When it comes to AI, trust, but verify
For the time being, at least, when using ChatGPT and other AIs, we need to remember a simple rule: trust, but verify.
I dislike the political baggage associated with the phrase, but there's not much I can do about that. It's good advice.
As a curious person, ChatGPT is an incredible resource. When I want to debug a computer programming problem or get into the weeds of a philosophical issue, I often start by engaging with the chatbot. I've learned important things this way, but I've also noticed ChatGPT making major mistakes. In one particularly bad “hallucination,” as they're called, ChatGPT invented a horrific quote and attributed it to someone who said no such thing.
These things happen, and technologists don't currently have a solution. For that reason, I strongly recommend double-checking any important claim made by one of these mechanical minds. Trust, but verify.