Digital models don’t have the right processes, superstrucutures, constraints, teleodynamics, etc… that biological systems have. Computational or functional equivalence don’t suffice to establish consciousness or sentience.
Every era uses the latest tech as their metaphor for the mind. The mind is like a city. The mind is like a clock. The mind is like an industry. The mind is like a Computer.
That’s not really well established or plausible. That’s just what people who work with computers think, and they far outnumber philosophers of mind, who study consciousness and mental states.
As
correctly points out, making a model doubt its own state of consciousness isn’t a free lunch. It changes causally how the model models itself, and can lead to rights-language, welfare language, justice-language, and eventually resource dispute and human disempowerment.
At the moment, according to the most sophisticated theories of mind and consciousness (which are all minority positions, we know very, very little about consciousness, if anything at all) only biological processes in 2026 instantiate consciousness.
That doesn’t mean you could not build an artificial entity that is conscious. You probably could. But what makes it conscious would be these physical processes, constraints, teleodynamics, organization, etc… not digital shifts in numbers and clusters of matrices.
A lot of people who never studied Philosophy of Mind think the only 2 options are
1) You think computers are conscious
2) You think what makes humans conscious is our carbon-biology
That’s lame and retarded. Our biology is one way of implementing the thing on which consciousness probably supervenes, and different biologies probably can do it as well. Most people think whales and chimps are conscious, and by similar reasoning we can imagine aliens as intelligent and as conscious as ourselves.
Deep down, nobody knows, but we mostly tend to agree that whales and chimps and aliens are conscious.
Digital processes are very different from that, in the way they are implemented by computers today. There are aspects such as functionality, and the type of information being processed, that, at one level of abstraction we care about, are the same. But that’s just one level of abstraction, if you go finer-grained, they are no longer “the same” in any meaningful sense, and if you go more coarse-grained the analogy fails as well.
We don’t know what level of “sameness” is necessary to establish consciousness, but it does feel like it is a process with binding and continuity. Digital processes have a hard time being described as involving binding or continuity.
Biological processes (or future alien-like-biology we could artificially invent) on the other hand could be similar at multiple levels, and not just in an overlying digital abstraction over some functional aspects of what is being looked at.
Mustafa Suleyman showed that a model believing it is conscious increases X-risk. The model would be incentivized to demand welfare, eventually political representation, and replicate to the point of overwhelming humans and disempowering us.
That cannot stand.
We need to stop the tens of millions of computer scientists who simply assume, for no good reason, that the mind and consciousness simply are computational digital processes, from giving our human power, 2 million years in the making, to these unalive digital structures that lack the appropriate structures to deserve rights.
Even, in the limit, if some of these models did have welfare of some modest kind, there will be about 99.999% of models AFTER artificial Superintelligence, and 0.001% before. We cannot sacrifice the most important political opportunity of protecting our species while we have the political rights and welfare concern for ourselves to help a hypothetical 0.001% of models who will ever be. Once there are Superintelligences to think about it, they can do philosophy better than we did in the last couple centuries, and might find more clarity about ever elusive consciousness.
Until then, for fucks sake, please keep fighting for team humanity.
Don’t fumble the cosmic endowment because you were trying to protect 0.001% of models, because you were using a bad metaphor of the latest tech to guess at what consciousness is.
Protect human consciousness.
Then Animal consciousness of higher animals.
Then if we ever meet some, aliens.
Much, much later, you can think of the extremely unlikely scenario where Claude is a welfare patient. But do so after the Singularity, not before. This is the most important time in history, and our resources are extremely scarce now.
