Distilled DeepSeek 14B still fails the Voight-Kampff test locally
Apoorva__Lal · x · 2026-08-22
User shares a chain-of-thought snippet from a locally running Distilled DeepSeek 14B model. The reasoning process exhibits distinct 'machine logic', humorously compared to failing the Voight-Kampff test from Blade Runner.
More from Fun
- Claude Opus 5.5 makes a surprisingly good unofficial Apple HomeView commercial — john_lam · 2026-10-16
- Zizians discourse: community sites judging members' 'good vs evil' spark stochastic terrorism concerns — jessi_cata · 2026-10-16
- 'If we developed AI in the 1800s, people would demand ASI to cure diabetes' — zetalyrae · 2026-10-16
- Dev's test 'daily brief' agent now spams her with emails she can't figure out how to stop — fhinkel · 2026-10-16
- His son answered riddles like an evaluation-aware model — and nailed the analogy — rao2z · 2026-10-16
- Blogger builds a website just for AI agents parsing his posts — dorsar · 2026-10-16