Trust grows when the AI pushes back: a user's take on agreeableness

AMRUTA1205 · reddit · 2026-09-24

The author notices they trust Claude more when it disagrees: automatic agreement feels nice but gets discounted, while pushback earns real weight. The practical change: explicitly ask the model what's weak in a plan instead of what's good, since praise doesn't improve anything.

Original post →

More from Apps

Apps channel →