|
Top-level
 LLM are not reliable enough to "check facts" this isn't what they are even designed to do well. What they are designed to do is generate plausible seeming streams of text similar to existing sets of text. That is all. There is no logic behind that, no verification. It's pure chance. Do not use them to check facts, please. 3 comments
 @futurebird by portraying LLMs as a panacea, the hypemongers have successfully kneecapped a lot of people's ability to consider alternative tools that are often well known, widely used for decades, and solve the specific problem they're targeting with a 0% fail rate. i've had to tell project managers way too many times, "no, we don't need to pay OpenAI to fuck up at doing what regex does perfectly. all we're trying to do is extract a consistently-formatted string for fucks sake." |
Gregory's Server
@futurebird
@ronaldtootall @hannu_ikonen
yeah LLMs aren't at all reliable for any of the use cases he proposed. 🙄😮💨 literally any modern word processor is more reliable for spelling/grammar checks, and the suggestion of using them for research is always nauseating.