

In my experience even simple LLM changes are wrong somewhere around half the time. Often in disturbingly subtle ways that take an expert to spot.
I just want to add: sailor’s reference to “expert” here is no joke. the amount of wild and subtle UB (undefined behaviour) you get in the C family is extremely high-knowledge stuff. it’s the sort of stuff that has in recent years become fashionable to describe as “cursed”, and often with good reason
LLMs being bad at precision and detail is as perfect an antithesis in that picture as I am capable of conceiving. so any thought of a project like this that pairs LLMs (or, more broadly, any of the current generative family of nonsense) as a dependency in it’s implementation is just damn wild to me
(and just incase: this post is not an opportunity to quibble about PLT and about what be or become possible.)




ah, a fellow 1937 poster that gets it
(I still don’t understand how people that post on/across the modern internet haven’t picked up the habit)