3 ms·
On the independence between consciousness and computational intelligence
- belter 4y ago"...Consciousness and intelligence are properties commonly understood as dependent by folk psychology and society in general. The term artificial intelligence and the kind of problems that it managed to solve in the recent years has been shown as an argument to establish that machines experience some sort of consciousness... ...Following Russell’s analogy, if a machine is able to do what a conscious human being does, the likelihood that the machine is conscious increases. However, the social implications of this analogy are catastrophic. Concretely, if rights are given to entities that can solve the kind of problems that a neurotypical person can, does the machine have potentially more rights that a person that has a disability? For example, the autistic syndrome disorder spectrum can make a person unable to solve the kind of problems that a machine solves. We believe that the obvious answer is no, as problem solving does not imply consciousness... ...Consequently, we will argue in this paper how phenomenal consciousness and, at least, computational intelligence are independent and why machines do not possess phenomenal consciousness, although they can potentially develop a higher computational intelligence that human beings..." [PDF]: https://arxiv.org/pdf/2208.02187.pdf https://arxiv.org/pdf/2208.02187.pdf
- IceMetalPunk 4y agoI only skimmed a bit of this, but I already notice one glaring problem with it. The author defines consciousness in several subjective qualia terms, then claims machines can't possess them because they are subjective qualia. But as the P-Zombie problem famously shows, we can't even show that people have qualia. If we can't provide evidence that humans have qualia, how can we possibly claim that machines do not -- or even cannot -- possess those same qualia? Isn't that a bit like saying "I have no way to know what color this apple is, but I claim it can't be red because I believe that only other apples that I also can't measure are red"? It also says that consciousness is hard to define because it's hard to measure; I think this is backwards. It's hard to measure because it's hard to define. Once we have a clear definition of what should be considered consciousness, then we will either be able to measure it, or be able to show it can't be measured, because we will know what it is. The most robust definition of consciousness I've ever heard is "the ability to synthesize new information from data" -- basically, integration of data is the key. It's not necessarily about intelligence; for instance, being able to take the image produced by the light on your retinas and pick out objects from there is an integration of data, and thus implies some level of consciousness. But more importantly, this definition allows for something we see in nature: it means that consciousness is not Boolean, but a continuous spectrum. A rock has zero consciousness, a human has quite a lot, but what about a sea sponge? Or a mosquito? Or a crab? Or a dog? Surely they all have more consciousness than a rock, but it's safe to say at least up to the crab level of these examples, they have less consciousness than a human. (Dogs are debatable.) So if that turns out to be as robust a definition of consciousness as it seems, then there is no reason to think machines can't be conscious; in fact, many modern AIs would be considered conscious already, as they integrate data to synthesize new information. Especially cutting-edge multimodal AIs that integrate different types of data. Do I think we have conscious AI already? I do. Do I think we have human-level conscious AI yet? No, not at all. Do I think we'll get there one day, maybe sooner than we expect? Absolutely.