Free version = freemium, like Spotify: paid subscriptions and investors foot the bill.
When today we hear “artificial intelligence”, we mean programs able to do
things that used to require a person: understand a piece of text, recognize
a voice, suggest the next word as you type a message, answer a question.
It isn’t a single technology. It’s a label that lumps together many very
different things.
In recent years “AI” has become, for almost everyone, a synonym for one
specific kind: the ones you can hold a conversation with, usually by
typing, like ChatGPT, Claude, Gemini.
You care about understanding it because it is already in your pocket. When
your phone completes a word for you while you type a message, when Google
Maps suggests a route that’s “three minutes faster”, when Netflix recommends
your next film, when spam ends up in the trash without you moving it: there
is some form of AI behind it. Most of these are boring but useful things
that have been saving you time for years.
All of this happened before ChatGPT came along. At the end of 2022, it
became the first AI product anyone could use, with no preparation, to have
a real conversation about any topic. That was the moment AI walked out of
research labs and into everyone’s hands.
When someone says “AI” in a practical context, they usually mean one of
these two things:
AI that does one thing well. The spam filter in your inbox, voice
recognition in Siri or Alexa, your keyboard’s autocomplete, photos that
recognize faces. These have been around for years, and they help you in
ways you no longer notice. Each one is specialized: Siri understands you
when you ask for a timer but it can’t write a poem, the spam filter
doesn’t know how to chat, autocomplete has no clue who your contacts are.
AI that talks (conversational). ChatGPT, Claude, Gemini, Copilot
and similar. Born in the last few years, these are the ones that made
noise. When the newspaper writes “AI”, almost always it means these.
In this guide we focus on the second. The reason is that this is the
category that has changed the most about what you can do today, without
being a developer: ask for information, get help writing, summarize
documents, learn new things at your own pace.
AI is surrounded by misconceptions. Some come from movies, others from
sensationalist headlines. Here are the four most common ones, out of the
way.
It doesn’t “think” like a person. It has no experiences of its own.
By default each conversation starts from zero, even if you talked to it
yesterday (some products add an optional An optional feature in some products that lets the AI remember things across chats. By default each new conversation starts from zero. feature, which we’ll
cover later). When it sounds convincing, it is doing something very
different from what your brain does.
It isn’t “always right”. It can make mistakes and it can make things
up, and when it does, it does so with the same confidence as when it
says true things. Recognizing when it’s getting things wrong is one of
the things you’ll take home from this manual.
It doesn’t “want” anything. It has no desires or intentions. It
responds to what you write. It isn’t judging you, it has no plans for
you. (A separate question is “but where does what I write go?”: that’s
about the company behind the AI, not the AI itself, and we cover it in
a dedicated lesson.)
It isn’t magic. It is a very sophisticated program, made by people,
with real limits and real costs (for whoever runs it).