know bank means the financial kind. Same word, different meaning, figured out from context.

Self-attention is what gave AI something that looks like understanding context. It is the single architectural idea that made modern AI possible.

Why the Combination Changed Everything

Here’s the thing nobody emphasizes enough: each of these three ideas existed on its own before GPT.

• Researchers had built generative models before.

• Unsupervised pre-training had been explored in smaller forms.

• The Transformer paper was published by Google, not OpenAI.

What OpenAI did was combine all three at massive scale. GPT-1 in 2018 showed the recipe could work. GPT-2 in 2019 showed it could write coherently. GPT-3 in 2020 was the moment the world saw what happens when you push this recipe to billions of parameters: the model started doing things it was never explicitly trained to do. Reasoning. Translation. Summarization. Rudimentary code generation. Researchers call these emergent abilities. Capabilities that appear, seemingly out of nowhere, once the model gets big enough.

ChatGPT in late 2022 was when the public caught on.

So when someone says “GPT changed AI,” they are not being dramatic. The specific combination of Generative + Pre-trained + Transformer at scale is the recipe that broke a decades-long logjam.

GPT vs. ChatGPT

Are they the same thing? No.

Here is the best analogy to understand the difference:

Think of a Laptop.

• GPT is the Processor (like Intel or Apple Silicon): It is the raw brainpower and technology that does the thinking.

• ChatGPT is the Laptop (like a MacBook or Dell XPS): It is the product wrapped around that processor with a screen and keyboard (an interface) that allows you to interact with it easily.

GPT is the model; ChatGPT is the application built using the GPT model.

The “Decoder” Secret

If you want to sound extra smart, know this: the Transformer architecture originally came with two parts, an Encoder (to understand input) and a Decoder (to generate output).

GPT models are actually Decoder-only models. They dropped the Encoder entirely. They are specialists in generating text: predict the next token, then the next, then the next, until they have built a whole sentence.

Different AI systems use different slices of the Transformer architecture. Google’s original BERT was Encoder-only (great for understanding and search). GPT is Decoder-only (great for generating). That single design choice is a big part of why GPT models feel so fluent when they write.

The Takeaway

You didn’t just learn what an acronym stands for. You learned the three ingredients that made modern AI possible:

• Generative: AI stopped classifying and started creating.

• Pre-trained: The internet itself became the training data, no humans needed to label it.

• Transformer: AI stopped reading one word at a time and started seeing the whole picture at once.

Each of these had been tried separately. Combining them at scale, between 2018 and 2020, is what OpenAI did. And it is the reason “GPT” became shorthand for modern AI.

The next time someone says “we’re in the GPT era,” you’ll know they don’t mean an acronym. They mean a recipe.

Coming Up

You now know what GPT stands for. But here is a subtle point we glossed over: GPT is just one example of a broader category called Large Language Models (LLMs). Claude, Gemini, Llama, and DeepSeek are LLMs too. So what exactly is an LLM, and why is it the engine behind every chatbot you use? In our next article, we’ll break down the engine behind ChatGPT, Claude, and Gemini and show you why LLMs are the defining technology of this decade.

AI for Common Folks - Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

In our last article, we explored Generative AI, AI that can create brand new content like essays, images, music, and code. We saw it transforming customer support, education, content creation, and software development.

But here’s a question that article left open: what’s actually powering all of that?

When ChatGPT writes your email, when Claude explains your kid’s homework, when Gemini summarizes a research paper. They’re all running on the same type of technology underneath. That technology is called a Foundation Model.

The Old Way vs. The New Way

In traditional AI (the old way), if you wanted to translate English to French, you built a “Translation Model.” If you wanted to summarize text, you built a “Summarization Model.” If you wanted to detect spam, you built a “Spam Model.”

One model, one job. Every new task meant starting from scratch.

Foundation Models changed the game completely. They are like a Swiss Army Knife: one single tool that can perform hundreds of different tasks, from writing code to composing poetry to analyzing legal contracts.

The Analogy: The “Super-Student” in the Library

Remember from our earlier article on What is a Model? We compared an AI model to a student who finished studying and walks into the exam with knowledge in their head?

Now imagine two types of students:

• Traditional AI (The Specialist): This student only studied one book: “How to Repair a Bicycle.” Ask them to fix a bike? Perfect. Ask them to write an essay on history? They have no idea. They are specialized.

• Foundation Model (The Generalist): This student has read almost every book in the library. History, math, coding, poetry, mechanics, medicine. Because they’ve seen so much, they’ve learned general patterns about how the world works.

Now, you can ask this “Super-Student” to fix a bike, or write a poem, or solve a math problem, or draft a legal brief. They have a broad “foundation” of knowledge that allows them to adapt to almost any request.

That’s why they’re called Foundation Models. They serve as the base upon which everything else is built. Just like you lay a concrete foundation before building a house, a hospital, or a skyscraper.

The Framework: Builders vs. Users

Understanding Foundation Models becomes much easier when you split the world into two groups: the people who make the engine, and the people who drive the car.

1. The Builder Perspective (Making the Brain)

These are the engineers at companies like OpenAI, Anthropic, Google, or Meta. They take massive amounts of raw materials (text from the internet, books, code repositories, images) and use sophisticated processes to train these models.

• The Goal: To create a model that learns general patterns about language, logic, and the world. It’s like feeding the “student” all those books.

• The Result: A Foundation Model (like GPT-4o, Claude, Gemini, or Llama).

2. The User Perspective (Putting the Brain to Work)

This is where most businesses and developers sit. They don’t need to know how to build the brain from scratch. They just need to know how to use it.

Think of the Foundation Model as a powerful Engine:

• One user might take that engine and put it into a Race Car (a chatbot for customer service).

• Another user might put it into a Truck (a tool to summarize legal documents).

• Another might put it into a Boat (an app that generates marketing copy).

• Another might put it into an Ambulance (an AI assistant helping doctors with diagnoses).

You don’t need to know how to build the engine. You just use its capabilities to solve your specific problem.

This is exactly what’s happening in the real world right now. Companies like Canva, Notion, Duolingo, and Khan Academy aren’t building their own Foundation Models. They’re plugging into existing ones (GPT, Claude, Gemini) and building products on top.

Types of Foundation Models

While Large Language Models (LLMs) are the most famous type, Foundation Models are expanding into new territories:

1. Large Language Models (LLMs): Trained on text. Good for writing, summarizing, reasoning, and coding. Examples: GPT-4o, Claude, Llama, Gemini.

2. Large Multimodal Models (LMMs): These can understand not just text, but also images, audio, and video. When you upload a photo to ChatGPT and ask “what’s in this image?” That’s a multimodal model at work. Examples: GPT-4o (handles text + images + audio), Gemini (text + images + video).

3. Image Generation Models: Trained on images paired with descriptions. They create visuals from text prompts. Examples: DALL-E 3, Midjourney, Stable Diffusion.

4. Code Models: Specialized for understanding and generating software code. Examples: Claude (Anthropic’s model powers Claude Code), GitHub Copilot (powered by OpenAI models).

All of these are Foundation Models — massive, general-purpose brains that can be adapted to specific jobs.

Why This Matters for You

You might be thinking: “Okay, but I’m not building AI. Why should I care about Foundation Models?”

Three reasons:

1. It explains why AI suddenly got good at everything. Before Foundation Models, AI was narrow — good at one thing, useless at the rest. Foundation Models are why your AI assistant can write an email and explain physics and debug code and plan your vacation. One brain, many skills.

2. It’s why the AI race is so expensive. Training a Foundation Model costs tens of millions to hundreds of millions of dollars. That’s why only a handful of companies (OpenAI, Anthropic, Google, Meta) can afford to build them. Everyone else builds on top of them.

3. It’s the reason AI will keep getting more useful. As Foundation Models get better, every product built on top of them gets better automatically. When GPT improves, every app using GPT improves. That’s the power of a shared foundation.

The Takeaway

A Foundation Model is a general-purpose AI brain:

• It is the engine inside the car.

• It is the student who read every book in the library.

• It is the Swiss Army Knife of the digital world.

It shifted AI from “specialized tools that do one thing” to “general intelligence that can be adapted to almost anything.”

And here’s the exciting part: we’re still early. The Foundation Models of 2026 are dramatically more capable than those of 2023. The ones coming in 2027 and beyond will make today’s look primitive. Understanding this technology now means you won’t be caught off guard as it transforms more of the world around you.

Coming Up

Now that you understand what a Foundation Model is, this massive, general-purpose brain, you’ve probably noticed we keep mentioning one name more than any other: GPT. GPT-3, GPT-4, ChatGPT. But what do those three letters actually stand for? And why did this particular approach become the dominant one? In our next article, we’ll decode the most famous acronym in AI and break down exactly how GPT works, letter by letter.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

We break down AI so it makes sense to real people. If you’ve been following along, our last article explored Predictive Modeling, how AI uses historical data to make educated guesses about the future. That’s AI looking backward to predict forward.

Today we’re flipping the script. What if AI could look at everything it has learned and create something entirely new?

That’s Generative AI. And it’s the reason AI went from a behind-the-scenes tool to something your parents, your boss, and your neighbor are all talking about.

Why This Feels Different from Everything Before

Think about what makes humans special. Our ability to create new things, right? For decades, people said this was the one thing AI would never be able to do.

Before Generative AI, artificial intelligence was mainly used for prediction and classification:

• Will this customer buy our product? (prediction)

• Is this email spam or not? (classification)

• Which movies should we recommend? (recommendation)

Useful, but limited. Now, Generative AI can write a poem, design a logo, generate a 3D model, compose music, or write working software. The creative barrier has been broken.

In fact, the images you see in this article and even the majority of this writing style was created with the help of AI tools. We’re using the very technology we’re explaining.

Four Ways Generative AI Is Already Changing the World

1. Customer Support That Doesn’t Make You Pull Your Hair Out

Remember the last time you needed help and had to wade through an endless phone tree or wait hours for a response?

Companies used to need huge call centers with dozens of employees handling problems. Expensive and frustrating for everyone.

Now, AI-powered chatbots handle the first level of support. When you contact customer service for a food delivery app or your internet provider, your initial questions are likely answered by an AI that understands what you’re asking and gives helpful responses.

The result? Companies reduce their support staff from 10 people to 2-3, while actually improving response times. The AI handles common questions. Human agents focus on the complex issues that really need their attention.

2. Content Creation That’s Indistinguishable from Human Work

“AI can’t be creative” was the mantra for years. That ship has sailed.

Today, if you read an article online, you often cannot tell if a human or an AI wrote it. The quality has improved that dramatically.

This extends beyond writing to image creation, video editing, music composition, and more. Artists and creators aren’t being replaced, but they’re increasingly working alongside AI tools that speed up their workflow.

For everyday people, this means you can generate professional-quality content without years of specialized training. Need a business proposal? A birthday card design? A custom bedtime story for your kids? Generative AI can help create it in minutes.

3. Education That Adapts to How You Learn

Generative AI is transforming education by providing 24/7 personalized learning support. Stuck on a complex topic? AI can explain concepts in multiple ways until you understand.

Tools like ChatGPT, Claude, and Gemini are already being used by students worldwide as study partners. Universities like Northeastern and the London School of Economics have integrated AI tutoring into their programs. Some of these tools use Socratic questioning, asking you guiding questions instead of handing you the answer, helping you actually think instead of just copy.

This newsletter itself is an example. We use AI tools every day to research, draft, and refine explanations so that complex topics reach you in plain language.

4. Software Development That’s Accessible to Almost Everyone

Coding used to be a highly specialized skill requiring years of training. Generative AI has dramatically lowered the barriers.

Today’s AI tools like Claude Code, Cursor, and GitHub Copilot can:

• Write functional code from a simple description

• Explain complex code in plain language

• Debug and fix errors

• Build entire applications with minimal guidance

Tasks that once required a team of five programmers might now be accomplished by two or three with AI assistance. More importantly, people who never thought they could create software are now building tools to solve their own problems.

The “no-code” and “low-code” movements are being supercharged by Generative AI, making software creation accessible to people without technical backgrounds.

Is Generative AI Just Another Tech Bubble?

Before investing your time in learning any technology, it’s smart to ask whether it has staying power. We evaluated Generative AI against five questions:

1. Does it solve real-world problems?
Yes. As we just saw, it’s already making a real difference in customer service, education, content creation, and software development. These aren’t trivial applications.

2. Is it useful in everyday life?
Yes. Unlike some technologies that only benefit specialized industries, Generative AI tools are immediately useful to almost everyone. Writing emails, learning new skills, creating content, understanding complex information. Daily value.

3. Is it creating economic impact?
Absolutely. Private AI investment hit $285 billion in the US alone in 2025, growing over 127% in a single year. Generative AI captures nearly half of all private AI funding globally. Major companies are restructuring entire divisions around AI capabilities.

4. Is it creating new job opportunities?
Yes. While there are legitimate concerns about job displacement, Generative AI is also creating entirely new career paths. The role of “AI Engineer” barely existed a few years ago. Now it’s one of the fastest-growing job categories.

New roles are emerging in prompt engineering, AI ethics, AI training and education, and specialized AI application development across industries.

5. Is it accessible to ordinary people?
Yes. This might be the most important part. Unlike previous waves of technology that required specialized knowledge, today’s Generative AI tools are designed to be used through natural language.

You don’t need to code or understand complex math. You simply talk to them in English, Hindi, Urdu, or whatever language you prefer. The technology is accessible to almost everyone, regardless of technical background.

All five answered yes. Generative AI is following the trajectory of truly transformative technologies like the internet, not temporary hype cycles.

Why This Matters Now

The Generative AI shift isn’t coming. It’s already here. But we’re still in the early days, comparable to where the internet was in the mid-90s. The most significant impacts and opportunities are still ahead.

By understanding these tools now, you position yourself to benefit from them rather than being caught off guard as they transform more aspects of work and life.

The good news? These tools are designed to be intuitive. You don’t need to understand everything about how they work to start benefiting from them today.

Coming Up

Now that you know what Generative AI is and why it matters, the natural next question is: what’s actually powering these tools? In our next article, we’ll break down Foundation Models, the massive pre-trained systems that make ChatGPT, Claude, and Gemini possible. If you’ve ever wondered why some AI tools are smarter than others, that one’s for you.

Inspired in part by CampusX’s Hindi-language AI education content.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

If you’ve ever heard someone say “we trained an AI model” and wondered what that actually means, this one’s for you. Not the buzzword version. Not the textbook version. The real version, explained the way you’d explain it to a friend over chai.

The Problem: Some Things You Can’t Explain

Normally, when a programmer wants a computer to do something, they write exact instructions. Step 1, do this. Step 2, do that. Like a recipe with precise measurements.

“Take the list of numbers. Compare the first two. If the first is bigger, swap them. Move to the next pair. Repeat.”

This works great for things where humans know the exact steps. Sorting numbers. Calculating taxes. Sending an email.

But what about recognizing a dog in a photo?

Try it right now. Look at a photo of a dog and explain, step by step, exactly how you know it’s a dog. Not “it has fur and four legs,” because so does a cat, a bear, and a wolf. What EXACTLY are the steps your brain takes?

You can’t write them down. Nobody can. Your brain does it instantly, but the process is invisible even to you.

So if you can’t explain it to yourself, how do you explain it to a computer?

The Breakthrough: Stop Explaining. Start Showing.

Back in 1949, an IBM researcher named Arthur Samuel had this exact problem. And his idea was simple but radical:

What if we stop writing instructions for the computer? What if we just show it thousands of examples and let it figure out the pattern on its own?

Like a grandma teaching you to cook. She didn’t hand you a formula. She let you try, told you how it turned out, and let you adjust.

That’s machine learning. That’s the whole idea.

The Chai Analogy: How the Learning Actually Works

Imagine you’re learning to make chai. There are things you can control: how much sugar, how much ginger, how long you boil the milk, how many tea leaves. Let’s call these your settings.

On your first try, you just guess. Two spoons of sugar, a small piece of ginger, boil for 3 minutes, one spoon of tea leaves. You taste it. Too sweet, no kick, kind of watery.

Now here’s the important part. You don’t throw everything out and start with a completely random guess. You think: “Too sweet means I need less sugar. No kick means more ginger. Watery means I should boil longer or add more tea leaves.” You adjust your settings based on what went wrong.

You try again. Better. Still not great. You adjust again. And again. After 30 cups, you’re making chai that people actually want to drink.

What just happened?

1. You had settings you could adjust (sugar, ginger, boil time, tea leaves)

2. You had a way to score the result (tasting the chai)

3. You used that score to figure out which settings to change and in which direction (too sweet means LESS sugar, not more)

4. You repeated this process until the score was good

That’s the entire structure of machine learning. Every single AI system in the world follows this pattern.

Now Replace Yourself With a Computer

In machine learning, the “settings” are called weights. They’re just numbers. Thousands of them, sometimes billions. Each one is like one of your chai settings: a small dial that slightly changes the final output.

The “thing being cooked” is called a model. It’s a program that takes an input (like a photo) and produces an output (like “dog” or “cat”). But the output depends entirely on where the weights are set. Same model, different weights, completely different results. Just like same kitchen, same ingredients, but different amounts of sugar and ginger give you completely different chai.

The “tasting” is called a loss function. It’s just a score that measures how wrong the model was. Show it a photo of a dog and it says “cat”? High score (very wrong). It says “dog”? Low score (good). The computer doesn’t “understand” dogs. It just has a number that tells it how far off it was.

The “figuring out which settings to change” is the clever part. Remember how you knew “too sweet” means reduce sugar, not increase it? The computer does something similar. It looks at the score and mathematically traces back through the model to figure out: which weights contributed to the wrong answer, and in which direction should I nudge each one to make the score a little better? This isn’t magic. It’s math. If turning a weight up made things worse, turn it down a little. If turning it down made things better, keep going that direction.

The “trying again” is called training. The computer looks at an example, makes a prediction, checks the score, adjusts the weights, and repeats. Not 30 times like your chai experiment. Millions of times. Across thousands of examples. Each time the weights get a little better. The score gets a little lower. The predictions get a little more accurate.

And Then Something Remarkable Happens

After enough rounds of tasting and adjusting, the model gets good. Show it a photo of a dog it has never seen before, and it says “dog.” Show it a cat, it says “cat.” Not because anyone wrote rules for what a dog looks like. Because the model adjusted its own settings, millions of times, based on millions of examples, until the patterns clicked into place.

Just like you can now walk into any kitchen, with any ingredients, and make decent chai without thinking about it. You don’t follow a recipe anymore. You have a feel for it. The model has its version of that feel: billions of finely tuned weights.

And here’s the part that matters. Once training is done, you lock in the weights. Now the model is just a program. Photo goes in, answer comes out. From the outside, it looks like any other software. The difference is nobody wrote the instructions. The machine found them by practicing.

Your grandma’s teaching method, at scale.

Want to See the Actual Math? Let’s Walk Through It.

Forget images and dogs for a minute. Let’s say you’re trying to predict how much chai your office will drink based on how many people show up.

You’ve noticed a pattern over the past few days:

You can probably see the pattern already: it’s roughly 2 cups per person. But pretend you don’t know that. Pretend you’re a computer that has to figure it out by guessing and adjusting.

Start with a random guess.

The model is the simplest possible formula:

prediction = weight x people

One input (people), one weight (some number we haven’t figured out yet), one output (predicted cups).

Let’s start with weight = 0.5. That’s our first guess.

Round 1: Predict and check.

2 people showed up, they drank 4 cups.

prediction = 0.5 x 2 = 1

We predicted 1 cup. The real answer was 4. Way off.

error = prediction - actual = 1 - 4 = -3

Negative means we predicted too low. The size (3) tells us how far off.

Now, which direction do we nudge the weight?

Our formula was: prediction = weight x people. We predicted too low. The input (people = 2) is fixed. The only thing we can change is the weight. If we make it bigger, the prediction goes up. We need it to go up. So the weight needs to increase.

But by how much? Machine learning uses a formula:

new weight = old weight - learning rate x error x input

The “learning rate” is a small number that controls how big each step is. Let’s use 0.1. Too big and you overshoot. Too small and you take forever.

new weight = 0.5 - 0.1 x (-3) x 2 = 0.5 + 0.6 = 1.1

The error was negative (too low), so the math automatically pushed the weight UP. No if-statement needed. The math handles the direction for us.

Round 2:

prediction = 1.1 x 2 = 2.2

error = 2.2 - 4 = -1.8 (still too low, but closer)

new weight = 1.1 - 0.1 x (-1.8) x 2 = 1.1 + 0.36 = 1.46

Round 3:

prediction = 1.46 x 2 = 2.92

error = 2.92 - 4 = -1.08

new weight = 1.46 + 0.216 = 1.676

Let’s skip ahead and see the pattern:

The weight crawls toward 2.0. The prediction crawls toward 4.0. The error shrinks toward 0.

Nobody told the computer the answer was 2 cups per person. It started at 0.5 and found its way there by repeatedly predicting, checking the error, and nudging the weight in the right direction.

Connect it back to the chai analogy:

• The weight (started at 0.5) is the chai setting. The sugar, the ginger, the dial you’re adjusting.

• The prediction is the chai you made this round.

• The error is you tasting it and knowing how far off it is.

• The update rule is you thinking “too weak, needs more.” Except here it’s a formula, not a feeling.

• The learning rate (0.1) controls how cautious you are. Small sips and small adjustments, not dumping the whole spice jar in at once.

This was one weight. One input. One simple formula.

A real AI model? Same exact process. Same loop. But instead of one weight, it has billions. Instead of “people to chai cups,” it’s “pixels to dog or cat.” Instead of multiplying one number, it passes data through layers of weights, each one getting nudged a tiny bit after every example.

The math gets bigger. The idea doesn’t change.

Predict. Check the error. Adjust the weights. Repeat.

A Note for Common Folks

This article is an attempt to explain how AI learns at the simplest level possible. We skipped a lot of nuance on purpose. Real AI systems are more complex, but the core loop you just read about is genuinely how they all work, from the simplest model to ChatGPT.

One important thing to understand: computers don’t see images, hear sounds, or read text the way you do. A computer only understands numbers. Specifically, everything inside a computer is 0s and 1s.

So how does AI handle different types of input?

Images are stored as grids of numbers. Each pixel has a number for how red it is, how green, and how blue. A 1000x1000 photo is just 3 million numbers. That’s what the model actually “sees.” Not a dog. Not colors. Just a grid of numbers.

Sound is stored as a sequence of numbers representing the wave of air pressure hitting a microphone, thousands of times per second. Your favorite song is just a very long list of numbers.

Text is converted into numbers through a process called tokenization. Each word or piece of a word gets assigned a number. “The cat sat” might become [458, 2093, 7721]. That’s what the model actually reads.

So when we say “a model takes an input and produces an output,” what we really mean is: numbers go in, math happens (weights multiply and add), and numbers come out. The model then maps those output numbers back to something humans understand, like the word “dog” or the next word in a sentence.

That’s why the same learning loop works for everything. Images, audio, text, medical scans, stock prices, language translation. It’s all numbers. And the model is doing the same thing every time: adjusting its weights to get better at turning one set of numbers into another.

If you understood the chai analogy, you understand how AI learns. The rest is just scale.

The Takeaway

Machine learning is not a computer “thinking.” It’s a computer adjusting its own settings, over and over, based on how wrong its guesses are, until the guesses get good enough. The same way you learned to cook, ride a bike, or throw a ball. Try, check, adjust, repeat.

The difference is speed and scale. You made 30 cups of chai. The computer makes 30 million guesses. You adjusted 4 settings. The computer adjusts 30 billion. But the process? Identical.

AI for Common Folks -- Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

Last time, we covered Semi-Supervised Learning, how AI can learn from a small number of labeled examples and a massive pile of unlabeled data.

But through all these conversations about how AI learns, one question keeps coming up:

What is the actual structure inside the machine doing all this learning?

When people say “the AI figured it out,” what is the it they’re referring to?

That’s a neural network. And once you understand what it is, everything else in AI (ChatGPT, Gemini, image generators, voice assistants) suddenly starts making sense.

The Big Reveal: It’s Simpler Than You Think

Here’s something the textbooks rarely tell you upfront.

When Jeremy Howard (founder of fast.ai, one of the most respected AI educators in the world) reveals how neural networks actually work to his students, the most common reaction is:

“Wait... is that ALL it is?”

Neural networks are powerful not because they’re mathematically exotic. They’re powerful because they do something incredibly simple an incredibly large number of times.

Almost everything a neural network does is just addition and multiplication. A lot of it. Done very fast.

What does that look like? Each unit in the network takes incoming signals (numbers), multiplies each one by a “weight” (a number that says how important that signal is), adds all the results together, and passes the total to the next layer. That’s it. Billions of tiny calculators doing grade-school math, over and over.

That’s the secret. Let’s build up to it.

The Analogy: Learning to Recognize Your Friend’s Face

Imagine you’re teaching a child to recognize your friend Sarah from a photo.

You show them 100 photos, some with Sarah, some without, and tell them “this is Sarah” or “this isn’t Sarah” for each one.

The child’s brain starts noticing patterns: Sarah has curly red hair. Her eyes are green. She usually smiles with her teeth.

At first, the child guesses wrong a lot. But every time you correct them, their brain quietly adjusts which features it pays attention to. Curly hair gets more weight. Background color gets less weight.

After enough photos, the child becomes pretty reliable.

A neural network does exactly this. Just with numbers instead of a child’s brain, and millions of examples instead of 100 photos.

The Three Parts of Every Neural Network

Every neural network in the world, from the tiny one in your spam filter to the massive one behind ChatGPT, has the same three-part structure.

1. The Input Layer: “Here’s What I’m Looking At”

This is where raw data enters the network.

• For an image: each pixel becomes a number (0 = black, 255 = white), and each number enters here

• For text: each word or piece of a word enters here

• For audio: sound frequencies enter here

Nothing clever happens in the input layer. It’s just the front door.

2. The Hidden Layers: “Where the Magic Happens”

This is where the network learns patterns. Each hidden layer takes the previous layer’s output and transforms it, mixing and combining signals to find increasingly complex patterns.

Think of it in stages:

• First hidden layer: detects simple features (”there’s a curved line here”)

• Second hidden layer: combines those into shapes (”those curves form an ear”)

• Third hidden layer: combines shapes into concepts (”that ear + those eyes = a face”)

The more hidden layers, the more complex the patterns the network can learn. This is why we call it deep learning: the network goes deep with many layers.

Modern AI systems have hundreds or even thousands of hidden layers.

3. The Output Layer: “Here’s My Answer”

The final layer makes a decision:

• “This image is a cat” (classification)

• “The next word is ‘the’” (language generation)

• “The sentiment of this review is positive” (analysis)

• “This email is spam” (filtering)

The Real Secret: Weights

Here’s the math secret, and it’s not scary.

Every connection between two neurons has a weight: a single number that says how much to trust that connection.

A weight of 2.0 means “pay close attention to this signal.”
A weight of 0.1 means “barely consider this signal.”
A weight of -1.5 means “this signal actually points the other direction.”

The entire job of training a neural network is just this: find the right numbers for every weight.

A typical large language model like Claude or GPT-4 has hundreds of billions of these weights. But each individual weight is still just a number, and finding the right set of numbers is what training is all about.

Think of it this way: the architecture is the instrument, and the weights are the music. Change the weights, and you’ve changed what the network does.

How It Learns: Hiking Downhill in the Fog

Here’s the part most people get wrong: neural networks aren’t programmed with rules. Nobody sat down and typed “if pointy ears AND whiskers, then cat.” The network figures out the rules itself, from examples.

Here’s how:

1. Start random. Every weight is set to a random number. The network starts as dumb as possible.

2. Make a prediction. Feed it an image. It guesses. Probably wrong.

3. Measure how wrong. A “loss function” calculates a single number representing the error, basically a score for how bad the answer was. High loss = very wrong. Zero loss = perfect.

4. Figure out which direction to improve. Using math called gradient descent, the network calculates: “If I increase this weight slightly, does the loss go up or down? Which direction makes me less wrong?”

The best analogy: hiking downhill in the fog. You can’t see the bottom of the valley, but you can feel which way the ground slopes under your feet. You take a small step downhill. Then another. Over time, you find your way to the lowest point.

The “valley” is the best possible set of weights. The fog is the fact that there’s no shortcut. The network has to feel its way there.

5. Adjust the weights. Nudge each weight slightly in the direction that reduces the loss.

6. Repeat millions of times. After enough examples and adjustments, the weights settle into values that make the network surprisingly accurate.

And here’s the thing: gradient descent nearly entirely relies on addition and multiplication. When students see the actual details, the most common reaction is: “Is that all it is?”

The Secret Ingredient: One Tiny Rule That Makes Everything Work

Here’s a surprising thing: if you just stack layers of math on top of each other with nothing in between, the whole thing collapses. It doesn’t matter if you stack 10 layers or 1,000. You end up with a network no smarter than a single layer. All that depth, wasted.

Imagine stacking 100 identical photo filters. The photo doesn’t get more detailed. It just gets darker. Same idea.

The fix is a tiny rule called an activation function, inserted between every layer. The most common one is called ReLU, and its entire job is this:

If the number coming in is negative, make it zero. If it’s positive, leave it alone.

That’s the whole rule. And that tiny step, repeated billions of times, is what gives neural networks the ability to learn curves, recognize faces, understand language, and generate images.

Here’s the intuition: without it, a network can only learn patterns that fit a straight line. With it, the network can bend and trace any shape, no matter how complex. The real world isn’t made of straight lines, so this matters enormously.

Different Networks for Different Jobs

As neural networks evolved, researchers found that different kinds of data work better with different structures. Here are the three types you’ll actually hear about:

CNNs: Built for Eyes

Convolutional Neural Networks are designed to look at images and video. They scan pictures in small patches, finding edges first, then shapes, then full objects, the same way your eye moves across a scene.

You use this: Apple Face ID, self-driving car cameras, doctors’ tools that detect tumors in scans.

Transformers: The Architecture Behind Everything Big

This is the breakthrough that changed AI. Instead of reading data one piece at a time, Transformers look at the whole thing at once and learn what to pay attention to. That’s why they’re so good at understanding context. They don’t just see the word, they see how it relates to every other word around it.

You use this: ChatGPT, Claude, Gemini, Google Translate, GitHub Copilot. And increasingly, image and video AI too.

Diffusion Networks: The Artists

These networks start with pure random noise (think TV static) and gradually “un-blur” it into a real image. They learn by practicing the reverse: taking a real image, adding noise step by step until it’s unrecognizable, then learning how to reverse that process.

You use this: Midjourney, DALL-E, Adobe Firefly, Stable Diffusion, Sora.

Despite their differences, all three architectures are built on the same foundation: layers of simple units adjusting their weights through feedback to recognize and create patterns. The specialization is in how they’re wired, not what they’re made of.

The honest 2026 picture: Transformers dominate. They’ve quietly taken over text, code, and increasingly images and video. If you hear about a major new AI product, there’s a good chance a Transformer is at the center of it.

The Limitations (Keeping It Real)

Great tools deserve honest assessments. Neural networks are not magic.

They need a huge amount of examples.
A child can learn to recognize a cat from 5 photos. A neural network might need 10,000. The more complex the task, the more data required. This is why big tech companies hoard data. It’s the raw material for their models.

They’re black boxes.
Ask a neural network why it classified that email as spam, and it can’t tell you. It just did. This is a serious problem in medicine, law, and anywhere decisions need to be explainable. Researchers are actively working on “explainable AI” to solve this.

They’re brittle in weird ways.
A neural network trained on millions of dog photos might confidently call a wolf a “husky” because wolves didn’t appear in its training data. This is called overfitting: the network becomes so tuned to what it’s already seen that it stumbles on anything new or slightly different. It’s why testing a model on fresh, unseen examples matters so much. Neural networks are pattern-matchers, not reasoners, and they fail in surprising ways when they encounter situations outside their training.

They’re expensive to train.
Training GPT-4 reportedly cost over $100 million and consumed electricity comparable to running thousands of homes for months. This is a real constraint, and not everyone can build or fine-tune large models.

But here’s what’s changing: a technique called transfer learning means you can take a massive pre-trained network that already understands general concepts (like what an edge, a texture, or a face looks like) and fine-tune it with a smaller amount of your specific data. It’s like teaching a seasoned expert a new specialty instead of training a complete beginner from zero. You don’t always need to start from scratch.

Try It Yourself

Want to feel neural network learning in action? Try this:

1. Go to Teachable Machine by Google (free, no code)

2. Click “Get Started” then “Image Project”

3. Create two classes (e.g., “thumbs up” and “thumbs down”)

4. Show your webcam 30-50 examples of each

5. Click “Train Model” and watch accuracy climb in real time

6. Test it live. Hold up your hand and see the neural network classify it

You just trained a neural network. What you watched happen (the accuracy rising as more examples were added) is gradient descent adjusting weights in real time.

The Takeaway

A neural network is a system of layers connected by weights. It learns by:

1. Making a prediction

2. Measuring how wrong it was (loss)

3. Adjusting its weights to be less wrong (gradient descent)

4. Repeating millions of times

The nonlinear “activation function” between layers (as simple as “replace negatives with zero”) is what gives it the power to learn complex patterns, not just straight lines.

The more layers, the more complex the patterns. That’s deep learning.

And that’s the system behind every AI product you use today, from the one that recognizes your face to the one that writes essays, composes music, and generates videos from a single sentence.

Coming Up

Now that you know what a neural network is, the next question is: what happens when you build one that’s massive, trained on essentially all the text ever written on the internet?

Next up: Large Language Models (LLMs), the specific technology powering ChatGPT, Claude, and Gemini, explained for normal people.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

We’ve covered what a Model is (the trained brain) and how Algorithms work (the learning process). Now the big question: why are companies spending billions teaching computers to learn?

They’re not doing it just to beat you at chess.

They’re doing it to see the future.

This brings us to one of the most valuable applications of AI: Predictive Modeling. It’s working behind the scenes every time Netflix recommends a show, your bank flags a suspicious charge, or Spotify creates a playlist that somehow knows your mood.

The Analogy: The Weather Forecast

You already use predictive modeling every morning when you check the weather app.

• Past Data: The app knows that for the last 50 years, when humidity is 90% and wind comes from the east in July, it usually rains.

• Pattern: High Humidity + East Wind in July = Rain likely

• Prediction: “80% chance of rain today. Take an umbrella.”

The computer doesn’t know it will rain. It knows that mathematically, rain is the most likely outcome based on what happened before.

That’s predictive modeling in a nutshell: find patterns in history, apply them to today, make an educated guess about tomorrow.

How It Actually Works

Let’s walk through a real example: predicting if a customer will cancel their streaming subscription.

Step 1: Gather Historical Data
Collect information on 100,000 past subscribers: how often they logged in, what they watched, how long they’ve been a member, and whether they canceled.

Step 2: Train the Model
Feed this data into an algorithm. The algorithm finds patterns:

• “Subscribers who haven’t logged in for 2 weeks AND skipped the last 3 recommended shows usually cancel”

• “Subscribers who added something to their watchlist in the last 7 days almost never cancel”

Step 3: Make Predictions
A current subscriber starts showing warning signs. We feed their activity into the model. The model applies its patterns and predicts: “78% chance of cancellation within 30 days.”

Now the company can send that person a personalized recommendation or a discount offer before they leave. That’s the entire process: historical data, pattern recognition, prediction on new data, then action.

The Two Types of Predictions

Predictive models answer one of two questions:

1. Classification: “Which category does this belong to?”

The model sorts things into buckets. Usually Yes/No, but can be multiple categories.

Examples:

• Email: Is this spam or not spam?

• Banking: Is this credit card transaction fraudulent? (Yes/No)

• Healthcare: Based on this scan, does this patient show early signs of a condition? (Yes/No)

• Customer: Will this subscriber cancel next month? (Yes/No)

2. Regression: “How much? What number?”

The model predicts a specific value.

Examples:

• Real Estate: What will this house sell for based on location, size, and recent sales? ($425,000)

• Rideshare: What should this Uber ride cost right now based on demand and distance? ($23.50)

• Retail: How many units of this product will sell next quarter? (10,000)

• Energy: How much electricity will this city need tomorrow at 3 PM? (4,200 megawatts)

Where You Encounter Predictive Modeling Daily

Your Bank Account:
Every time you swipe your credit card, a model runs in milliseconds predicting: “Does this transaction look like fraud?” Your location, spending history, and the merchant type all become inputs. If the model flags it, your card gets frozen before the thief finishes checkout.

Your Music:
Spotify’s Daylist changes multiple times a day. It predicts your mood based on the time of day, your listening history, and what millions of similar users play at the same hour. Monday morning gets focus music. Friday evening gets party hits. That’s predictive modeling reading your patterns better than you read yourself.

Your Shopping:
Amazon predicts what you’ll want before you know you want it. Its models are so confident in their predictions that the company has patented “anticipatory shipping,” where they start moving products toward your area before you even click “buy.”

Your Health:
UnitedHealth and other insurers now use predictive models to flag patients at risk of hospitalization. Your age, conditions, prescription history, and recent visits become inputs. The model predicts who needs outreach before an emergency happens. (This is also why AI in healthcare is one of the most debated topics right now.)

Your Commute:
Google Maps predicts traffic using current conditions and years of historical patterns. It knows that this specific highway slows down every Tuesday at 5:15 PM, and it reroutes you before you hit the jam. Google recently started using AI to predict flash floods the same way, turning old news reports into data that saves lives.

The Prediction Isn’t Perfect

This is crucial to understand: predictions are probabilities, not certainties.

When a model says a subscriber will cancel, it might mean “78% chance of cancellation.” That’s not 100%. Sometimes the model is wrong. The subscriber might have just been on vacation.

A patient flagged as high-risk might be perfectly healthy. A “guaranteed” sunny day might surprise you with rain. A transaction flagged as fraud might be you buying something unusual on a trip.

We measure model quality by testing it: hide some historical data, ask the model to predict it, compare predictions to reality. A model that’s right 95% of the time is excellent. One that’s right 51% of the time is barely better than a coin flip.

The Limitations (Keeping It Real)

Predictive modeling has real constraints:

Historical bias: If past data reflects bias (certain groups were denied loans unfairly, certain neighborhoods were over-policed), the model learns and repeats that bias. Amazon scrapped an AI hiring tool in 2018 because it penalized resumes that included the word “women’s,” since it was trained on a decade of male-dominated hiring data.

Assumes patterns continue: Models assume the future looks like the past. They fail when something unprecedented happens. COVID-19 broke nearly every predictive model in existence because no historical pattern could account for the entire world shutting down simultaneously.

Correlation isn’t causation: A model might find that ice cream sales predict crime rates. Both rise in summer. But ice cream doesn’t cause crime. Good data scientists catch these traps. Bad ones build products around them.

Only as good as the data: Missing or inaccurate data leads to wrong predictions. Garbage in, garbage out. A model trained on data from one country may completely fail in another.

The Takeaway

Predictive Modeling is the bridge between data and decision-making.

• It uses algorithms to find patterns in historical data

• It creates a model that applies those patterns to new situations

• It helps us make educated guesses about the future

It’s not a crystal ball. It’s statistics at scale: finding what usually happens and betting that it’ll happen again. The companies that do it well (Netflix, Spotify, Google, your bank) feel like they can read your mind. The ones that do it poorly feel like that friend who always gives confidently wrong advice.

Coming Up:
We’ve built a strong foundation: AI, Machine Learning, Models, Algorithms, and Predictive Modeling. But how does the AI actually learn these patterns under the hood? In the next edition, we’ll explore Neural Networks, the architecture inspired by the human brain that makes all of this possible. If you’ve ever heard someone say “deep learning” and wondered what makes it “deep,” that one’s for you.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

An Algorithm is a step-by-step set of instructions that tells a computer exactly how to solve a problem or complete a task. Think of it like a recipe, but for machines.

Hey Common Folks!

We just learned that a Model is the “finished product” of AI, the thing you actually interact with when you use ChatGPT or get a Netflix recommendation.

But how does a model learn? What’s the actual process that transforms raw data into intelligent predictions?

That’s where Algorithms come in.

You’ve heard this word blamed for everything: why you spent 3 hours on TikTok, why your loan was denied, why you saw that specific ad for sneakers. People whisper it like it’s a mystical force: “The Algorithm did it.”

Let’s demystify this. An algorithm isn’t a sentient being plotting against you. It’s just a set of instructions. That’s it.

The Analogy: The Chef’s Recipe

Think about baking a cake:

1. The Ingredients (Data): Flour, sugar, eggs, chocolate. Raw stuff that can’t do anything on its own.

2. The Recipe (Algorithm): The instructions that say: “Mix flour and sugar. Add eggs. Bake at 350 degrees for 30 minutes.”

3. The Cake (Model): The finished result you actually eat.

In AI:

• We feed Data (ingredients) into an Algorithm (recipe)

• The algorithm processes that data, finds patterns, learns

• It produces a Model (cake) we can use

You interact with the cake, not the recipe. But without the recipe, there’s no cake.

Traditional Algorithms vs. AI Algorithms

Here’s where it gets interesting.

Traditional Software (Rigid):
A calculator follows fixed rules:

• Input: 2 + 2

• Rule: Add them

• Output: 4

The algorithm never changes. It does exactly what it’s told, every time.

Machine Learning (Adaptive):
AI algorithms are designed to change themselves based on data. It’s like a recipe that rewrites itself to make the cake taste better every time you bake it.

The algorithm looks at examples, adjusts its approach, and gradually improves, without a human manually updating the rules.

Three Types of Algorithms You’ll Hear About

1. Decision Trees (The Flowchart)

Imagine playing “20 Questions”:

• Is it an animal? (Yes)

• Does it bark? (No)

• Does it meow? (Yes)

• Conclusion: It’s a cat.

A Decision Tree splits data into smaller branches based on simple Yes/No questions until it reaches an answer. It’s simple, logical, and easy to explain.

Used for: Loan approvals, medical diagnosis, customer segmentation.

2. Neural Networks (The Brain Mimic)

This is the heavy hitter behind Deep Learning and modern AI.

Imagine a massive web of interconnected switches:

• Input comes in (a picture of a face)

• Data passes through layers of these switches

• Each layer looks for something: edges, shapes, eyes, noses

• Final layer makes a decision: “This is Alex”

The algorithm learns by adjusting the strength of connections between switches. Stronger connections = more important patterns.

Used for: ChatGPT, image recognition, voice assistants, self-driving cars.

3. Gradient Descent (The Hiker)

This is the algorithm that trains neural networks.

Imagine you’re on a mountain at night, blindfolded, trying to reach the bottom (the best answer):

• You feel the ground with your foot

• If it slopes down, you step that way

• You keep feeling the slope (Gradient) and stepping down (Descent)

• Eventually, you reach the lowest point

This is how AI learns: it makes a guess, measures how wrong it is, and adjusts to be slightly less wrong next time. Repeat millions of times.

Why Do We “Blame” The Algorithm?

When people say “The Instagram Algorithm,” they mean a specific set of rules designed to maximize your engagement:

• Input: Your past likes, watch time, shares

• Algorithm: A formula predicting: “If we show this video of a Golden Retriever, there’s a 90% chance they’ll watch it.”

• Action: Show the video

It feels like manipulation, but it’s just math predicting your behavior based on your history. The algorithm optimizes for what you click, not what’s good for you.

Common Algorithms in Plain English

• Linear Regression: Draws a straight line to predict numbers. Used for: house prices, salary predictions.

• Logistic Regression: Separates things into categories. Used for: spam vs. not spam, pass vs. fail.

• Decision Trees: Asks yes/no questions to classify. Used for: loan approvals, medical diagnosis.

• Random Forest: Many decision trees voting together. Used for: more accurate classifications.

• Neural Networks: Layers of math mimicking brain connections. Used for: images, language, complex patterns.

The Limitations (Keeping It Real)

Algorithms aren’t perfect:

Garbage in, garbage out: An algorithm trained on bad data produces bad results.

Bias amplification: If historical data contains bias, the algorithm learns and repeats that bias.

Not truly “intelligent”: Algorithms follow patterns. They don’t understand meaning or context the way humans do.

Overfitting: Sometimes algorithms memorize training data instead of learning general patterns, then fail on new data.

The Takeaway

An algorithm is just a tool, the “how-to” guide for a computer.

• It tells the computer how to process data

• It defines how a model learns and improves

• It’s math and logic, not magic or conspiracy

Understanding this takes the mystery out of it. When someone blames “the algorithm,” they’re really blaming a set of instructions doing exactly what it was designed to do: optimize for a specific goal.

Coming Up:
Now you know what Models are and how Algorithms train them. But what’s the point of all this learning? In the next edition, we’ll explore Predictive Modeling, how AI uses patterns from the past to predict the future.

AI for Common Folks. Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

We’ve covered the umbrella (AI), the engine (Machine Learning), how computers learn (Deep Learning), the fuel (Data Science), and the three ways AI learns (Supervised, Unsupervised, and Semi-Supervised).

But when you open ChatGPT, or when Netflix recommends a movie, or when your bank approves a loan — what are you actually interacting with?

You’re interacting with The Model.

In the AI world, people often confuse “Algorithm” and “Model.” They use them interchangeably, like “Engine” and “Car.” But they’re different things. Today, we’re defining exactly what a Model is, because this is the “product” that companies are actually building, selling, and competing over.

The Analogy: The Student and the Exam

Think about a student preparing for a math exam.

1. The Study Method (Algorithm): How the student learns — flashcards, practice problems, tutoring. This is the process of improving.

2. The Textbooks (Training Data): The material they study from.

3. The Student on Exam Day (Model): Once studying is done, they walk into the exam. They’re not holding the textbook anymore. They’re holding the knowledge in their head.

The Model is the student’s brain after they’ve finished studying.

When you ask ChatGPT a question, you’re not running the training process again. You’re asking the “graduated student” to use what they already know to give you an answer.

What Does a Model Actually Look Like?

If you could crack open an AI model file (like a .bin or .pytorch file) and peek inside, what would you see?

Not miniature brains. Not videos.

Numbers. Billions of them.

A model is simply a Parameterized Math Function. Remember high school math?

Where:

• x is the input (e.g., house size)

• y is the output (e.g., house price)

• m and b are the Parameters (the learned values)

When we “train a model,” we’re finding the perfect numbers for m and b so the equation fits the data accurately.

• In a simple model: You might have 2 parameters

• In GPT-4: You have hundreds of billions of parameters

The “Model” is just that massive list of numbers saved in a specific structure. That’s it.

The Three Stages of a Model’s Life

Every model goes through this lifecycle:

1. Initialization (The Blank Slate)
We create the architecture (the structure), but it knows nothing. The weights are random numbers. It’s essentially a baby brain.

2. Training (The Education)
We feed it data. The model makes a guess, gets it wrong, and the algorithm adjusts those numbers slightly. This happens millions of times until accuracy improves.

3. Inference (The Job)
Training is done. We “freeze” the numbers — they stop changing. This static file (the trained model) goes into an app. When you type a prompt, the model uses those frozen numbers to calculate an answer.

Why Are Some Models “Smarter”?

Why is GPT-4 smarter than a simple spam filter?

It comes down to Capacity:

Shallow Models (Simple):

• Like Linear Regression — draws a straight line through data

• Great for simple predictions (house prices based on square footage)

• Fails at complex tasks

Deep Models (Complex):

• Like Deep Neural Networks — many layers stacked together

• Can learn incredibly complex patterns

• Powers language understanding, image recognition, creative generation

More parameters + more layers + more training data = more capable model.

Models You Use Every Day

• ChatGPT / Claude / Gemini: Large Language Models (LLMs) with billions of parameters

• Face ID: A vision model that learned your facial features

• Spotify Discover Weekly: A recommendation model predicting what you’ll enjoy

• Google Search: Multiple models ranking and understanding your queries

The Limitations (Keeping It Real)

Models aren’t magic — they have real constraints:

Only as good as their data: A model trained on biased data learns biased patterns.

Frozen knowledge: Once trained, a model doesn’t learn new things unless retrained. That’s why ChatGPT has a “knowledge cutoff.”

Black boxes: Complex models often can’t explain why they made a decision. They just... work.

Size vs. speed tradeoff: Bigger models are smarter but slower and more expensive to run.

The Takeaway

When you hear “OpenAI released a new model,” translate that in your head to:

“OpenAI finished training a massive mathematical function and saved the resulting list of numbers into a file that we can now use.”

• Algorithm: The recipe for learning

• Data: The ingredients

• Model: The finished cake

You eat the cake, not the recipe. You use the model, not the training process.

Coming Up:
Now that you know what a Model is, how does it actually learn? In the next edition, we’ll explore Algorithms — the step-by-step processes that turn raw data into intelligent models.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

We’ve covered the two main ways computers learn:

• Supervised Learning: The teacher stands over the student, correcting every mistake (high effort, requires answer keys).

• Unsupervised Learning: The student is left alone with a pile of books and told to “figure it out” (low effort, but harder to guide).

But what if there’s a middle ground? What if you’re a busy teacher who only has time to grade a few papers out of a hundred? Can the computer still figure out the rest?

Yes. This is called Semi-Supervised Learning, and it’s the secret weapon of big tech companies.

The Problem: Labels Are Expensive

Why don’t we just use Supervised Learning all the time? Because giving the computer the “answer key” is expensive and exhausting.

Imagine you want to build an AI that detects rare diseases in X-rays.

The Data: You can easily download 100,000 X-ray images from a hospital database. That’s the easy part.

The Labels: To tell the computer which X-ray shows the disease, you need a highly paid doctor to look at every single image and mark it. You can’t afford to pay a doctor to label 100,000 images.

This is where Semi-Supervised Learning saves the day. You pay the doctor to label just 1,000 images, and the AI uses that knowledge to figure out the remaining 99,000.

Real-World Proof: Modern deep learning has shown you can train state-of-the-art models with surprisingly small labeled datasets—sometimes as few as 150 images per category. Semi-supervised learning pushes that efficiency even further.

The Google Photos Example

The best example—and one you probably use—is Google Photos.

Step 1 - Unsupervised Grouping: You upload 5,000 photos of your family. Google’s AI looks at them and notices, “Hey, this face appears in 500 photos. That face appears in 200 photos.” It doesn’t know who they are, but it knows they’re the same person. It groups them together using clustering.

Step 2 - The Supervised Nudge: You click on one photo and type “Dad.”

Step 3 - Semi-Supervised Magic: The AI takes that one label (”Dad”) and instantly applies it to the other 499 photos in that group.

You did 1% of the work (labeling one photo), and the AI did 99% of the work (labeling the rest). That’s Semi-Supervised Learning in action.

How Does It Work?

It follows a simple logic:

1. Cluster First: The AI looks at all the data (labeled and unlabeled) and groups similar things together. It notices that Data Point A is very similar to Data Point B.

2. Propagate the Label: If you tell the AI that “Point A is a Cat,” the AI assumes that since Point B looks exactly like Point A, then Point B must be a Cat too.

The key assumption: data points that are close to each other probably share the same label.

2026 Update: Self-Supervised Learning Takes Over

Here’s where things get exciting. Since around 2018, a cousin of semi-supervised learning has become the backbone of modern AI: Self-Supervised Learning.

What’s the difference?

• Semi-Supervised: You give the AI a few labels, and it uses unlabeled data to fill in the gaps.

• Self-Supervised: The AI creates its own labels from the data itself—no human labels needed at all.

Real-World Example: How ChatGPT Learned to Write

ChatGPT wasn’t trained by humans writing “correct answers” for billions of sentences. Instead, it used self-supervised learning:

1. Take a sentence: “The cat sat on the ___”

2. Hide the last word (”mat”)

3. Train the AI to predict it

4. Repeat this billions of times with internet text

The AI creates its own “quiz” from raw text, learning language patterns without anyone labeling a single sentence. This is why GPT-4, Claude, and Gemini could train on trillions of words without hiring millions of human teachers.

Why this matters: Self-supervised learning is the reason AI exploded in the 2020s. It unlocked the internet’s massive, messy, unlabeled data.

Spot It in the Wild: Where You’re Already Using It

You interact with semi-supervised and self-supervised learning every day:

• Spotify Discover Weekly (it knows what songs you like and finds similar unlabeled music)

• Gmail spam filter (you mark a few emails as spam; it learns patterns to catch the rest)

• Medical diagnosis tools (doctors label a few images; AI extends that knowledge across millions)

• ChatGPT, Claude, Gemini (all trained with self-supervised learning on massive unlabeled text)

When Semi-Supervised Learning Fails

Just like any machine learning technique, semi-supervised learning has limitations you need to watch for:

1. Garbage In, Garbage Out (Amplified)

If your small labeled dataset is biased or wrong, the AI will spread that bias across all the unlabeled data.

Example: If you label 10 photos of “doctors” and they’re all men, the AI might learn “doctor = male” and mislabel female doctors in the unlabeled set.

2. The “Close Together” Assumption Can Break

Semi-supervised learning assumes similar-looking things have the same label. But what if they don’t?

Example: Huskies and wolves look similar, but they’re not the same. If your labeled data only has huskies, the AI might confidently—and wrongly—label wolves as “husky.”

3. Domain Shift

If your unlabeled data comes from a different source than your labeled data, the AI can get confused.

Example: You label 100 professional X-rays (high quality, well-lit). Then you feed the AI 10,000 unlabeled phone photos of X-rays (blurry, poor lighting). The patterns don’t transfer well.

The Fix: Always check your unlabeled data’s quality and diversity before letting the AI loose on it.

Why This Matters Now

We live in a world where data is cheap, but labels are expensive.

• We have billions of tweets, but we don’t know the sentiment of all of them.

• We have millions of hours of YouTube video, but we don’t have transcripts for all of them.

• We have endless medical images, but we can’t afford experts to label every one.

Semi-Supervised Learning allows companies to unlock the value of massive, messy datasets without hiring thousands of humans to manually tag every single file.

And Self-Supervised Learning is why AI could suddenly read, write, code, and converse in 2023-2026 without needing labeled “correct answers” for every sentence on the internet.

Try It Yourself

Want to see semi-supervised learning in action? Here’s a simple experiment:

1. Open Google Photos (or Apple Photos)

2. Upload 50+ photos with at least 2-3 people appearing multiple times

3. Wait for the app to cluster faces

4. Label just one photo of each person

5. Watch the AI instantly label dozens more

That’s semi-supervised learning working for you—right on your phone.

The Takeaway

Semi-Supervised Learning is the “work smarter, not harder” approach to AI.

• Supervised: Requires a teacher for every lesson.

• Unsupervised: No teacher at all.

• Semi-Supervised: The teacher gives a few examples, and the student figures out the rest by association.

• Self-Supervised (2026 Bonus): The student creates their own practice tests from the material itself.

This is how your phone organizes your memories, how medical AI detects diseases without bankrupting the hospital, and how ChatGPT learned to write without anyone grading billions of essays.

AI for Common Folks — Making AI understandable, one concept at a time.

Learn More

Want to dive deeper into practical AI? Check out the free fast.ai course, which inspired several examples in this article and teaches you to build real AI applications from day one.

Previous articles in this series:

• What Is Artificial Intelligence?

• What Is Data Science?

• What Is Machine Learning?

• What Is Deep Learning?

• What Is Supervised Learning?

• What Is Unsupervised Learning?

Subscribe now

Leave a comment

Hey Common Folks!

Last week we talked about Supervised Learning—the kind where we hold the computer’s hand and show it the right answers. Today we’re going somewhere more mysterious.

What happens when you don’t have an answer key? What if you have mountains of data but no one has labeled any of it? What if you don’t even know what questions to ask?

That’s where Unsupervised Learning comes in.

The Toy Box Analogy

Imagine dumping a bucket of toys in front of a toddler: red blocks, blue balls, yellow cars, green stuffed animals. You don’t tell them the names. You don’t explain what goes with what. You just watch.

What happens?

The toddler starts sorting. Maybe all the round things go in one pile. Maybe all the red things go together. Maybe the soft toys get separated from the hard ones.

The child doesn’t know the words “ball” or “block,” but they’ve discovered something profound: these things are similar to each other, and those things are different.

That’s Unsupervised Learning. The machine groups data based on similarities it discovers, without anyone telling it what the categories should be.

The Key Difference: No Labels, No Answers

In Supervised Learning, we showed the computer 10,000 emails and told it “this is spam” or “this is not spam.” We provided the answers.

In Unsupervised Learning, we just dump 10,000 emails on the computer and say “find the patterns.” We don’t tell it what spam looks like. We don’t even tell it to look for spam.

The computer might discover: “Aha, there’s a group of emails with similar characteristics—they all have words like FREE MONEY, they come from weird addresses, they have lots of exclamation points!!!”

It found the pattern. We just didn’t tell it what to call it.

The Three Superpowers of Unsupervised Learning

Since we’re not predicting specific answers, Unsupervised Learning typically does one of three jobs:

1. Clustering: The Automatic Organizer

This is the most common use. The AI looks at your data and automatically groups similar items together.

The Student Example: Imagine plotting 1,000 college students by their grades and attendance. You don’t label anyone as “high achiever” or “struggling.” But when you look at the chart, you see natural clusters: one group with high grades and high attendance, another with low grades and spotty attendance, and a middle group coasting along.

The AI draws circles around these groups automatically. It discovered three types of students without anyone teaching it the categories.

Real-World Use—Amazon’s Recommendations: Amazon doesn’t manually sort you into “tech enthusiast” or “new parent.” Instead, their AI notices you buy the same types of products as certain other customers, groups you with them, and recommends what that group typically buys next. You’re in an invisible club you didn’t know existed.

2. Anomaly Detection: The Digital Security Guard

Instead of finding what’s similar, the AI hunts for what’s weird. It learns what “normal” looks like, then flags anything that doesn’t fit.

The Credit Card Example: Your bank doesn’t have a list of “fraud transactions” to train on. Instead, it learns your normal pattern: $50 at the grocery store in Indiana, $30 for gas, $15 at Starbucks.

Then one day, boom—a $5,000 charge in Las Vegas.

The AI sees this as an outlier, way outside your normal pattern. It doesn’t need to be told “this is fraud.” It just knows “this is weird,” and freezes your card.

3. Association: The “People Who Bought This Also Bought” Engine

This finds rules hidden in your data. It discovers that when X happens, Y tends to happen too.

The Famous Example: Walmart’s data team discovered something bizarre in their transaction data. Men who bought diapers on Friday evenings also tended to buy beer.

No one programmed this rule. The algorithm discovered the pattern: new dads stopping for diapers were also grabbing beer for the weekend.

Netflix’s Secret: When you finish watching Inception, Netflix suggests Interstellar. Not because someone manually linked these movies, but because the algorithm noticed people who watched one usually watched the other. It associated the two based purely on viewing patterns.

The Big Challenge: How Do You Know It’s Right?

Here’s the uncomfortable truth about Unsupervised Learning: you can’t always tell if it’s right.

In Supervised Learning, if the AI calls a cat a dog, we correct it immediately. Wrong answer.

In Unsupervised Learning, the AI might group your customers by shoe size instead of spending habits. Is that wrong? Technically no—it found a pattern. But is it useful? Probably not.

This is why human expertise still matters. The AI finds patterns we never knew existed, but humans have to interpret whether those patterns actually mean something valuable.

Where You’re Already Using It

You interact with Unsupervised Learning more than you realize:

Netflix and Spotify recommendations work by clustering users with similar tastes and suggesting what others in your cluster enjoyed.

Google Photos automatically groups pictures of the same person together, even though you never labeled anyone. It learned to recognize faces and found the pattern: “these 50 photos all contain the same face.”

Credit card fraud detection flags unusual purchases based on your personal spending patterns, not a pre-labeled list of “fraud types.”

Spam filters got their start with Supervised Learning, but many now use Unsupervised Learning to catch new spam tactics no one has labeled yet.

The Takeaway

Unsupervised Learning unlocks the value hidden in raw, unlabeled data. It finds patterns we didn’t know to look for.

While Supervised Learning needs a teacher, Unsupervised Learning is the self-starter—the algorithm that explores data on its own and surfaces insights humans might never have discovered.

It clusters similar things. It spots weird outliers. It discovers associations we didn’t see coming.

Coming Up: We’ve covered learning with a teacher (Supervised) and learning alone (Unsupervised). But what about learning through trial and error—getting rewards for good choices and penalties for bad ones? That’s Reinforcement Learning, the technique teaching robots to walk and AI to master video games. We’ll explore it next.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Hey Common Folks!

We’ve covered the umbrella (AI) and the engine (Machine Learning). Now we’re zooming into the most popular way machines actually learn: Supervised Learning.

If Machine Learning is the school, Supervised Learning is the class where the teacher gives you the answer key before the exam.

Think about it: when you learned to read as a child, someone didn’t just hand you a pile of books and say “figure it out.” They pointed at an apple and said “Apple.” They pointed at a banana and said “Banana.” They supervised your learning by giving you the correct answers.

That’s exactly how Supervised Learning works.

What Makes It “Supervised”?

The word “supervised” means there’s a teacher involved. In technical terms, we train the computer using labeled data:

Data = The question (a picture, an email, a patient’s symptoms)

Label = The correct answer (cat, spam, cancer)

We show the computer thousands of examples where we already know the right answer. The computer’s job is to find the pattern connecting the input to the output.

Example: We show a computer 10,000 emails. For each one, we’ve already marked it as “Spam” or “Not Spam.” The computer studies these examples and learns: “Aha! Emails with words like ‘FREE MONEY’ and ‘CLICK NOW’ tend to be spam.”

After training, we can show it a brand new email it’s never seen, and it correctly predicts: Spam.

The Training Process: How It Actually Learns

Let’s say we want to predict whether students will get job placements based on their grades and IQ scores. Here’s how Supervised Learning works:

Step 1: Gather Labeled Data

We collect data on 1,000 students: their CGPA, IQ scores, and whether they got placed (Yes/No). The “Yes/No” is our label—the correct answer.

Step 2: Split the Data

We divide our 1,000 students into two groups:

Training Set (800 students): The computer studies these examples WITH the answers. It learns the mathematical relationship between good grades and job placement.

Test Set (200 students): We hide the answers. The computer makes predictions, and we check how many it got right. This tells us if it actually learned or just memorized.

Step 3: Learn and Adjust

The computer makes a prediction, checks if it was right or wrong, and adjusts its internal “thinking.” It repeats this millions of times until it gets really accurate.

The Two Flavors of Supervised Learning

Supervised Learning solves two types of problems. Think of them as two different subjects in school:

1. Classification (Sorting into Buckets)

This is when the answer is a category—Yes or No, Cat or Dog, Spam or Not Spam.

The question: “Which bucket does this belong in?”

Example: Is this email spam? Is this tumor malignant or benign? Will this customer cancel their subscription?

The answer is always one of a limited set of options. There’s no “half-spam.”

2. Regression (Predicting a Number)

This is when the answer is a continuous number—not a category.

The question: “What number should this be?”

Example: What will this house sell for? What temperature will it be tomorrow? How much revenue will we make next quarter?

The answer could be any number: $450,000, 72 degrees, $1.2 million.

Quick way to remember: Classification = Categories (this OR that) Regression = Real numbers (how much, how many)

Where You’re Already Using Supervised Learning

You interact with Supervised Learning dozens of times a day:

• Email spam filters → Classification (spam or not spam)

• Credit card fraud detection → Classification (fraudulent or legitimate)

• House price estimates on Zillow → Regression (predicting dollar amounts)

• Medical diagnosis tools → Classification (disease present or not)

• Weather forecasts → Regression (predicting temperature, rainfall amounts)

The Catch: The Labeling Bottleneck

Supervised Learning is powerful, but it has one big limitation: someone has to label all that data first.

That spam filter? Someone had to manually mark thousands of emails as “Spam” or “Not Spam” before the computer could learn.

That medical AI? Doctors had to review thousands of X-rays and mark which ones showed tumors.

This is expensive, time-consuming, and if humans make labeling mistakes, the AI learns those mistakes too. Garbage in, garbage out.

The Takeaway

Supervised Learning is the most common and reliable form of Machine Learning because we define what “correct” looks like. We’re the supervisor, providing the answer key.

If the AI is predicting a category (Yes/No, Cat/Dog), it’s Classification.

If the AI is predicting a number (price, temperature, score), it’s Regression.

Next time your email app catches a phishing attempt or Google Maps predicts your arrival time, you’ll know: that’s Supervised Learning doing its job—pattern recognition trained on millions of labeled examples.

Coming Up:

But what happens when we don’t have the answer key? What if we just dump a pile of data on the computer and say “find the patterns yourself”? That’s the world of Unsupervised Learning, and we’ll explore it next.

Was this helpful? Reply and let us know what AI/ML/Data Science concept confuses you the most!

AI for Common Folks — Understand AI in plain English

Subscribe now

Leave a comment

Data Science is using scientific methods and tools to extract useful insights from data. It’s the process of turning raw information into decisions - like a doctor diagnosing a patient’s health using symptoms, tests, and medical history.

AI for Common Folks
Jan 26, 2026

Hey Common Folks!

We’ve talked about the “brain” (AI), the “learning engine” (Machine Learning), and the “complex neural wiring” (Deep Learning).

But today, we are zooming out to look at the entire hospital where all this diagnosis and treatment happens. We are talking about Data Science.

You’ve heard the phrase “Data is the new oil.” It’s a cliché, but it’s true. However, crude oil sitting in the ground is worthless. You need to refine it into gasoline, plastic, and jet fuel before it powers anything.

Data Science is that refinery. It is the process of taking raw, messy information and turning it into gold.

What Is Data Science?

Data Science is the field of using scientific methods, algorithms, and systems to extract knowledge and insights from data—both structured (like spreadsheets) and unstructured (like customer reviews or images).

Think of it this way: Data Science is like being a doctor for organizations.

The Doctor Analogy

Imagine you’re not feeling well and you visit a doctor. Here’s what happens:

1. Symptoms (The Problem)
You tell the doctor: “I’m always tired, constantly thirsty, and losing weight.”
→ In business: “Our sales are dropping,” “Customers are leaving,” “The machine keeps breaking.”

2. Medical History (Historical Data)
The doctor asks: “When did this start? Has it happened before? Any family history?”
→ In Data Science: You look at past performance, previous issues, patterns over time.

3. Running Tests (Data Collection)
The doctor orders blood tests, X-rays, maybe an MRI—gathering evidence.
→ In Data Science: You pull data from databases, surveys, sensors, website logs, customer calls.

4. Diagnosis (Data Analysis)
The doctor analyzes all the test results and finds the root cause: “You have Type 2 diabetes.”
→ In Data Science: “Your sales are dropping because new customers aren’t returning after their first purchase.”

5. Treatment Plan (The Model/Solution)
The doctor prescribes medication, lifestyle changes, and a monitoring plan.
→ In Data Science: You build a model or create a strategy—”Send personalized follow-up emails to first-time buyers within 48 hours.”

6. Monitoring & Follow-up (Deployment & Evaluation)
The doctor checks: “Are your blood sugar levels improving? Do we need to adjust the medication?”
→ In Data Science: “After implementing the email campaign, did repeat purchases actually increase?”

7. Explaining to the Patient (Communication)
A good doctor doesn’t just write a prescription—they explain what’s wrong, why it happened, and how the treatment works.
→ A good Data Scientist translates complex findings into plain English for executives: “If we fix the onboarding experience, we’ll retain 20% more customers—that’s $500K in annual revenue.”

Data Science uses AI and Machine Learning as tools, but it also involves statistics, visualization, domain expertise, and a lot of human judgment.

The “Secret Sauce” Ingredients

Data Science isn’t just one skill; it’s a mix of three things:

1. Computer Science (The Tech Skills)
You need to code to handle massive amounts of information. Python and SQL are the most common languages. You don’t need to be a software engineer—just comfortable enough to wrangle data and automate analysis.

2. Math & Statistics (The Logic Skills)
You need to know if the patterns you see are real or just random luck. This is where statistics comes in—understanding averages, probabilities, correlations, and whether results are statistically significant.

3. Domain Knowledge (The Real-World Skills)
This is crucial and often underrated. If you’re analyzing cricket data, you need to know what a “run rate” is. If you’re analyzing cancer data, you need to know biology. The best insights come when you combine technical skills with subject-matter expertise.

Just like a doctor needs to know medicine (domain knowledge), anatomy (science), and how to use medical equipment (tools)—a Data Scientist needs all three ingredients.

How Does It Actually Work? (The Lifecycle)

Data Science isn’t just “running code.” It’s a lifecycle. Based on how experts break it down, here’s what happens behind the scenes:

1. Define the Question (The Symptoms)

Before touching any data, you need a clear question:

• “Why are customers leaving our service?”

• “Which patients are at highest risk for complications?”

• “What causes this factory machine to break down?”

No question = no direction. This step sounds obvious, but most failed data projects skip it.

2. Data Collection (Running the Tests)

You gather data from databases, spreadsheets, APIs, sensors, surveys—anywhere relevant information exists.

Example: A retail company investigating falling sales might collect:

• Purchase history

• Website clickstream data

• Customer service call transcripts

• Weather data (ice cream sells better when it’s hot!)

• Competitor pricing

3. Data Cleaning (The Janitor Work)

Real data is messy. It has missing values, typos, duplicates, and errors. This unglamorous step takes 60-80% of a Data Scientist’s time.

Example:
One customer entered their age as “25,” another as “Twenty-Five,” a third accidentally typed “250,” and a fourth left it blank. A Data Scientist has to fix all of this before the computer can use it.

The saying in the field: “Garbage In, Garbage Out.” If you feed bad data into a model, you get bad predictions—just like a doctor making the wrong diagnosis from contaminated lab samples.

4. Exploratory Data Analysis (Finding Patterns)

This is where you “interview” the data by creating graphs, charts, and statistical summaries to find hidden patterns.

You might discover:

• Ice cream sales spike when sunglasses sales spike (both driven by summer weather)

• Customers who spend 10+ minutes on your website are 5x more likely to buy

• Machine breakdowns happen more often on night shifts

This is like a doctor noticing your symptoms all point toward one condition.

5. Modeling (Creating the Treatment Plan)

This is where Machine Learning often comes in. You feed clean data into an algorithm to create a model—a mathematical formula that can make predictions.

Examples:

• Predict which customers will cancel their subscription next month

• Forecast how many flu cases a hospital will see this winter

• Recommend which movie you’ll watch next on Netflix

Just like a doctor prescribes treatment based on medical evidence and past cases, a Data Scientist builds models based on historical patterns.

6. Evaluation (Does the Treatment Work?)

Just because a model makes predictions doesn’t mean they’re accurate. You test it on new data to see how well it performs.

If your model predicts 100 customers will leave but only 10 actually do, that’s a problem. Back to the drawing board—just like adjusting medication that isn’t working.

7. Deployment (Putting It Into Action)

Once the model works, you deploy it—meaning you put it into an app, website, or system where it runs automatically in the real world.

Examples:

• Credit card companies use fraud detection models in real-time

• Spotify’s recommendation algorithm runs every time you open the app

• Self-driving cars use models to recognize stop signs

8. Communication (Explaining to the Patient)

The final step—and another underrated one—is explaining your findings to people who don’t speak “data.”

A great Data Scientist takes complex analysis and turns it into a simple, actionable story:

“If we send discount emails to customers who haven’t purchased in 30 days, we’ll recover 15% of them—that’s $200K in revenue. Here’s a simple chart showing the pattern.”

Charts, visuals, and plain English matter just as much as the code. It’s like a doctor explaining your diagnosis so you actually understand and follow the treatment plan.

Data Science in Your Daily Life

You interact with Data Science dozens of times every day, whether you realize it or not. Here are three real-world examples:

1. The “Diapers and Beer” Phenomenon (Retail)

Supermarkets use Data Science for Association Rule Learning (finding what items are bought together).

A famous example: stores discovered that men who bought diapers on Friday evenings often bought beer too.

The Diagnosis: Dad is tired, picking up supplies for the baby, and grabbing a treat for himself.

The Treatment: The store places beer right next to the diapers. Sales go up.

That’s Data Science—finding unexpected patterns and turning them into profit.

2. Uber’s Surge Pricing (Transportation)

Ever wonder why your Uber costs more when it’s raining? It’s not just greed—it’s a Data Science model balancing supply and demand in real-time.

The model predicts:
“It’s raining in downtown. Demand will spike 40%. Supply is low. Temporarily increase price to encourage more drivers to get on the road.”

Just like a doctor adjusting medication dosage based on how your body responds, Uber’s algorithm adjusts prices based on real-time conditions.

3. Who Gets the Loan? (Banking)

When you apply for a loan, a bank officer doesn’t just look at your face. They feed your data—age, salary, credit score, past debts, employment history—into a model.

The model compares you to thousands of past customers:

• If you look like people who paid back their loans → Approved

• If you look like people who defaulted → Rejected

This is Credit Risk Assessment—a classic Data Science application that protects both the bank and borrowers.

The “Confusing” Job Titles

You will hear different job titles thrown around. Here is the cheat sheet:

Data Engineer (The Lab Technician)
Builds the “pipes” and infrastructure to move data from sources (databases, apps, sensors) into storage systems. They make sure data is available, clean, and flowing properly—like a lab tech ensuring all equipment is working and samples are properly prepared.

Data Analyst (The Diagnostician)
Looks at the data to tell you what happened in the past.
Example: “Sales dropped 10% last month in the Midwest region.”

Think of them as the specialist running initial tests and reporting findings.

Data Scientist (The Treatment Planner)
Looks at the data to tell you what will happen in the future or what you should do.
Example: “If we don’t change the price, sales will drop another 15% next quarter. But if we offer a limited-time bundle, we can reverse the trend.”

They’re like the doctor who diagnoses AND prescribes the treatment.

Machine Learning Engineer (The Specialist Surgeon)
Takes models created by Data Scientists and turns them into production systems that run at scale. They ensure the “treatment” works reliably for millions of “patients” simultaneously.

The Takeaway

Data Science is the bridge between raw numbers and real-world decisions.

• AI is the engine.

• Machine Learning is the transmission.

• Data Science is the car, the driver, and the map—getting you to your destination.

It turns the chaos of customer reviews into product improvements.
It turns patient medical records into life-saving diagnoses.
It turns website clicks into personalized recommendations.
It turns noise into knowledge.

The best part? You don’t need a PhD to understand the concepts or use Data Science thinking in your own work. The mindset—asking good questions, looking for patterns, testing ideas with data—is something anyone can learn.

Just like you don’t need to be a doctor to understand when you need medical care, you don’t need to be a Data Scientist to recognize when data could solve your problem.

Was this helpful? Reply and let us know what Data Science concept confuses you the most!

AI for Common Folks — Understand AI in plain English.

Subscribe now

Leave a comment

Deep Learning is a technique where we build complex, layered structures (called Neural Networks) that allow computers to learn from vast amounts of data without us having to tell them what to look for. It’s the technology behind self-driving cars, FaceID, ChatGPT, and voice assistants—allowing machines to see, hear, and create like humans.

Hey Common Folks!

We’ve covered the umbrella (AI) and the engine (Machine Learning). Now, we’re going to talk about the rocket fuel that has made AI the hottest topic on the planet for the last decade: Deep Learning (DL).

If Machine Learning is about teaching computers to find patterns, Deep Learning is about building a “brain” that can find patterns so complex that humans can’t even describe them.

When you hear about a computer beating a world champion at the game Go, or your phone unlocking by scanning your face, or Siri understanding your accent—that isn’t just generic AI. That’s Deep Learning.

Deep Learning vs Machine Learning: What’s the Difference?

The most critical difference comes down to one thing: Features.

Imagine you want to build a system to tell the difference between a Car and a Bus.

Machine Learning (The Manual Teacher)

In traditional Machine Learning, you (the human) have to be the expert. You have to tell the computer specific rules or “features” to look for.

• You tell it: “Look for the number of wheels.”

• You tell it: “Look at the length of the vehicle.”

• You tell it: “Look for the number of windows.”

This is called Feature Extraction. The computer learns the math to separate cars from buses based on the features you gave it. But if you forget to tell it about “height,” it might confuse a tall van with a bus. The computer is limited by your ability to describe the object.

Deep Learning (The Automatic Learner)

In Deep Learning, you don’t define the features. You just throw thousands of pictures of cars and buses at the system.
The system looks at the raw pixels and figures it out on its own.

• It figures out: “Hey, this long rectangular shape usually goes with the label ‘Bus’.”

• It figures out: “These two circular patterns (wheels) spaced far apart mean ‘Bus’.”

It performs Representation Learning. It automatically extracts the features without human intervention.

The Key Difference: In Machine Learning, you tell the computer what to look at. In Deep Learning, the computer figures out what’s important on its own.

How Deep Learning Works: Neural Networks Explained

Deep Learning uses something called an Artificial Neural Network (ANN).

Imagine a corporate hierarchy or an assembly line.

1. The Input Layer (The Entry Level):
This is where the data comes in. If it’s a picture of a face, these are the raw pixels.

2. The Hidden Layers (The Middle Managers):
This is where the magic happens. The data passes through multiple layers of “neurons.”

• The first layer might just detect lines and edges (curves, straight lines).

• The next layer combines those lines to identify shapes (circles, squares, eyes, noses).

• The deeper layers combine those shapes to identify complex objects (a human face).

3. The Output Layer (The Boss):
This layer gives the final decision: “This is a photo of Alex.”

We call it “Deep” Learning simply because we stack many, many of these hidden layers on top of each other. The deeper the network, the more complex patterns it can recognize.

How Does It Actually Learn? The Training Process

Here’s the key thing most people get confused about: Deep Learning still needs a teacher during training.

Think of it like teaching a child to recognize animals using flashcards.

Training Phase (You’re the Teacher):

1. Show the flashcard
You hold up a picture of a dog.

2. Tell them the answer
You say “DOG” out loud. (This is the correct label you provide.)

3. They make a guess
The child looks at the picture and says “Cat!” (Wrong!)

4. Measure how wrong they are (Loss Function)
You say, “No, that’s wrong. The right answer is DOG, not CAT.” The child’s brain calculates how wrong they were. Were they completely off, or kind of close?

5. They adjust their thinking (Backpropagation)
The child’s brain tweaks itself slightly. It thinks: “Okay, pictures with floppy ears and wagging tails are more likely to be dogs, not cats.” Next time they see similar features, they’ll guess differently.

6. Repeat thousands of times
You keep showing flashcard after flashcard. Dog, cat, dog, bird, dog, dog, cat... After seeing thousands of examples WITH your corrections, the child gets really good at recognizing animals.

Testing Phase (They Work Alone):

Now you show them a picture of a dog they’ve NEVER seen before—no label, no help.
The child confidently says “DOG!” ✓

The Deep Learning Process Works the Same Way:

During Training:

• We show it 10,000 pictures of dogs (labeled “DOG”)

• We show it 10,000 pictures of cats (labeled “CAT”)

• The network looks at each picture one by one, makes a guess, gets corrected, and adjusts

• Then it goes through ALL 20,000 pictures again... and again... and again

• Each complete pass through all the data is called an Epoch

• Models typically train for 10-100+ Epochs until they get really accurate

After Training:

• We show it a NEW picture it’s never seen

• It correctly identifies “DOG” on its own

• We don’t give it the answer anymore—it learned the pattern

The Three Steps Happening Inside (For Each Picture):

Step 1: The Guess (Forward Propagation)
The neural network looks at a picture and makes a guess based on its current “knowledge” (the connections between neurons).

Step 2: The Grade (Loss Function)
The system compares what it guessed to the correct answer we provided:

• What it guessed: “CAT”

• What we told it: “DOG”

The Loss Function measures how wrong it was. Think of it like grading a test:

• Totally wrong answer → Big red X → High error score

• Close but not quite → Partial credit → Medium error score

• Perfect answer → Gold star → Zero error

Step 3: The Correction (Backpropagation)
The network takes that error score and works backward through all its layers, slightly adjusting the connections (called weights) between neurons to make a better guess next time.

This loop—Guess, Grade, Correct—happens for every single picture in the dataset.

The Magic Part:

Yes, during training we give it Input (pictures) AND Output (correct labels). The network learns to find the patterns that connect them.

The “automatic feature learning” means we don’t tell it “look for floppy ears” or “look for wet noses”—it figures out THOSE details on its own by examining millions of pixels. But we absolutely DO tell it “this picture = dog, this picture = cat” during training.

Once trained, it can identify dogs in brand new photos without any help.

The Three Types of Neural Networks

Just like there are different types of vehicles for different jobs (trucks for hauling, Ferraris for speed), there are different neural networks for different data:

1. ANN (Artificial Neural Networks):
The basic version. Good for simple data like spreadsheets or numbers.

2. CNN (Convolutional Neural Networks):
The “Eyes” of AI. These are designed specifically for images and videos. They’re brilliant at scanning a photo to find patterns, like identifying a tumor in an X-ray or a stop sign for a self-driving car.

3. RNN (Recurrent Neural Networks):
The “Ears” and “Memory” of AI. These are designed for sequential data like text, audio, or time. They remember what happened previously to understand what’s happening now (like predicting the next word in a sentence).

Where You’re Already Using Deep Learning

You interact with Deep Learning technology every day:

• Face ID / Face Unlock → CNNs recognizing your unique facial features

• Voice Assistants (Siri, Alexa, Google Assistant) → RNNs understanding speech patterns

• ChatGPT and AI Chatbots → Deep neural networks generating human-like text

• Self-driving cars → Multiple neural networks processing camera feeds in real-time

The Takeaway

Deep Learning is the technology that allows computers to perform tasks that we used to think only humans could do—seeing, hearing, and creating.

It’s powerful, it’s complex, and it requires massive amounts of data. But at its core, it’s just a system of layers trying to minimize its own mistakes.

Was this helpful? Reply and let us know what AI term confuses you the most!

AI for Common Folks,
Understand AI in plain English.

Leave a comment

Subscribe now

If AI is the destination, Machine Learning is the vehicle that gets us there. And you’re already using it dozens of times a day without realizing it.

Hey Common Folks!

In our last edition, we learned that Artificial Intelligence (AI) is the big umbrella term for machines acting smartly. But how exactly do they get smart? They don’t just wake up one day knowing how to drive a car or recommend your next Netflix binge.

They have to learn.

Today, we’re zooming in on the most important circle inside that AI umbrella: Machine Learning (ML).

Machine Learning vs Traditional Programming: The Big Shift

To understand why Machine Learning is revolutionary, we need to look at how we used to talk to computers versus how we talk to them now.

The Old Way: Traditional Programming (The Recipe)

For decades, if we wanted a computer to do something, we had to give it a specific “recipe.”
We gave the computer the Input (ingredients) and the Rules (recipe), and the computer gave us the Output (the cake).

Example: If you wanted to write a program to add two numbers, you had to write the rule: If user gives 2 and 2, perform addition. Result is 4.

The problem? You have to write code for every single scenario. If you want a computer to recognize a dog, you have to write rules for tail length, ear shape, and fur color. But what happens when you show it a Poodle after you wrote rules for a German Shepherd? The program fails. You can’t write enough rules to cover the real world.

The New Way: Machine Learning (The Detective)

Machine Learning flips the script. Instead of giving the computer the rules, we give it the Input and the Output (the answers), and we ask the computer to figure out the Rules itself.

The Analogy:
Imagine teaching a child to identify a “dog.” You don’t hand the child a dictionary definition of a canine.

You point to a Golden Retriever and say, “Dog.” You point to a Pug and say, “Dog.” You point to a cat and say, “No, not dog.”

Eventually, the child’s brain spots the patterns—the snout, the paws, the bark—and learns to recognize a dog they’ve never seen before.

This is Machine Learning. We feed the computer thousands of photos (Data) and tell it which ones are dogs (Answers). The machine acts like a detective, finding the hidden patterns that make a dog a dog.

The Three Types of Machine Learning

Not all learning happens the same way. In the world of ML, there are three main ways machines learn. Think of them as different teaching styles:

1. Supervised Learning (The Classroom with an Answer Key)

This is the most common type. We act as the “supervisor” or teacher. We give the computer data that includes the right answers.

How it works: We show the computer data about students—their IQ and their Grades (Input)—and tell it who got a job placement and who didn’t (Output/Answer Key). The computer learns the relationship between grades and getting a job.

Real Life Examples:

• House Prices: Predicting if a house will sell for $500k based on its size and location (This is called Regression—predicting a number).

• Spam Filters: Predicting if an email is “Spam” or “Not Spam” (This is called Classification—sorting things into buckets).

2. Unsupervised Learning (The Solo Explorer)

Here, we throw the computer into the deep end without an answer key. We give it data, but no labels. We say, “Here’s a pile of data. Find the patterns yourself.”

How it works: Imagine you dump a bucket of mixed coins on a table. You don’t need to know the names of the coins to sort them. You can group them by size or color. That’s Unsupervised Learning.

Real Life Example:

• Customer Segmentation: A bank looks at millions of transactions and groups customers into “Savers,” “Spenders,” and “Investors” without being told those groups exist beforehand.

3. Reinforcement Learning (The Gamer)

This is learning by trial and error. The AI is an “agent” placed in an environment. If it does something good, we give it a reward (like a digital cookie). If it messes up, it gets a penalty.

The Analogy: It’s exactly like training a dog. If the dog sits, it gets a treat. If it jumps on the couch, it gets a “No!” Eventually, it learns what to do to get the most treats.

Real Life Examples: Self-driving cars learning not to crash, or robots learning how to walk.

How Machine Learning Actually Works: The Math Behind It

When we say the machine “learns,” it isn’t thinking like a human. It’s using math to draw a line through data.

If you plot points on a graph—say, “Study Hours” vs. “Exam Score”—Machine Learning is essentially trying to draw the best possible line that passes through those points.

Once that line is drawn, if you tell the machine you studied for 5 hours, it looks at the line and predicts your score.

That’s it. It’s not magic; it’s statistics on steroids.

Where You’re Already Using Machine Learning

You interact with Machine Learning every single day:

• Netflix recommendations → Supervised Learning predicting what you’ll watch next

• Spam filters → Classification sorting emails into spam or not spam

• Voice assistants (Siri, Alexa) → Learning to understand your speech patterns

• Amazon product suggestions → Unsupervised Learning finding patterns in shopping behavior

• Self-driving cars → Reinforcement Learning improving through millions of practice miles

The Takeaway

Machine Learning is the shift from telling computers what to do to teaching computers how to figure it out.

• It’s Supervised when we give it the answers.

• It’s Unsupervised when it finds patterns on its own.

• It’s Reinforcement when it learns by trial and error.

Next time Netflix suggests a movie you end up loving, you’ll know: that wasn’t a lucky guess. That was a Machine Learning model acting like a detective, analyzing your history to predict your future.

Coming Up:
We’ve covered the engine (ML), but what happens when we upgrade that engine to mimic the human brain? Next, we dive into the “Deep” end with Deep Learning and Neural Networks.

Subscribe now