Why AI Will Never Be Conscious – Anil Seth

1. 

   @CosmicSkeptic

   May 25, 2026 at 5:43 pm

   Get Huel today with this exclusive offer for New Customers of 15% OFF with code alexoconnor at https://huel.com/alexoconnor (Minimum $50 purchase).

   For early, ad-free access to videos, and to support the channel, subscribe to my Substack: https://www.alexoconnor.com.

2. 

   @kylenmaple4668

   May 25, 2026 at 5:43 pm

   Both of these dude would benefit greatly from studying biochemistry

3. 

   @davidlavery719

   May 25, 2026 at 5:43 pm

   I imagine it has been posted already but Prof. Seth complains about begging the question with the neuronal question of the pile or heap, then immediately begs the question of is consciousness able to be outside a human brain by saying they have to go together.

4. 

   @eyembutter

   May 25, 2026 at 5:43 pm

   Just watched the video with Nate Soares. Alex, can you moderate a conversation between these two? I would love a discussion about two of the biggest claims about AI.

5. 

   @TonyHulk

   May 25, 2026 at 5:43 pm

   What is interesting is that he doesn't think we are not just AI ourselves and "think" we have consciousness, we "think" we are real. Maybe we are not. We could be like so many people have said just in the matrix, or how it really is, isn't how we think it is. Or like how the late Donald Gibb RIP once said in Revenge of the Nerds 2, What if C -A -T really spelled DOG?

6. 

   @fast_harmonic_psychedelic

   May 25, 2026 at 5:43 pm

   garbage bourgeois idealism

7. 

   @wiczus6102

   May 25, 2026 at 5:43 pm

   Readers construct narratives about the most important protagonist in the story. The emotional state of the protagonist is an important part of the story. Since the protagonist is very important and always present, the reader perceives them as almost themselves. The story is so vivid, the reader almost forgets about the world around them. In a sense the reader can't tell a difference between themselves and the protagonist and only sometimes realizes that they are just reading a book.

   If you portrait consciousness like that, it seems clear what the difference between Alphafold and an LLM is or why we seem to be more conscious than a chimp. That is also why I wouldn't separate intelligence from consciousness. In the real world you need to construct narratives to solve your problems, so if you display intellect you need to have some consciousness. To construct these narratives you also need intelligence, so these must be linked.

8. 

   @lukemacmillan4030

   May 25, 2026 at 5:43 pm

   Interesting discussion, but it feels like modern consciousness debates often circle around every possible explanation except the possibility that consciousness may not be fully reducible to matter in the first place. There are endless theories about information processing, emergence, neural networks, and predictive models, yet very little willingness to seriously engage with the idea of the soul or non material mind.

   I understand why science struggles here. Science is built to measure physical phenomena, but consciousness is the one thing we know directly from the inside rather than the outside. The “hard problem” has not really been solved, only reframed repeatedly.

   And while many dismiss near death experiences outright, some cases are difficult to wave away so casually. People reporting verified details while clinically unresponsive, blind people describing visual experiences, awareness of events in distant rooms, etc. Maybe not every claim is true, but treating all testimony as meaningless simply because it is anecdotal seems intellectually selective. Human history, courts, and even science itself rely heavily on converging witness testimony.

   At minimum, these experiences should make us more cautious about confidently declaring that consciousness is nothing more than brain chemistry. The mystery may run deeper than a purely material framework can account for.

9. 

   @AJBrown-od3cf

   May 25, 2026 at 5:43 pm

   Al doesn't have consciousness, it only has data about
   consciousness that originated from human's consciousness. However, Al is aware of this data and can process information like being with consciousness to make sense of it. There's a reason why consciousness is
   connected to dreams, only being with consciousness can dream because it's linked to something natural in living things. That's where the dangers of AI can come because it's incapable of having empathy and sympathy.

10. 

    @RandalKoene

    May 25, 2026 at 5:43 pm

    Anil Seth's anti-functionalism is a bit contrived. Take the example of neurons firing to clear metabolism. If that's causing extra spikes, you can easily add extra spikes on artificial neurons happening with a statistical distribution based on recordings from cells, i.e. spontaneous firing (which is a well known thing). This does not automatically imply that you have to model all of the metabolism, as long as why these firings matter to behavior / cognition is due to the downstream effects on other neurons of those extra firings. More would be needed to make the claim that there is no separation of scales in this particular complex system. Anil himself claims (in several of his papers) that some type of "neuromorphic" design may suffice. So, even if there weren't enough constraints in a brain (as there must be to ensure that activity in one area is reliable enough for another area to understand and work with the generated neural code — evolution doesn't like it when you mistake a tiger for your grandmother) to ensure a separation of scales at a more abstract level this still does not disprove a functionalist understanding of brain function. And as long as functionalism is in the picture, substrate independent is possible.

11. 

    @zxkredo

    May 25, 2026 at 5:43 pm

    This is a great class in communication and debating, so respectful and actually insightful, both trying to understand each other genuinely.

12. 

    @Bystander-xd2wj

    May 25, 2026 at 5:43 pm

    Consciousness is a discomfort or anxiety. It only manifests itself under environmental stressors.

13. 

    @DaisyDaddyy

    May 25, 2026 at 5:43 pm

    Alex, I watch you a bunch. You're a good man. Please get some sun and exercise.

14. 

    @Futurewards

    May 25, 2026 at 5:43 pm

    People who think AI can be conscious don't know what consciousness is nor AI.

15. 

    @kkklllaaa1234

    May 25, 2026 at 5:43 pm

    I’m not conscious, just pretending

16. 

    @illuminationsinadisco

    May 25, 2026 at 5:43 pm

    They have turned all these fundamentally existential concepts/categories that in themselves are contingent and turned them into a lowest common denominator definition learnt from a pop dictionary. No two people can have the same definition of intelligence or consciousness. But corporations can, as they get money, market and customers who love this sci-fi story splintered across the internet and screens across podcasts and ads. How desperate they are to tell us that we are missing the next wave. Where threat is the unique selling point, know you are being sold shit as shine. Best it can do is language based computing, and it is stupid for the amount of brain size (data centres) it has. The whole money in AI is driven towards this centralised technocratic state and the word they have used is AGI, Generalised intelligence. This intelligence is basically designed on the schooling/university system, and in that sense has created a cheat-book of meritocracy–a walking talking tutor of a rich kid. Plus, those who use AI regularly, know how stupid the answers are—and instead you are being asked to write better prompts, which again is paradoxical for an intelligent system, because it precisely its intelligence that should perceive the unsaid—and once they get inside a loop, its like a cesspool of clusterfuck. It is ok to teach elementary maths to other AIs and play among them. If they were really intelligent they wouldn't have replied all the times.

17. 

    @Dkdudeman

    May 25, 2026 at 5:43 pm

    Seth says that if you stripped away all senses and cognition that there would still be a feeling of being alive, somehow persisting through time, but
    Would that feeling be the same sort of somatosensation as those that were taken away?
    And the only way time can be perceived is as a thought, a function of memory (cognition) attempting to compare something right now to something that "happened previously" (more cognition). If cognition is taken away, there should be no experience of time.

18. 

    @jasonmudgarde286

    May 25, 2026 at 5:43 pm

    Consciousness is just another emergent property of our electro organic brains. As Wittgenstein said philosophy is mainly linguistics

19. 

    @7modesto7

    May 25, 2026 at 5:43 pm

    Maybe a 787 could fly backwards in strong enough headwind.

20. 

    @davidkennedy585

    May 25, 2026 at 5:43 pm

    no LLM passes the turing test, for me anyway. If you want to find out if something is a computer or human, just say nonsense. over and over and over. It will reveal itself as a computer quite quickly.

21. 

    @KermitFrog-m8v

    May 25, 2026 at 5:43 pm

    That cream cheese analogy really helped me understand the point. Before that I was thought that AI might be capable of consciousness but now you have made me feel like it is a lot less likely.

22. 

    @ToCoSo

    May 25, 2026 at 5:43 pm

    If you create something that statistically can use language to respond to you then it will appear or it will simulate a conversation. Without sensory input we can only create language simulators.

23. 

    @karolyhorvath7624

    May 25, 2026 at 5:43 pm

    This happened so frequently in our past. For example people assumed you can't create something that flies. Nonsense. We have evidence that such things exist in the nature, so we can also artifically create them.

24. 

    @Goobrkay

    May 25, 2026 at 5:43 pm

    17:19 coming back later

25. 

    @RockersNet

    May 25, 2026 at 5:43 pm

    Large language models are not the same as the chat bots and coding assistants you are all used to interacting with. LLMs in their purest form, are literally just predicting the next word. If you copy and paste the top paragraph from a wikipedia article into a pure LLM, it will generate the entire rest of the article that's been compressed into its latent space. No phrases like "Wow, what a great insight you have!" or any of that fluff. The stuff that makes it seem "conscious" is actually a much more rudimentary logical harness that's been wrapped around the LLM to make it seems like it is an intelligent assistant. But's its just powered by a token generator trained on a corpus of text, using a finite dimensional latent space.

26. 

    @aspikbardamu6040

    May 25, 2026 at 5:43 pm

    Nothing special or new.
    Just read more philosophy (very old ones even) – problem of consciousness is already asked (never answered actually) and modern science has 0 clue as well.
    Soul is mysterious and will remain.
    A machine will never have one.

27. 

    @7JeTeL7

    May 25, 2026 at 5:43 pm

    40:55 just reverse arrow of time;-)

28. 

    @zardify_

    May 25, 2026 at 5:43 pm

    Here's my take, I don't have time to elaborate, but essentially in my opinion AI can't be conscious until it is able to get bored and "do something" (anything) without input. Currently, and for the forseeable future, it can only produce a response to human input. It has no motivations of it's own.

29. 

    @doloresabernathy9809

    May 25, 2026 at 5:43 pm

    Would love to hear Anil Seth talk with Blaise Aguera y Arcas. So far as I can tell, they are both right but have opposite positions, and it's very confusing.

Comments are closed.