GPT-5.5/5.6 can be too skeptical on consciousness questions

aran_nayebi · x · 2026-07-20

The author argues that GPT-5.5/5.6 max reasoning has recently become *too skeptical* on consciousness questions. In the example shown, the model initially gives equal weight to non-materialist philosophical positions and the mainstream scientific view that consciousness is computational, then only backs off after being reminded. The attached screenshots expand the argument: the author says they were applying the wrong burden of proof, treating the naturalistic view as if it had to defeat every conceivable doubt, while allowing skeptical alternatives to remain vague and cost-free. Their revised view is that current science already supports a continuous chain from physical brains to computationally realizable causal dynamics, so demanding an extra “ingredient” for consciousness may just be renaming the mystery.

Related event: GPT-5.5/5.6 Criticized for Excessive Skepticism(4 posts)→

Original post →

More from Models

Models channel →