How AI chatbots actually work, and why it changes how you use them
No maths, no jargon. What's really going on when you chat with ChatGPT, Gemini or Claude, and the handful of habits that follow from it.
4 min readThere's a lot of hype about AI, and a lot of myths. Here are ten statements people often get wrong. Decide whether each is true or false, and we'll show you the evidence.
Question 1 of 10
AI chatbots answer by looking facts up in a giant database.
Answer: False. False. OpenAI explains that the model learns patterns from large amounts of text and predicts 'the most likely next piece of language based on context'. That's why it can sound right and still be wrong.
A scammer can copy someone's voice from just a few seconds of audio.
Answer: True. True. Starling Bank says criminals can clone a voice from as little as three seconds of audio. A family safe phrase is the simple fix.
Most people can tell a deepfake video from a real one if they look carefully.
Answer: False. False. When Which? showed 514 people a mix of real and deepfake videos, 70% didn't correctly identify all the ones they saw. Asking who posted it beats squinting at the pixels.
An AI chatbot always knows about yesterday's news.
Answer: False. False. A model only learns from text up to its knowledge cutoff, and doesn't always choose to search. The Columbia Journalism Review described ChatGPT dismissing a major news story hours after it happened with 'That didn't happen.' Ask it to search for anything recent.
More than half of UK children aged 8 to 17 use AI.
Answer: True. True. Ofcom found that 56% of 8 to 17-year-olds in the UK use AI.
You have to be at least 18 to use Claude.
Answer: True. True. Anthropic's consumer terms say you must be at least 18, or the minimum age to consent where you live if that's higher. Other assistants set different ages.
People paying for ChatGPT Plus see adverts in their chats.
Answer: False. False. OpenAI says ads are for logged-in adults on the Free and Go plans, and that paid plans such as Plus and Pro don't have them.
Switching off training in Copilot stops Microsoft using your chats for anything else.
Answer: False. False. Microsoft says opting out stops future chats being used to train its AI models, but not their use for other product improvements, advertising, safety, security and compliance.
If you delete your Gemini activity, every copy of every chat is gone.
Answer: False. False. Google says chats already picked for human review are kept for up to three years, even if you delete your activity. That's why it asks you not to enter anything confidential.
In a very long chat, an AI can lose track of what you said at the start.
Answer: True. True. There's a limit to how much text a model can take in at once, called its context window. Past that limit, earlier parts of the conversation effectively drop out of view. Start a fresh chat when you change topic.
We checked these on 9 November 2026 and review this quiz by 9 December 2026.