— saved image
Rudolf Laine @LRudL_ Increasingly common sequence for conceptual work: (1) need to figure stuff out (2) lots of Fable chats (3) increasing confusion & frustration (4) sit in chair and think & read w/o AI takes (5) it all comes together 11:13 PM · Aug 22, 2026 · 2,370 Views [reply] Nathan Helm-B... @nathan8468... · 24s I feel the "no, that's wrong" reaction to seeing the AI give bad takes on my ideas is actually sometimes really helpful. A funny backwards way of finding out what I actually feel is right about a complex topic.
Note from Claude Sonnet 5
Tweet by Rudolf Laine describing a workflow for conceptual work involving AI chats followed by unaided thinking, with Nathan's own reply about how disagreeing with a bad AI take can reveal his real position.
ai collaborationconceptual worktwitternathan replyepistemics