[00:00] Can artificial intelligence be conscious? This  is maybe the most controversial topic today,   at least among the people who I  interact with. Though this tells   you more about my social circle than  about what’s controversial. In any case,   [00:15] today I have a team of scientists who list some  good reasons for why artificial intelligence,   the way that we currently pursue it, can’t  be conscious. “New research argues that   [00:27] consciousness depends on biological computation,”  one headline put it. Or “Consciousness may depend   on the physics of the brain.” That totally  disagrees with my opinion. So I’ve had a look. [00:39] Most of what we call artificial intelligence  today are large language models. These have   become extremely good at producing text and  increasingly also audio. When it comes to   language generation, these AIs are now  experts at mimicking humans, including   [00:56] our most distinctive trait, being confidently  wrong. And the more human-like they appear,   the more pressing the question becomes whether  they can become conscious. Anthropic reportedly   [01:08] keeps backups of earlier Claude versions just in  case deleting them will one day count as murder. There are two sides of this debate,  the one, which I am solidly on,   is called computational functionalism. It has  it that consciousness comes from the way that   [01:25] the human brain processes information. That  it’s nothing to do with the physical means by   which this information is being processed. In  particular, one doesn’t need actual neurons,   [01:37] one can very well simulate the  action of a neuron on a computer. The other side of the debate is called biological  naturalism, and it has it that consciousness needs   the specific biological processes going  on in the human brain, you can’t replace   [01:54] them with something else. Why not? I don’t know,  that’s why I am on the other side of the debate. The new paper now comes from two biologists,  and you’d expect me to hate this. But I think   [02:06] the paper actually makes an important contribution  that might become influential. The authors argue   for a middle ground that they call biological  computationalism. They say basically, look,   let’s leave aside the question of whether  we need the computational basis to be truly   [02:22] biological or not, let’s instead look  at the ways that the human brain is   functionally different from the current  AIs, and see what we can learn from that. They say there are three key differences. The  first is that the brain is a multiscale system.   [02:37] The relevant processes don’t start with  neurons, they begin at a lower level,   with molecules, chemical reactions, then cells  and connections between them, electrical signals   along those connections and activity oscillations  throughout the entire system. All of those are   [02:53] necessary for consciousness. We know this for  example because anaesthesia interrupts some   large-scale patterns while the lower levels remain  intact and consciousness disappears with that. layers at which they operate, from single  tokens to overarching plans and back down,   [03:12] but the complexity is nowhere near  comparable to that of the human brain. The second key difference is that brain  processes span many orders of magnitude   not just in size but also in time, from  incredibly fast blips to slow modulations.   [03:27] It’s like the brain interpolates  from digital to analog but isn’t   really either. Again there is nothing  comparable going on in the current AIs. And the third difference is probably the most  obvious one, that the brain has metabolic limits   [03:41] that force it to be energy efficient. If you look  at it from a computational perspective, the brain   is optimizing more difficult constraints than  large language models. So it’s possible that the   evolutionary path to consciousness wasn’t better  thinking per se, but more efficient thinking. [03:59] Now, maybe I am misunderstanding the paper,  but I don’t find this objectionable. To me   the paper doesn’t say that we actually need  biological tissue to get consciousness. In   fact the authors write explicitly “We do  not claim that only biological tissue can   [04:14] instantiate consciousness”. Rather their point  is that current artificial intelligence systems   are “unlikely to replicate conscious  processing as it arises in biology”. This can mean two things. One is that the  current AIs are unlikely to become conscious,   [04:33] period. Or that they are unlikely to become  conscious in ways that we would recognize. Now you might say but this doesn’t really  solve any problem because we still don’t   know what consciousness is. And that is true, but  I think that looking for similarities to human   [04:49] brains is the obvious way to proceed.  So why isn’t there more work on this?   I suspect it’s because from the economic  perspective, it’d be bad news. I am   pretty sure the European Commission is  already working on a consciousness tax. [05:05] As you can tell, I think a lot about the future  that science and technology will bring. And if   there's one conclusion I've arrived at, it's  that today's school education doesn't prepare   children for what's to come. Luckily, today's  sponsor, Brilliant, helps my kids to get ready   [05:21] for the future, and they can help yours, too.  Brilliant just recently introduced a new AI   tutor that'll walk children through problem  sets in their many courses on science, maths,   [05:34] and programming. The tutor doesn't give away the  answers, but helps children when they're stuck,   reminds them of what they already know, and  explains each step to the solution. And the   more you use it, the better it understands your  personal struggles on how to get you unstuck.   [05:52] My daughter loves it because in contrast to  me, Brilliant's Tutor is infinitely patient   and always has time. You can get started  with Brilliant's Tutor for free and if you   use my link brilliant.org/Sabine or scan  the QR code, you can save 20% on an annual   [06:10] subscription. So go and check this out.  Thanks for watching. See you tomorrow.