Hacker News Reader

Comment by halJordan | Hacker News Reader

What actually insane is what assumptions you allow to be assumed. These non sequitors that no human would ever assume are the point. People love to cherry pick ones that make the model stupid but refuse to allow the ones that make it smart. In compete science we call these scenarios trivially false, and they're treated like the nonsense they are. But if you're trying to push ant anti ai agenda they're the best thing ever

View on HN

> People love to cherry pick ones that make the model stupid but refuse to allow the ones that make it smart.

I haven't seen anybody refuse to allow anything. People are just commenting on what they see. The more frequently they see something, the more they comment on it. I'm sure there are plenty of us interested in seeing where an AI model makes assumptions different from that of most humans and it actually turns out the AI is correct. You know, the opposite of this situation. If you run into such cases, please do share them. I certainly don't see them coming up often, and I'm not aware of others that do either.

The issue is that in domains novel to the user they do not know what is trivially false or a non sequitur and the LLM will not help them filter these out.

If LLMs are to be valuable in novel areas then the LLM needs to be able to spot these issues and ask clarifying questions or otherwise provide the appropriate corrective to the user's mental model.

Reply on news.ycombinator.com