laanwj on Nostr: impressive, yes but also limited for real-world use cases, for example the issues ...
impressive, yes
but also limited for real-world use cases, for example the issues with hallucination make it absolutely terrible for technical documentation or critical software projects
like sure you can write software faster with a LLM, but then someone has to maintain it and fix the bugs 😅
a lot of impressive examples in papers and such, are cherry-picked from a lot of slop
remember the hype about self-driving cars? there seemed to be a lot of progress in a short time, then it just stopped: driving a vehicle in the real world conditions is much more involved than expected
but also limited for real-world use cases, for example the issues with hallucination make it absolutely terrible for technical documentation or critical software projects
like sure you can write software faster with a LLM, but then someone has to maintain it and fix the bugs 😅
a lot of impressive examples in papers and such, are cherry-picked from a lot of slop
remember the hype about self-driving cars? there seemed to be a lot of progress in a short time, then it just stopped: driving a vehicle in the real world conditions is much more involved than expected