8 minute read
AI agents
What is an AI agent, and can it actually do things for me?
An agent is a model in a loop with tools. What that gets you, what it does not, and why the word is oversold.
Explained properly
The bit everyone skips.
The courses teach you to use AI well. These explain what is actually happening behind the button, and what it costs — in plain English, with the sources, and with an honest “nobody knows” where that is the truth.
9 minute read
Tokens, and what AI actually costs
What is a token, and why am I being charged for them?
Why AI is billed by the token, what that means in real money, and why the same question can cost a penny or a pound.
8 minute read
Why it makes things up
Why does AI invent things, and why does it sound so sure?
The mechanism behind a confident wrong answer — not just the warning that it happens.
7 minute read
Context windows
Why does it forget what I told it five minutes ago?
What a context window is, why a tool loses the thread, and why bigger is not automatically better.
8 minute read
Where your data actually goes
If I paste something into an AI tool, who can see it?
Training versus answering, what retention means, and what “we do not train on your data” does and does not promise.
7 minute read
Free versus paid
Is the paid version actually worth it?
What you really give up on a free tier, and the handful of situations where paying earns its money.
8 minute read
How a model is trained
How does an AI model actually learn anything?
Pre-training, fine-tuning and human feedback — the three stages, and what each one does to the thing you end up talking to.
7 minute read
Chatting with your own documents
How does it answer questions about a file I uploaded?
What retrieval actually does, why it reduces invention without removing it, and the failure that looks most convincing.
7 minute read
Reasoning models
What does it mean when a model says it is thinking?
What the thinking actually is, when it earns its extra cost and time, and when it is simply slower.
7 minute read
Open and closed models
What does an “open” AI model actually mean for me?
What open weights give you, what the word hides, and when any of it matters to an ordinary user.
7 minute read
Running AI on your own computer
Can I run one of these on my own laptop?
What is realistic on ordinary hardware, what it costs in practice, and the good reasons to bother.
7 minute read
Pictures, sound and video
It can look at photos now. How good is it actually?
What changes when a model takes images, audio or video — and the specific ways it still gets them wrong.
8 minute read
What AI costs the planet
Is using AI bad for the environment?
What actually uses the energy, why the per-question figures you have seen are unreliable, and what genuinely reduces the impact.
8 minute read
Scams and deepfakes
How do I know if a video or a voice is real?
What has become cheap to fake, what the old advice gets wrong now, and the checks that still work.
8 minute read
Copyright and training data
Did they have the right to train on all that? And who owns what it makes?
Two unsettled questions, what is actually known about each, and what that means for you in practice.
7 minute read
Bias in AI
Is AI biased, and would I notice if it were?
Where bias comes from, what it looks like in ordinary use, and why the confident tone makes it harder to spot.
8 minute read
AI and jobs
Is AI going to take my job?
What is actually changing in work, without the doom or the boosterism, and what is worth doing about it.
7 minute read
The rules, and who makes them
Is any of this actually regulated?
What already applies to AI in the UK, what is genuinely new, and what it means for an ordinary user.
8 minute read
Making a picture that never existed
How does it draw something from a sentence I typed?
How picture tools actually work, what they are still bad at, and whose work they learned from.
7 minute read
Somebody could copy your voice
Could a scammer really sound like my daughter?
How little recording it takes, what it is genuinely useful for, and the one rule that works whether or not a call is fake.
7 minute read
Video you cannot take at face value
That clip looks real. How would I even know?
What generated video can do now, why the giveaways keep expiring, and what to check instead of squinting.
6 minute read
Who wrote the song it just made?
It wrote me a song in thirty seconds. Is that allowed?
How music tools work, what they are genuinely good for, and the argument about what they learned from.
Looking for a single word rather than a whole piece? The glossary has every term in plain English, with a link to the explainer that covers it properly.
Trying to decide between the big models rather than understand one? Which one should I use? walks through the handful of differences that actually change what you can do.
More will follow. If there is something about AI you have never had explained properly — or somewhere this section is wrong — that is exactly what it is for, and worth telling us about.