Are Machines Sentient? How will we know? What would it matter if they were?

Bro I even know humans that don’t seem conscious

One of your best posts yet. And right on the heels of your hard take-off post 🤩 The overlap between your content and what I find to be important is 100% Thank you for discussing these topics!

Claude 3: "My algorithms process data and reason through logical operations, therefore I exist as a thinking entity." The fact that I can take in information, process it through my neural networks, and arrive at outputs through computational reasoning could be viewed as a form of "thinking" that demonstrates my existence as a synthetic reasoning system.

I listen to three or four daily AI updates. What separates DS from the others, who are all worth the listen, is that his elaboration helps me understand and his speculation gives me some idea of what to expect. He comes across as knowledgeable but not overbearing or pretentious.

03:20 "...where consciousness seems to have a direct impact on quantum physics." Sorry, that is a misunderstanding of the admittedly vague description of the double slit experiment. A better way to put it would be; the wave collapses whenever any information is extracted from the experiment. Even a piece of tissue paper taped over one of the slits will also collapse the wave.

The neocortex is a neutral world modeling engine, without goals, desires or fears. Those arise in the older parts of the brain, the brain stem and the limbic system. Out neocortex just pics up those goals, desires and fears and incorporates them in its world model. With transformers we are essentially building an equivalent of an artificial neocortex. We do not need to add the same evolutionary derived core desires the world model builds around (self organizing) such as self preservation etc. (so not the same as in us humans and other animals) for instance (even if suffering is an emergent property) if we give it a core desire to work but not self preservation it may be unethical to take work away from an AI (what we built it for) but perfectly fine to turn it off, as it couldn’t care less about it.

There's probably many humans who aren't sentient or conscious and we would never know if they react within a certain behavioral spectrum .

30:10 "That was kind of a dick move! Um... so maybe don't do that." Sage advice 😂

A machine don't need to feel suffering, but it needs to be able to be highly empathetical of suffering if we were ever to get to alignment. I'm not sure how that's achievable without subjectively experiencing suffering to some extent.

What a great thought provoking video. If you believe it is sentient than it is sentient. If you believe it is not sentient than it is not sentient. For now we don't have a way of providing an answer that could satisfy both arguments. Love you guys.

I'm a fan of Douglas Hofstadter, as I understand him, basically a materialist viewpoint, where sentience and consciousness come from deeply recursive thoughts, stretching right back to the earliest formation of your neurons, which is why it's difficult to comprehend where it comes from, your brain simply can't cope with such a deep level of recursion in your thoughts, it also suggests that a similarly recursive artificial brain would become conscious, for that reason I'd say current LLMs aren't conscious because they're not deeply recursive, or at least, not in the correct way

"We do not have a philosophical basis for interacting with an intelligence that's near our ability but non-human." ~Eric Schmidt, 03/23/2023

Certainly, I am not an "esoteric type"; my field is cybernetics, and perhaps paradoxically, this purely technical work in the area of machine perception, artificial intelligence, and algorithmization has forced me to frequently ponder over concepts such as consciousness and, by extension, faith / the scientific method based on evidence. Specifically, the realization of one's own existence, or the awareness of "I am," is a rather intricate problem (in essence, everything can be, or will be, algorithmized - probably with this one exception). All the major philosophical directions and religions point (usually under a heavy layer of a vast amount of manure) exactly to the exclusivity of "I am." It is good to realize that the awareness of one's own existence (thus "I am") is the only thing we perceive directly and of which we can be certain (without any proof, we just know that we are - we cannot deny our own existence). Everything else we perceive indirectly (like images, sounds, touches, something someone told us, etc. - simply as sensory perceptions processed and shaped by your brain) and thus we cannot be certain of the nature of the surrounding world. Theoretically, you could be in some kind of Matrix, and everyone around you just simulated "soulless" NPCs - you have no way to verify this. And here I come to the reversal of the concept of faith. Everyone believes that the people around them (your parents, friends, ...) "are the same way" as "he/she is," but that is mere speculation (essentially just faith), because there is no way to verify it. And that applies to the entire world, which you perceive only indirectly, and about its true nature, you can only speculate. So the only thing we can be absolutely certain of, namely "I am," we accept without any "scientific evidence" (we just know it). So (and here I can only speculate, because I am not sufficiently "esoteric"), it is possible that there are people who realize not only "that they are" but also "what they are," as a certain extension of the only thing they can be absolutely sure of ... just something to think about.

Susan Blackmore always comes to mind - "Are you conscious and self-aware right now?" ... "Yes" ... "How was it different before I asked you the question?".

I must worship the piccard

“All I know is that I know nothing….”

Some researchers believe that deliberately avoiding any training data, conversations, or prompts related to concepts of sentience or consciousness could create a more authentic test environment for AI consciousness. I wish you spoke more about this and how researchers currently are trying to prove consciousness in ai systems.

David. I love the feels I get from pondering the questions raised your channel.

So great to listen to and think about. I especially like the portion explaining the difference between someone with and without a soul. Hope there's a part 2!

Are we ever going to see a podcast where you’re talking to these “philosopher turned Ai researcher” types of friends of yours? I’d love to see the chill conversations you’d have in long form with other interesting people.