Guides: using AI without being fooled
Short, practical answers to the questions people ask about trusting ChatGPT, Claude and Gemini.
How to spot AI hallucinations
To spot an AI hallucination, look for over-specific details, numbers that do not add up, sources you cannot open, and answers that change when you ask again. Then check one claim yourself: ask the AI how sure it is, click its sources, and make it show its working for any numbers.
How to check AI answers
Check AI answers with three habits. Ask "how sure are you, out of ten, and what would make you wrong?". Make it show every step and use a calculator or code for numbers. Click every source and find the line it claims to support. Match the effort to the stakes.
What not to paste into ChatGPT
Do not paste passwords, API keys, client details, confidential contracts, medical information or ID numbers into an AI chatbot. A simple test: would you read it out loud to a stranger on a bus? If not, swap sensitive details for placeholders like "Client A", and check your app's training-data setting.
Why ChatGPT agrees with everything you say
ChatGPT and other chatbots tend to agree with you too easily. To get honest feedback, ask the AI to argue against your idea, name its weakest part and say what a sceptic would object to. Describe options neutrally, and start a fresh chat if the conversation has become one-sided.
The 30-Day AI Habit: the full plan
The 30-Day AI Habit is a 15-minute-a-day course in four weeks: Stop the Mistakes, Better Answers, Win Back Time and Your AI Routine. Each day has one rule and one small task, plus quizzes that train you to spot errors in confident AI answers. Missing a day does not restart the plan.
Is ChatGPT accurate? How much should you trust AI answers
ChatGPT and other AI assistants are useful but not reliably accurate. A BBC- and EBU-led study of more than 3,000 AI answers to news questions found 45% had at least one significant issue and 31% had serious sourcing problems. Treat answers as drafts, and verify anything involving money, health, law or public claims.
How to make ChatGPT sound like you
To make ChatGPT sound like you, give it a brief (who you are, what you need and why), paste a sample of your own writing and ask it to mirror that voice, then ask for a second draft as a tough editor. Tell it the shape of the answer you want before it starts.
How to use AI for email, meeting notes and summaries
Use AI for the first draft of tricky emails, to turn messy meeting notes into action items, and to summarise long documents. Make it flag gaps instead of inventing them, check any summary against the original, and save your best prompts as fill-in-the-blanks templates you can reuse.
Does AI really save time? How to measure it honestly
To find out whether AI saves you time, time one real task twice: the old way, and the AI way including the time it takes to check the work. Compare the totals. If you only count the AI's typing speed you are fooling yourself, because checking is part of the job.
AI glossary: plain-English definitions
This glossary defines the key terms for using AI safely: AI hallucination, cognitive surrender, sycophancy, prompt, custom instructions, training on your conversations, placeholder, source click, pre-flight check, fresh chat, prompt library and spaced repetition.
Ready to build the habit? The 30-Day AI Habit is a 15-minute-a-day course with one rule and one small task each day. Open the course