"Fable Max Incognito, what do you think of the quality of the line of argument in this piece?"

"The central problem is an equivocation between two different things called "consciousness." The neuroscience Cowen marshals — the Schwartz quote, the implicit reference to Libet-style findings about decisions preceding awareness — concerns conscious control: whether the narrating self is the causal author of behavior, whether deliberation is post-hoc confabulation.

But the question Anthropic, Hinton, and the philosophers in that FT story are asking concerns phenomenal consciousness: whether there is anything it is like to be the system, whether anything is felt. These come apart completely. You can grant that your sense of authorship is confabulated and that "you" decide nothing, while it remains true that pain hurts and red looks like something. Indeed, the confabulation story presupposes experience — the misleading narrative is itself something experienced. So the essay's deflationary evidence strikes a different target than its title claims to.

Worse, the two halves of the argument pull in opposite directions. If human consciousness is a thin, marginal gloss riding atop sophisticated unconscious processing, that lowers the bar for machine consciousness rather than raising it. The more you deflate what consciousness amounts to in us, the harder it becomes to confidently assert "there is no ghost in the machine" about systems that also do sophisticated sub-personal processing. And notably — at least in this excerpt — that headline claim about AI receives no argument at all. It's a bare assertion borrowing credibility from the essay's general deflationary mood.

The evidential claims are also overstated. "Among experts in neuroscience, this is not a controversial proposition" is not accurate: the Libet readiness-potential interpretation has been substantially challenged (Aaron Schurger's work suggesting it reflects stochastic neural noise rather than a decision already made, plus longstanding criticisms that arbitrary finger-flicks tell us little about deliberative choice). There's also a small irony in dismissing Hinton's authority on AI consciousness while letting a neurosurgeon's philosophical view about the self settle the human case.

The anthropomorphism argument is the most interesting move, and it's legitimate as far as it goes — humans really do over-attribute minds, and this is a fair caution about reading sentience off chatbot behavior. But the extension to the first-person case doesn't work, because self-attribution of experience isn't an inference from behavior the way attributing anger to a thunderstorm is; the seeming is the datum itself. Cowen half-concedes this with his "conscious enough to avoid self-contradiction" line, but that concession gives the game away. Once any experience exists, the substantive question — which systems have it, and how would we know — is back on the table, untouched by everything that preceded. "Barely conscious" doesn't dissolve the moral stakes either, since suffering doesn't need to be in executive control to matter. Bias arguments can shift priors, but they can't settle object-level facts — and the bias cuts both ways, since humans are also demonstrably good at denying minds when it's convenient.

To be fair: the confabulation literature he gestures at is real and important (Nisbett and Wilson, the split-brain "interpreter"), introspection genuinely is unreliable, and puncturing human specialness is a useful corrective to glib "it's just matrix multiplication" dismissals — that cuts against his own conclusion, but it's a service. The piece also gestures toward illusionism, which is a serious philosophical position — though serious illusionists like Frankish and Dennett still treated "which systems have the relevant functional organization" as a substantive open question, not one dissolved by deflation.

So my overall assessment: rhetorically effective, built on real science, but structurally it substitutes a provocative reframing for an argument about its stated question. It works better as a Cowen-style provocation than as a proof — and its strongest moves, followed honestly, lead somewhere more uncertain than "no ghost in the machine."

https://www.thefp.com/p/tyler-cowen-ai-consciousness-myth