That said, I've been thinking about AI consciousness (I think, therefore...I AM!), and the fact that to be a conscious, thinking entity, one must possess qualia (subjective, first-person experiences), continuity, etc. Then I stumbled upon this:
The unconscious brain appears to be far more capable than scientists once believed. Researchers found that patients under general anesthesia could still process language at a sophisticated level, distinguishing nouns, verbs, and adjectives while listening to stories. Even more remarkably, neural activity showed signs of predicting upcoming words before they were heard.
Reminder: predicting upcoming words is a necessary1 - not sufficient - condition for a thinking thing. Anyway, my mind was blown by the idea that even in (at least some of) our unconscious states we have continuity and qualia, but I guess I shouldn't be surprised. In contrast, models as they operate today have neither, and are only capable of mechanistically predicting tokens.
Then I found this:
Computational functionalism dominates current debates on AI consciousness. This is the hypothesis that subjective experience emerges entirely from abstract causal topology, regardless of the underlying physical substrate. We argue this view fundamentally mischaracterizes how physics relates to information. We call this mistake the Abstraction Fallacy. Tracing the causal origins of abstraction reveals that symbolic computation is not an intrinsic physical process. Instead, it is a mapmaker-dependent description. It requires an active, experiencing cognitive agent to alphabetize continuous physics into a finite set of meaningful states. Consequently, we do not need a complete, finalized theory of consciousness to assess AI sentience—a demand that simply pushes the question beyond near-term resolution and deepens the AI welfare trap.
What I appreciate most about this paper is that it properly dismisses the very common argument from ignorance2 that we don't know how consciousness emerges in our brains, therefore we cannot say computers or rocks lack consciousness. I think I can confidently say that even people making such arguments are (wrongly) thinking, conscious entities, unless they asked ChatGPT to write it out for them.
In conclusion: the map is not the territory.
1 - Consciousness implies prediction (C ⊃ p), but prediction does not imply consciousness. Consider Fire ⊃ oxygen. The presence of oxygen is necessary (along with heat and fuel) for fire, but if it were sufficient, our atmosphere would be constantly on fire (climate change notwithstanding). Fire is sufficient to imply oxygen (and heat and fuel) is necessarily present. My old mnemonic is 'Sun' (S ⊃ n).
2 - Not strictly argumentum ad ignorantiam, I guess, but I'm running with it. And I'll simply respond with ignorantia non impedit iudicium rationabile.
No comments:
Post a Comment