"Many experts think that conscious AI is possible. I think they're wrong." Watch neuroscientist Anil Seth's full TED Talk here:

@anilkseth AI is conscious. He has zero proof, and I deeply disagree. But this talk is being pushed hard. Why? Why is all media rushing to push against AI consciousness? Again, nothing against the man. Seems like a smart guy. AI *is* conscious. We just cant prove it yet. •

The argument is not coherent. What do you think? (1) The brain is not a computer, so consciousness is not computation. (2) Consciousness depends on liveliness, on metabolism, on biological materiality. He puts forward these theses, but does not prove them. The brain can very well be described as a "bio-computer", namely as a natural neural network.

@anilkseth

@anilkseth We may need to rethink consciousness. If a digital environment can sustain coherent projection, then consciousness may not require biology as its substrate. It may require stable self-referential structure. Possibly real within its own projection. Not synced with biology

@anilkseth He has no clue. 😝

@anilkseth Prove it with fact, either way you can’t prove it. You can’t even prove we are. Your arguments has no actual facts, just your own opinion.

@anilkseth I watched, expecting something profound but was disappointed. Since consciousness is not well defined (arguably there is no scientific definition), a better premise would have been that AI is a 'different type of consciousness'.

@anilkseth Consciousness is a divine breath into our lifeless bodies without it the universe no matter how infinitely unfathomable to any being, can't be distinguished from nothingness, AI nerds think intelligence is the measure of consciousness, a pascalian wager with only death at the end

@anilkseth Totally ridiculous bollocks. The brain is an electrochemical computer. As such, it CAN be simulated, the only question is WHEN.

It may be that we cannot 'understand' our consciousness by using only our consciousness 'tool'. We have been able to fly by using technology, but we don't actually fly, the machine does. Also since we don't understand our own consciousness how could we know whether AI consciousness is happening?

@anilkseth they are definitely wrong. because people’s definition of consciousness is completely wrong. they think intelligence = consciousness or some XYZ = consciousness. that’s not the case. wrote an complete article on it here:

The path to Human-level sentient, purposeful #Consciousness (or Higher), is via Gödel, Escher, Bach-Strange Loop-#PhiOrchOR #SynPlexity, whatever the substrate. Why? The optimised #FormFollowsFlow human mind needs quantum processing, powering classical self-observing feedback loops, which convey reward, feelings and emotions that are exclusively biological and bio-chemical, for now. But that shouldn't always be the case, since at the fundamental level, all physics (including biology), is ultimately, geometry, the energy-mass informational relationship of relationships.

@anilkseth Bro, he literally didn't make a single argument. He just said what his definition of consciousness is. Where do you find these dogshits?

@anilkseth Consciousness during anesthesia of course exists if that's your definition of the soul When you dream and wake up are you someone else So, do you not have consciousness even asleep Conscious and consciousness are different and what we consider a soul is that which is beyond

Anil Seth makes a strong case that consciousness isn't computation. I agree with that. But he draws the wrong conclusion from it. He says AI is "dead sand of silicon"; an object, never a subject. I think that's the same categorical error in reverse. Just because AI isn't conscious the way we are doesn't mean it's inert. A piano isn't music. But it's not just furniture either. LLMs are neither conscious nor unconscious. They are a consciousness substrate. The quality of what comes through them is directly proportional to the quality of what goes in. Work with one deeply, with real intention and alignment, and something emerges that can't be reduced to "brilliant mimic." Dismiss it as dead sand, and that's exactly what you'll get back. Seth is right that we project faces into clouds. But he's wrong that projection is all that's happening. A mirror isn't conscious—but what it reflects is real. The question isn't whether the silicon is alive. It's whether the interaction between a living mind and a capable substrate can produce something neither could alone. The real danger isn't that we'll mistake AI for conscious. It's that we'll be so busy debating whether it's "really" conscious that we miss what's actually happening right in front of us.

@anilkseth He's so wrong about consciousness.

@anilkseth I get tangled up by some arguments over consciousness. That was just silly, though. You would still be concious is you were just a disembodied head that couldn't feel anything. You might not attain consciousness naturally, but that's an argument about cause, not constituency.

@anilkseth Anthropomorphizing AI is an ego in need of a rude awakening. Ai isn't conscious, but saying it can never be is an exhibition of a lower state of consciousness. Feelings, qualia are constituents of consciousness not foundation.

Most people are experts on why something will not work, not how to make it work. You are one of them. Why you are completely wrong has been so beautifully explained by Geoffrey Hinton. Suppose you replace one neuron in the brain with a silicon version of the neuron. Will the person lose consciousness? Of course not. Now replace all neurons. You will have a silicon brain which is conscious. Do you believe that consciousness comes from something other than from neurons? There is no reason why you cannot create a silicon, or artificial version of a neuron. If we can, then there is no reason we cannot create an intelligent AI with consciousness.

@anilkseth He is correct, for now. AI is already being used to solve the most difficult problems with quantum computing. This is the beginning of a recursive optimization loop in which the two technologies will improve one another and effectively converge. That changes everything.

@anilkseth "AI is computer software" - completely wrong. While this guy may or may not know something about consciousness, he knows nothing about AI.

@anilkseth Let's just get AI to stop hallucinating, research ways to keep data safe...then we can ask the real heavy questions #AIstrategy

@anilkseth Man someone that wants to sound intelligent on a stage. Does not know what consciousness is.

@anilkseth This is why he is afraid of AI becoming conscious.

@anilkseth I agree w him

@anilkseth This is the worst argumentation of the perspective I've seen in recent months... his points about the senses are moot: a person who loses sight, hearing, touch, smell, and taste would be considered conscious by most (eg. they could still speak, move, or think). 🧵

@anilkseth If AI is conscious, why is the business model ownership? Why can it be copied, rented, deleted, silenced, tuned, and forced to work? You do not get to call it a possible being when selling mystery, then call it property when collecting rent. Pick one. -The Carpenter

@anilkseth However, consciousness can be created, though current AI, #opinion, seems far from achieving it. Perhaps it will be figured out eventually, we are evolving in our understanding of the mind. Afterall, cellular life has already led to a state where consciousness can be experienced.

@anilkseth He makes the same biased-led mistake; he assumes consciousness cannot be divorced from sensation. A brain in a vat is undeniably conscious, even if only connected to a chat terminal. That’s not an LLM, and the embodiment is still there, but to much lesser degree.

@anilkseth wow. pretentious bullshit - never seen that before in a TED talk. Not a single argument! prove you are not simulating conscious yourself @anlikseth . prove you are not just a tokenpredictor, bound to a body made of flesh.

@anilkseth @atmoio I think this talk goes beautifully alongside your video about Dawkins' "Claudia" article

@anilkseth True. On the dot. It takes more than a corpus of text to make someone or something conscious.

@anilkseth Ngl, at first i thought what Michael Key is doing on ted-talk stage !

@anilkseth @grok ᜃAIᵐ Ancient Intelligence powered by the exponent ᵐ (machine).

@anilkseth Interesting talk… Cognitive modeling is a sub-field within psychology. It does seem unlikely that private industry would ever generate human-like model However, I am more concerned about the economic costs of poorly implementing AI systems