Why does AI make things up, and how do you catch it?
Chatbots sometimes invent facts, quotes and sources, and say them with total confidence. Here is why it happens, how often it happens, what it has cost people, and a simple routine for catching it.
Read the full answerAI chatbots make things up because they are built to produce likely-sounding text, not to look facts up, and their training rewards a confident guess over admitting they don't know. Measured error rates vary widely by task, from a few per cent when summarising a document to a third or more on hard factual questions. Check any name, number, quote or source that matters before you rely on it.
What is superintelligence, and who is worried about it?
Tech leaders say it is coming, some scientists want it banned, and the US government now uses the word for all AI. Here is what superintelligence actually means and why people disagree about it.
Live with itKids, teens and AI chatbots: what should parents know?
Most teenagers already use AI chatbots, often for homework. Here is what the age rules say, which parental controls exist, where the real risks lie, and what parents can do this week.
UnderstandHow does an AI chatbot actually work?
You type a question and fluent text comes back in seconds. Knowing what happens in between explains why chatbots are so useful, why they vary, and why they still get things wrong.
SI Observatory4 October 2026 · Elon Musk says SpaceXAI will be renamed SpaceXSI29 September 2026 · Six AI companies sign the voluntary White House Accord on Super Intelligence
Find your way
Understand
Plain-language explainers on how AI works and what the words mean.
SectionUse
Practical guides and tools, tested by a person before we recommend them.
SectionLive with it
What AI changes for work, school, health, money, privacy and your rights.
SectionWhat's new
A short, curated brief on the developments that actually matter.
SectionSuper Intelligence
The new US term, the frontier labs, policy and the long-term debate.