You ask a chatbot a simple question. The answer comes back in seconds: clear, polite and completely confident. But is it right?
That is the question most of us now face several times a week, often without noticing. AI drafts our emails, summarises letters, answers questions about bus times, bills and symptoms, and increasingly sits behind the decisions banks, employers and shops make about us. The tools are genuinely useful. They are also very good at sounding right when they are not.
The good news is that you do not need to understand how AI is built to use it sensibly. You need a quick habit for deciding how far to rely on any one answer.
Why a confident answer is not the same as a correct one
A chatbot produces the kind of answer that a correct answer would sound like. Often that is also what the correct answer is. Sometimes it is not, and the tone gives you no warning either way.
Two real cases show what that can cost. In Mata v. Avianca (New York, 2023), lawyers filed a court document citing earlier decisions that did not exist. They had been produced by ChatGPT, and when the lawyers asked whether the cases were real, it said they were. The court fined the lawyers and their firm $5,000.
In Moffatt v. Air Canada (2024), an airline's website chatbot told a bereaved customer he could claim a discount after travelling. That was not the airline's policy. The airline argued the chatbot was responsible for its own words; a tribunal in British Columbia rejected that argument and ordered compensation.
If careful professionals and a major airline can be caught out by fluent answers, the rest of us can too.
Five questions to ask before you rely on an AI answer
These take seconds once they become a habit.
- What happens if it is wrong? A poor film suggestion costs nothing. A wrong answer about medicine, money, the law or safety can cost a great deal. The higher the stakes, the more checking an answer needs.
- Could it actually know this? Be especially careful with anything current (prices, timetables, news), local (your area, your GP surgery), personal (your account, your case) or private (records an organisation holds).
- Can I check it? The best uses of AI are often ones you can confirm quickly, such as checking a bus time on the operator's own website.
- Where did this come from, and does that source exist? If a chatbot gives a source, open it. Make sure the page, report or law is real and says what the chatbot claims.
- Would I accept this from a confident stranger? Imagine the same answer from someone at the bus stop who sounds very sure. If you would check it first, check the chatbot too.
Match your checking to the stakes
- Low stakes and easy to check (rewording a letter, recipe ideas): use AI freely, then read the result before you use it.
- Moderate stakes or hard to check (travel plans, document summaries): treat the answer as a starting point and confirm the key facts from the original source.
- High stakes (health, money, legal rights, safety, anything sent in your name): use AI to prepare your questions, then let an official source or a qualified person settle the answer.
One trap to avoid: asking the chatbot "Are you sure?" is not a check. It may simply give you a different answer with the same confidence. If the answer matters, verify it somewhere else.
When the voice on the phone sounds like someone you love
The same habit protects you from AI-enabled scams. Voices and faces can now be faked convincingly, so a familiar voice is no longer proof of who is calling. If you get an urgent request for money or codes, even from someone you know, hang up and call back on a number you already have. In the UK you can call 159 to reach your bank's fraud team safely. Agree a family code word in person, and never write it down online.
A plain-English guide for the rest of us
AI: Should I Trust It? One Street, Six Lives, and What AI Means for You by Edward Omoba follows six neighbours through a year of meeting AI at work, at home, at the bank, at the doctor's and on the phone. Their stories show where AI genuinely helps, where it misleads or excludes, and what ordinary people can do about it.
The 159-page Premium PDF covers checking AI answers, challenging automated decisions, stopping voice-clone and deepfake scams, and your rights in the UK, the US, the EU and beyond. It includes a printable Scam and Impersonation Action Sheet, a Family AI Agreement and a Personal AI Decision Sheet.
Explore the guide, view real sample pages and get the PDF download.
General information for ordinary readers. This article and the book are not legal, financial, medical or professional advice. Laws and services change; check the current position with an official source or a qualified adviser.