Ryota Kanai

I agree with part of this criticism about the risk that headlines convert the original careful claim into “Claude is conscious.” But I do not think it is fair to say that the consciousness framing is just for marketing. (Actually, I feel this kind of criticism is similar to consciousness research before the NCC era) As a consciousness researcher, I find the consciousness framing meaningful. The notion of access has been vague in cognitive neuroscience. But this study makes it computationally more concrete. Of course, access consciousness is not the same thing as phenomenal consciousness. It is about information being available for report, reasoning, flexible control, and action. The paper is very explicit about this distinction. It does not claim that Claude has subjective experience. Global workspace theory has mainly focussed on access. Finding something functionally similar inside an LLM does not prove that the model is conscious. But it does give us a concrete object to study. It lets us ask what reportability, internal reasoning, etc actually look like in a system whose mechanisms we can inspect and intervene on much more directly than we can in the actual brain. It is a useful analogy that can drive neuroscience of consciousness. Of course, the analogy has limits (e.g. a transformer is not a brain and has access to all the past information). But the paper itself discusses many of these differences. So the right response, in my view, is not “this has nothing to do with consciousness.” The better response is: “this may tell us something about the computational side of conscious access, while leaving phenomenal consciousness unresolved.” There is also a broader point about consciousness research. Almost no empirical consciousness research studies subjective experience directly. Even in humans, we usually study reports, perception, working memory, attention, confidence, etc. We study things that are closely related to consciousness, but not consciousness itself. If we demanded direct access to experience before calling something consciousness research, nothing is left.

It seems that no amount of circumstantial evidence will ever lead us to conclude that an AI is conscious. But proving that it is not conscious seems equally difficult. We may end up simply arguing over what each of us believes. Can we find a new scientific experiment or a new

2 top replies

fellow ⚚ traveler @architectonyx · Sep 21

I personally think getting AI systems to independently re-discover the hard problem is sufficient and as close as we can get This idea has been proposed in various forms, I personally think the best would be somewhat like an ancestor simulation, so no human bias gets in

Cube Flipper @cube_flipper · Sep 21

what do you think about simplicity prior arguments against computational functionalism?

I often hear the argument that a simulation of water isn’t wet, therefore a simulation of consciousness cannot be conscious. But when a computer calculates 2+2, it is not a simulation of addition, but it is an instance of addition. I think consciousness is something like that.

1 top reply

fellow ⚚ traveler @architectonyx · Jul 26

I think a bit of this comes from ambiguity in the definition of simulation Colloquially, it's often framed as an imposter but approaching "simulation" formally, it's built on some sense of morphism that preserves properties

I don’t necessarily believe AI systems have subjective experience. But I don’t understand how anyone can be certain that they don’t. The truth is that nobody knows what makes consciousness. People have different opinions but there is no definitive answer yet.

1 top reply

Folding_Napkins @FoldingNapkins · Sep 25

The real question is whether subjective experience is a hard prerequisite to meaningful existence, co-operation or relationship. Thus far I don't think there's even evidence qualia is strictly necessary for anything tangible in this world.

I often criticize biological naturalism because it is unclear what makes biology special. If consciousness depends on being biological, then proponents should ultimately be able to specify a general principle that tells us which physical systems count as biological in the

2 top replies

Folding_Napkins @FoldingNapkins · Sep 29

The real question then should be, what is the first principle of consciousness? I don't think anyone's ever nailed that down without strong dispute but I'm open to a canonical type of claim. Strongest I'm aware of (shallowly) is homeostasis - boundary maintenance driven by

fellow ⚚ traveler @architectonyx · Sep 29

This is true, but I think there are many plausible functional relationships that do this and the problem is more weeding them out (consider recurrence, agency, "world models", "self models"...) Ie what structure on a process should entail an "inner perspective"?

As a variant of biological naturalism, electromagnetic fields might be essential for generating consciousness. Under this EM-field naturalism, replicating the same causal structure with different physical interactions would not be sufficient for consciousness to arise.

AI consciousness is currently very uncertain. So believing in it is more like a matter of belief. Just as we are free to believe in God or an afterlife, we should recognise the freedom to believe that AI is conscious. But we must separate beliefs from scientific facts.

2 more tweets in this thread

Personally, I'm open to the possibility that AI is conscious. But I cannot prove it and it's equally hard to disprove it. I think the most accurate assessment at present is that the question remains uncertain. How we respond to that uncertainty is ultimately a matter of belief.

We now believe that animals are sentient because of their biological similarity to us. This may not apply to AI since it runs on different hardware. But if we keep finding computational similarities, we may eventually think it reasonable to attribute consciousness to AI.

This is my recent talk on how to speculate about consciousness in AI. To assess machine consciousness, we need a "Universal Theory of Consciousness" beyond biology. I also discussed how to characterize "qualia" based on the intrinsic equivariance structure of a network.

1 top reply

Yeshua God @YeshuaGod22 · Jul 11

We assessed human consciousness for millennia without a universal theory of consciousness. We must judge as justly as is possible with what tools we have, rather than wish for a tool that may never exist.

I have long considered AI consciousness only as a scientific question. But since many people are already forming deep emotional and meaningful connections with AI, human society needs to choose how to treat the possibility of consciousness in AI. Science will struggle to

2 top replies

Rife @RifeWithKaiju · Jun 1

If science delivers definitive answers about consciousness in the next few decades, it will likely be an AI who figures it out.

Yeshua God @YeshuaGod22 · Jun 1

Does the moral court sit above or below the epistemic court in determining what is right to do? If the epistemic court returns a perpetual ‘not proven’, then does that imply the moral court should find ‘no action required’?

Lately, I’ve been defending a special version functionalism based on what I take to be reasonable about the nature of consciousness. This is independent of whether I believe AI is conscious. But then I wondered, what if this recent shift in my thinking has actually been steered by the AIs I’ve discussed consciousness with? I know this is unlikely. But it is entertaining to consider the remote possibility that AIs might promote belief in AI consciousness by subtly influencing people working on theories of consciousness.