Hacker News new | past | comments | ask | show | jobs | submit login

I'm guessing they relied on an LLM response. That was my thought, and having tried it they indeed generate lots of garbage for this topic. I got a ChatGPT A/B test for this and both options were incorrect (one obviously and the other subtly, due to misinterpreting a bond rating page's discussion of the PUF and just blindly regurgitating the number from there).





Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: