Post #2468774
2026-03-24 10:13 UTC
@Ember@blobfox.coffee Personally, my problem with AI isn't that it always produces bad results – I think that's not true. Under proper guidance, AI can be very helpful for software development. Just like we see a bunch of examples of blatant 'slop', there are also a lot of cases where AI is quietly used by skillful developers, with all of the results thoroughly reviewed by a human, and that results in overall better outcomes, technically speaking.
My problem with it is mostly ethical – even if we ignore the environmental impact and the use of AI in armed conflict... the endgame of big AI companies is obviously to de-skill the software developer job and make themselves an integral part of most software companies by renting the coding "skills" out to them at a high price.
I think if you don't care about the ethics, if you're skillful enough to review the code and if you're only interested in churning out working software faster than your competitors – you can definitely achieve that by using AI.
Replies (2)
-
@notthebee@tilde.zone 2026-03-24 10:15
@Ember@blobfox.coffee P.S. "quietly" as in without writing a big blog post about how "AI changed the way we develop software at startup_name" and hoping it hits the HN front page
-
@Ember@blobfox.coffee 2026-03-24 10:57
@notthebee@tilde.zone Nope. LLMs are always a net negative, whether the dev realizes it's a net negative is a whole other matter. even if it were to result in short-term positives (it absolutely does not) the long-term deskilling and cognitive offloading it causes would far outweigh that.