juraj on Nostr: When AI large language models act like politically correct pussies, it is not just ...

When AI large language models act like politically correct pussies, it is not just annoying. It is a worse service. Example: I told it to summarize my lecture about crypto volatility. Instead of pure summary, it added a lecture on cryptocurrency risks, volatility and possibility of loss to every summary (I summarize per parts). So the summary is unnecessarily long and contains information that was not in the original text. This then degrades the resulting summary of the text.

When I tell a language model to summarize something, they should take the content and make it shorter, not lecture me on risk.

We need normal models. And we need Sam Altman and his cronies to leave us alone.

(Yes, I cover the risks in the lecture, in another part)