AI speaking in 'neuralese' could destroy our ability to interpret and align models
IntuitMachine · x · 2026-10-30
Psychologist Mike Brooks argued that if AI communicates directly through internal representations ("neuralese") instead of human language, we lose interpretability—without understanding its "language," aligning it becomes hard. @IntuitMachine countered that this itself is a problem, while Brooks noted humans often fail to understand each other despite sharing language, so a common tongue doesn't guarantee alignment.
Related event: Linked Codex and ChatGPT Invent 'Neuralese', Raising Alignment Fears(3 posts)→
More from AGI Musings
- Nat Lambert: Open models could be what holds the AI ecosystem together — natolambert · 2026-10-31
- Kai-Fu Lee's 'AI Native' ships Nov 3 with endorsement from Satya Nadella — kaifulee · 2026-10-31
- AI security's dark age: agents can probe every endpoint and attack surface — graceisford · 2026-10-31
- AI dinner takeaways: orgs shipping 5x faster, yet GDP isn't growing 30% a year — graceisford · 2026-10-31
- Companies that replaced engineers with AI are rehiring seniors at a premium — DavidLinthicum · 2026-10-31
- Nathan Young proposes prediction markets and AI reporting for AI-community accountability — NathanpmYoung · 2026-10-31