AI, in plain words

My dad pointed at the washing machine and asked if its "AI" was the same thing as ChatGPT. Fair question. The honest answer is no, not even close, but I couldn't blame him for asking. The same two letters are stamped on washing machines, on the autocomplete in my code editor, and on trillion-dollar research labs. When a word means that many things, it has stopped meaning much. So let's strip it back.

This is post 1 of 8 in the Foundations series. We start at the word and work down: AI, then machine learning, then deep learning, then the stuff you actually touch every day.

It started as a grant pitch nobody could deliver

The word got coined in 1955, in a funding proposal. A mathematician named John McCarthy wanted money for a summer workshop, the Dartmouth Summer Research Project on Artificial Intelligence, which ran in 1956. The pitch was gloriously naive. Get a few sharp people in a room for one summer and work out how to make machines reason, use language, and solve problems on their own.

They didn't crack it. Nobody has, seventy years on. But the name stuck.

And for the next six decades, "AI" mostly meant research that didn't quite work. Chess programs that eventually got good. Handwriting readers that worked on a clear day. Funding droughts so regular they earned their own name, the AI winters. The word carried a faint smell of overpromising, and mostly it had earned it.

Then one launch rewired the word

ChatGPT went live on November 30, 2022. It hit 100 million users in two months, faster than any consumer product before it. After that, "AI" in normal conversation quietly settled into one meaning: a thing that reads or writes text and answers you like a person would.

That's not the textbook definition. The textbook still says "machines that perform tasks requiring human intelligence." But that's not what someone means when they say they "asked AI to plan a trip." They mean an LLM. Probably ChatGPT, Claude, or Gemini.

That gap is where the trouble lives. "AI-powered" on a landing page might be a 70-billion-parameter language model. It might also be a ten-line if statement somebody decided to call smart. Same sticker, very different box.

So why is everything suddenly AI?

Money, mostly. We're in the post-ChatGPT gold rush, and right now the word itself gets rewarded. Investors like it, customers click it. So your calendar app "uses AI" to suggest meeting times. Your toothbrush "has AI" to grade your brushing. Most of these aren't language models. A good few aren't even machine learning.

The other reason is that the word has no real edges. There has never been an agreed line between "rule-based software" and "AI." The spam filters of 2005 used statistical learning and we called them software. The same code in 2026 gets called AI and nobody blinks. The label just drifts with the mood of the room.

So here's the rule of thumb I use. When a product says "AI," read it as marketing and move on. When you actually want to know what's inside, ask one question. Is it a language model, a smaller machine-learning model, or a hand-written rule? The answer tells you what it can and can't do.

One picture that clears most of it up

Here's the bit that made it click for me. AI is the umbrella. Sit machine learning under it, and deep learning under that. Every "AI" thing you actually use lives at one of those nested layers, never the vague top-level "AI" floating on its own.

That nesting is exactly where post 2 picks up: what each layer really means, and why people swap the three words around even when they shouldn't.

So next time something brags that it's AI-powered, you've got the only follow-up that matters. Which layer? The washing machine, for the record, is a rule. A very confident rule.