Hacker News new | ask | show | jobs
by lurking_swe 467 days ago
except we can guarantee (with tests) that the generated instructions from the compiler are bug free 99% of the time. Pretty big difference there.
1 comments
sanswork 466 days ago
The tool currently being bad doesn't matter to the argument of whether or not they make devs dumb.
link
lurking_swe 466 days ago
if i can one day have a similar level of confidence in LLM output, as i do in a compiler, then i will also call it an abstraction. until then…i wait. :)

But i agree, it’s unrelated to devs being dumb.

link