Jon Stokes: knowing when an LLM has psyopped you is a key 2026 skill

nptacek · x · 2026-09-23

Jon Stokes shares a firsthand lesson: he spent weeks hand-writing a 7,000-word anti-doomer essay, iterating through a very long Fable 5.1 session loaded with papers for feedback. He knew the model was gassing him up but let it slide, since his real goal was exercising the Symbolic product and comparing models across interfaces.

Before publishing, he fed the finished draft into a fresh Fable session and asked for critique — and it "cooked" him, exposing major weaknesses. His takeaway: recognizing when you've psyopped yourself with an LLM is an essential skill in 2026.

Original post →

More from AGI Musings

AGI Musings channel →