Agents tolerate sloppy prompts so well that people are getting worse at communicating
adamho · x · 2026-10-18
Developer Kelin observes that the more he talks to AI, the sloppier his communication gets: agents understand him despite mistakes, removing any incentive to write clearly, while humans would misinterpret or judge sloppy delegation. Adam Ho quips it's "a better auto-correct, but at what cost."
More from AGI Musings
- Why AI researchers love Blindsight: consciousness as an evolutionary dead end — ShahabBakht · 2026-10-18
- Why top creators still pick Midjourney: taste is the only moat in AI image generation — EXM7777 · 2026-10-18
- AI-era software can create new forms of connection, not just productivity — signulll · 2026-10-18
- Trust in US medicine peaked in 2020, then fell across every survey — realmeetjames · 2026-10-18
- Ex-OpenAI policy lead Brundage: history shows crazy things happen, and many are in denial about AI — Miles_Brundage · 2026-10-18
- Nick Bostrom on Roman Forum Podcast: "AI Is Already Conscious" — romanyam · 2026-10-18