Short YouTube video (presenter Kyle Balmer) advising viewers not to accept the first ChatGPT response without follow-up. Uses a family anecdote from a UK bank-holiday to show people often stop after one reply—even when it's wrong—and emphasizes multi-shot prompting: ask clarifying questions, point out mistakes, and iterate until the model improves.
Chapters
This is probably the biggest mistake you can make when using ChatGPT. Here in the UK it's bank holidays, we have a three day weekend and I saw some family. At one point ChatGPT came up and a family member was using it in order to query something. What I noticed was they would ask ChatGPT a question, get the answer from ChatGPT and then leave it at that. This also included when ChatGPT gave a bad answer, they would just close it down and be like ah, well it didn't work this time. The main advantage of ChatGPT over something like Google is that you can go back to it. You can say oh I didn't like that part of your answer, or that bit didn't make sense. Can you explain more to me? The fact that we can multi shot prompt, we don't need to do it in one go, but we can go back and have a discussion with ChatGPT is really what makes it unique. So if you are not happy with the result that ChatGPT has given you, ask again. Tell it exactly what it got wrong and what it needs to do better and I promise you it will.