What is AI?

You have probably used AI a dozen times today, and called it none of those times "AI". You unlocked your phone with your face. Your inbox quietly binned three junk emails before you saw them. Maps sent you a different way to work because a lorry had jackknifed on the ring road. None of that felt like the future. It felt like Tuesday.

That is the first thing to know about AI: most of it is invisible, and most of it is already here. The chatbots grab the headlines, but AI has been doing quiet, useful work in your pocket for years. This lesson is about seeing it clearly: what the word actually means, where it is already helping you, and what is coming next.

What "AI" actually means

Strip away the hype and there is one simple idea underneath.

Old software follows rules a person wrote. A human sits down and types out every step: if the total is over 100, apply the discount; if the date is past due, flag it red. The computer does exactly what it is told, no more, no less. This runs the world, and it is not AI. It cannot handle anything nobody thought to write a rule for.

AI learns patterns from examples instead. Nobody can write a rule for "is this a photo of a cat". Where would you even start, whiskers? So we flip it around. Show the software a few million labelled photos, let it work out for itself what "cat" tends to look like, and it gets good at spotting cats it has never seen before. It was never told the rule. It learned the pattern.

That is the whole trick, and it is worth saying plainly: AI is software that learns from examples rather than being handed every rule. Everything else is detail. You do not need maths to get this. You have done the same thing your whole life: you know a friend's handwriting without ever writing down a rule for it.

The AI you already use

Here is the fun part. You do not need to sign up for anything to start noticing AI, because you are surrounded by it. A normal day, roughly:

Your face unlocks the phone. The camera does not match a stored picture pixel for pixel. It learned the pattern of your face and recognises it in bad light, with a new haircut, half asleep. That is AI.

Netflix, Spotify, and YouTube know what is next. "Because you watched", the playlist that somehow fits your mood, the video that autoplays and is annoyingly perfect. All of it is a model that learned your taste from what you (and millions of others) played, skipped, and finished.

The spam filter. Nobody wrote a rule listing every dodgy email. The filter learned what junk looks like from billions of examples, and it keeps learning as the junk changes. It feels like nothing, which is exactly the point.

Maps reroutes you. It reads live traffic, predicts how long each road will take, and quietly picks the fastest one. The ETA that is eerily accurate is a prediction, not a lookup.

Your camera takes a good photo in the dark. Night mode, portrait blur, the shot that looks better than the scene did. The phone is not just capturing light, it is using AI to fill in and clean up what the tiny lens could not catch.

Your bank texts you about a weird charge. A model watched your normal spending, spotted something that did not fit, and flagged it in real time. Same idea keeps your card working abroad instead of freezing at the worst moment.

Autocorrect and predictive text. Every time your phone guesses the next word, finishes your sentence, or fixes "teh", a small language model is at work. It is a cousin of the big ones you will meet later in this course.

None of this asks you to do anything. It just runs. Which is why so many people who say "I have never used AI" use it before breakfast.

So why does the new wave feel different?

If AI has been around this whole time, why does ChatGPT and Claude feel like something changed? Because the AI above mostly sorts, scores, or predicts: is this spam or not, how long is this drive, which video next. Useful, but narrow. Each model does one job.

The new wave is generative AI, and the clue is in the name: it makes new things. Give it a blank page and it writes the email, drafts the report, builds the table, answers the question in full sentences, even writes working code. It does not pick from a list. It produces something that was not there a second ago.

That is the leap. For the first time, the AI does not just tidy your world behind the scenes, it sits across the desk and does work with you. This whole course is mostly about that kind of AI, and about Claude in particular. Exactly how it manages the trick (models, training, why it sometimes makes things up) is the next lesson. For now, one line is enough: generative AI creates; the older AI mostly sorts.

How the words fit together

People throw around "AI", "machine learning", and "generative AI" as if they were the same thing. They are not, but they are not rivals either. They are nested circles, one inside the next.

Artificial intelligence is the widest circle: any software doing something we would once have called clever. Machine learning sits inside it, the learn-from-examples approach behind nearly all AI worth talking about today. Generative AI sits inside that, the newest and most exciting corner, where the software does not just recognise patterns but uses them to create. Claude, ChatGPT, and Gemini all live in that innermost circle.

You do not need to police these words in the office. But knowing the shape stops you nodding along to nonsense, and helps you ask the one question that matters at work: what is this actually doing, and can I trust it with this task?

What is coming next

AI moves fast, so any list dates quickly. But two waves are worth picturing: the near ones you will likely see soon, and the bolder ones a bit further out.

Coming soon, and grounded. The clearest shift is from AI that answers to AI that acts. Today you ask Claude a question and it replies. Increasingly it will carry out a whole task for you: book the trip, reconcile the invoices, pull the numbers from three systems and draft the report, then hand it back for a check. These are often called agents, and Part 5 of this course gets you using them. Alongside that: live translation good enough to hold a real conversation across languages through an earbud, medical screening that spots a problem on a scan earlier than a tired human eye, and a quiet copilot in every work tool you already use, drafting, summarising, and catching mistakes as you go.

Bolder, and further out. Look past the next couple of years and the picture gets more sweeping. Cars that genuinely drive themselves on ordinary streets, not just on a demo route. A tutor for every child that adapts to exactly how that child learns. Drug discovery compressed from a decade to a year or two because AI can sift possibilities no lab could test by hand. And, more slowly than the headlines claim, AI that reasons across many fields at once rather than being sharp at just one.

A fair word of caution, because this course respects your time and your judgement: the near list is close to certain, the far list is not. Some of it will arrive early, some late, some in a shape nobody predicted. The useful stance is neither breathless nor dismissive. It is curious: learn what the tools can do today, and you will spot the real changes coming before the hype does.

Why the words matter at work

You will sit in meetings where "we should use AI for this" gets said with great confidence and no detail. Being able to ask "which kind, and what would it actually do?" is quietly powerful. It is the difference between chasing a buzzword and solving a problem.

That is the real aim of this first part of the course. Not to turn you into an engineer, but to give you a clear enough map that you can make good calls: where AI genuinely helps, where it does not, and what is safe to trust it with. Everything after this builds on that.