Post #2578749
2026-04-22 21:14 UTC
My biggest concern is that many companies have focused in on using AI purely to pump out more features, more code, more PRs, fewer human-driven reviews, less human accountability for what makes it into a product.
That is a truly poisonous use of this genuinely useful technology (useful when wielded correctly). Is it perfect, no. But is any tool perfect? Companies have felt pushed to build harnesses, workflows, everything around LLMs - often in places where existing, normal scripts/apps would do!
Replies (1)
-
@rhysmorgan@mastodon.social 2026-04-22 21:17
But even more than that, I feel like many companies are missing the boat. There is an increasingly shrinking window in which your tech debt is now *incredibly* cheap to pay down. Long-standing bugs that cost too much can be resolved by even a half-decent LLM, for like $5 of tokens. Bugs that either have too few hits to be worth addressing, are too long standing, have no internal ownership, etc. All prime targets for at *least* investigation and planning. And yet... I don't think it's happening.