2026-09-14 22:03 UTC
@zzzeek@fosstodon.org @glyph@mastodon.social If we're being as charitable as possible toward the LLMs here, the consistent finding is usually phrased as "LLM code generation amplifies what is already there", and the effectively-unavoidable corollary is that the percentage of teams/people who already had good-enough practices in place is so low that it is recklessly irresponsible to deploy this technology at scale.
Replies (1)
-
@ubernostrum@infosec.exchange 2026-09-14 22:08
@zzzeek@fosstodon.org @glyph@mastodon.social Also, my direct personal experience of trying to level up the practices and the team culture to try to make them ready for the benefits LLMs are supposed to bring is that it doesn't work because managers are always getting pressure from directors who are getting pressure from execs/board who are getting pressure from investors to show evidence of "we're an AI-first company now" and so the practices and culture get set aside in favor of a token-spend leaderboard.