Getting facts wrong: a generative AI demo

When ChatGPT first burst onto the scene and rumours started to spread that it could make mistakes, I came up with a simple fact that I could use to verify those rumours. Several years later I still find ChatGPT and other LLMs struggling with a simple piece of information that is relatively easy to verify, for example by using a traditional search engine. I have used this experience as an introduction —hopefully an entertaining one — when giving talks on AI and cybersecurity. This video is taken from one of those. I hope you enjoy it and find it useful if you know people who may be putting too much faith in AI output.