Everyone claiming that generative models' (dubious) capability with software is transferable to other realms is making an unsubstantiated claim.
-
Everyone claiming that generative models' (dubious) capability with software is transferable to other realms is making an unsubstantiated claim. But then again, it's just another permutation of engineer's disease.
Coding is the easiest thing for a language model to do.
-
Everyone claiming that generative models' (dubious) capability with software is transferable to other realms is making an unsubstantiated claim. But then again, it's just another permutation of engineer's disease.
Coding is the easiest thing for a language model to do.
@mttaggart "Claude crashed the plane, but here's how that's good, actually." — every mouth-breathing Hacker News poster in six months.
-
Everyone claiming that generative models' (dubious) capability with software is transferable to other realms is making an unsubstantiated claim. But then again, it's just another permutation of engineer's disease.
Coding is the easiest thing for a language model to do.
It's like this: between all examples of, say, a contract in the training corpora, what would you imagine the variance is between them? Consider variance as diction, syntax, and even conceptual structure.
Now consider the variance between all examples of, like, a login process.
It's so, so much harder to distill not-code text to produce useful output in similar shapes.
But of course, the insecure nature of the example code also guarantees a tendency to insecurity in model output as well.
-
R ActivityRelay shared this topic
