nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqxksp9jnwt3d9a5m2wmvjm0ag64vesa304ztgd6k3w2g7lk2r69nqel0kf8 I read the news about this with an air of "oh look, another 'LLM AI is evitable so let...
@nprofile1q... @nprofile1q...
I feel the important bit is that it transfers the onus from Tovards to the contributors.
Ethical and environmental questions aside¹, the problem with LLM's is users who outsource their thinking, leaving others to verify usability. As verification is 90% of the work, the end-result is glorifying those who offload their work not just LLM's, but their colleagues.
Now if the code is bad (and LLM code usually is), it's the coder who's bad.
_
¹ which I don't belittle.
I feel the important bit is that it transfers the onus from Tovards to the contributors.
Ethical and environmental questions aside¹, the problem with LLM's is users who outsource their thinking, leaving others to verify usability. As verification is 90% of the work, the end-result is glorifying those who offload their work not just LLM's, but their colleagues.
Now if the code is bad (and LLM code usually is), it's the coder who's bad.
_
¹ which I don't belittle.
1
