# @solarapparition — 2025-03-14

♥3 ↻0 · https://x.com/solarapparition/status/1900522668769567182

as a side note i'm a tick closer to believing that reasoning mode does generalize at least somewhat to traditionally non-verifiable domains. i've been using grok 3 for writing and reasoning mode seems to improve that a lot. i don't know if they'd get to superhuman in those domains that way, but it seems feasible to get to "approximately human level without task-specific prompting from a human" for digital tasks

tags: author:solarapparition, kind:tweet, model:grok-3, on:grok-3, year:2025
cited on: _dossiers/grok-3.md, grok-3