Why judging matters more than ever
A chatbot will answer almost any question in a few seconds, neatly laid out and in a calm, confident voice. That changes what is hard.
For most of history, the hard part was getting an answer at all: finding the right book, the right person, the right page. Now getting an answer is the easy part. Knowing whether to trust it is not. That is why we think judging (deciding whether an answer is correct, trustworthy and useful) is the most important AI-ready skill.
Answers got cheap. Judging didn't.
The better AI gets at producing answers, the better you will need to get at judging them. A wrong answer from a chatbot can look just like a right one. There is no hesitation in its voice and no warning sign in its layout.
Producing an answer now costs almost nothing. Trusting a wrong one can still cost a lot, as the lawyers in Why AI gets things wrong found out.
Try this: the next time you get an AI answer you are going to use, stop and ask yourself one question: how would I know if this were wrong? If you have no idea, you are not ready to rely on it yet.
The more you lean on it, the more judging matters
The students in the maths study on Use AI as a tutor weren't lazy. They accepted answers they hadn't properly judged, including the most important judgement of all: do I actually understand this now, or does it just look finished? A neat explanation can feel like understanding when it isn't, and the test is where you find out.
Try this: after a chatbot explains something to you, close the chat and explain it back in your own words, out loud or on paper. If you can't, you haven't understood it yet. You've only seen an explanation.
What judging actually involves
Judging isn't one big skill. It's a set of steps you can practise.
- ✓ Break it down. You can't properly check an answer if you only look at the final result. Break it down lets you practise splitting a problem into smaller logical steps.
- ✓ Check that each step follows. An answer can contain correct calculations and still reach the wrong conclusion if the reasoning has a gap. Checking the answer gives you worked answers to judge.
- ✓ Check the facts elsewhere. Names, numbers, dates and sources need to be checked against a reliable source other than the chatbot. Fact-check what AI tells you shows how.
- ✓ Ask what shaped the answer. Even when an answer is factually correct, ask: whose perspective does it reflect? What information might be missing? What would another reliable source say? Why AI gets things wrong explains where bias can come from.
Try this: pick one of the practice pages and do five problems today. Judging improves the way any skill improves: by practising, getting things wrong sometimes, and noticing why.
The short version
Getting answers is no longer the hard part. Knowing which ones to trust is. Start with one habit this week: before you use an AI answer, ask how you would know if it were wrong.
Last reviewed: 25 September 2026