Hey everyone! I'm a bit of a newbie here, and I'm looking for some advice on how to get ChatGPT to recognize when it's not being factually accurate. I've tried prompts like telling it to value factual accuracy and to explicitly mention when it's unsure about something. Unfortunately, I haven't seen much improvement. Ideally, I'd love for it to provide footnotes at the end of its responses to highlight any potentially inaccurate or unverified information. What do you all think? Any tips or tricks?
2 Answers
Honestly, ChatGPT often hallucinates and doesn't really have a grasp on reality. While regular updates can help somewhat, it’s built to generate text based on patterns, which means it can create details that aren’t accurate. Just keep that in mind when using it!
You might have better luck with a model that focuses on reasoning. Those usually catch mistakes more often, but they’re not perfect either. Just keep experimenting!
Totally! I learned the hard way too—had it send me bios for a conference, and it made up titles and affiliations completely out of nowhere. Glad I double-checked!