[00:00] I'm convinced that artificial intelligence  will become conscious eventually. There have   now been new rumors again that the current AI  might be conscious already until little notice   paper that basically argues that they may be  conscious not while you use them but during   [00:17] the prior phase of training. So let's have  a look. Claude can watch its own thoughts,   but connected to the right tools. It can do  something more useful. It can make your thoughts   turn into reality. Our sponsor Higgsfield just  released Seedance 2.5 and you can connect it to   [00:34] Claude. I tried it. It really only takes about a  minute. You just add it to the Claude connectors.   Here we give it one prompt. Direct a 30-second  film in one take about a physicist who finds a   [00:47] thought in her head that isn't hers. Claude plant  the shots and Seedance 2.5 generated what you're   watching right now. It does this in one take  or 30 seconds without awkward cuts. Same face,   [01:01] same light, same room. And if a detail is wrong,  you can fix that one detail without having to   regenerate the entire thing. This used to be the  part where AI videos fell apart, but Seedance 2.5   [01:15] makes it work. Links in the description below.  So, go and direct something. Thanks for watching.   See you tomorrow. And now back to the science  news. Claude is the chatbot made by Anthropic   and it’s certainly the one that has attracted the  most chatter. Richard Dawkins, author of the “God   [01:30] Delusion” claimed in an essay in May that he  can’t rule out that Claude is conscious. His   argument is basically that he thinks Claude  is so good that if this isn’t consciousness,   then what do we even need consciousness for  which, if nothing else, tells us that Dawkins   [01:45] hasn’t spent much time with Large Language Models.  But Anthropic’s CEO Dario Amodei himself has said   “We don’t know if the models are conscious…  But we’re open to the idea that it could be.” [01:58] It’s not just words. In a new paper that just  appeared, Anthropic researchers say they found   a small, special part of Claude’s internal  activity that works like a silent scratchpad.   [02:11] Claude can put information there and use it  for reasoning, without ever outputting it. This is very interesting, because it means  that Claude has a sort of internal life   and is capable of introspection. It’s  one of the necessary requirements for   [02:26] global workspace theory, that is one of  the leading theories for consciousness. More formally, the Anthropic team called  the internal workspace the J-space,   [02:38] where the J stands for Jacobian matrix,  that’s the mathematical object that the   researchers calculate to identify the space.  The Anthropic researchers now showed that if   they change what is in Claude’s J-space,  Claude’s answer changes. For example if   [02:54] Claude has internally worked out that the animal  which spins webs is a spider, and the researchers   replace “spider” with “ant,” Claude answers  as if the animal had six legs, not eight. [03:07] One Anthropic researcher, Jack Lindsey,  already claimed he had evidence for Claude’s   introspection last year. The new paper now is  a fuller analysis that also identifies exactly   [03:20] where the introspection happens, namely in said  J-spaces. It’s not just Claude. Another group   likewise injected words into Llama’s internal  activity and found that Llama also seemed to   [03:32] take note of it. Probably a similar thing  is going on with all Large Language Models. But not everyone is convinced. Researchers  from New York University have challenged   these claims. They tested this with three  LLMs and found that the models could not   [03:48] reliably distinguish a change that was  made to their internal activity from an   ordinary prompt. So they are saying it’s  just yet another way to give them input. Then we have Erik Hoel, a neuroscientist  who works on consciousness. He has written   [04:05] an interesting paper in which he argues that  today’s large language models are probably not   conscious. We’ve heard this before, of course,  but his argument is new. Hoel compares Large   Language Models to lookup tables. Such “lookup  table” algorithms are the classical examples   [04:21] that demonstrate that observing input and  output alone cannot tell you whether a   system is conscious. It’s like looking up Chinese  translations doesn’t mean you understand Chinese. [04:33] Hoel now argues that by way  of computational structure,   Large Language Models are much closer to  lookup tables than they are to the human   brain. They are stochastic input-output machines.  Much of the autonomy that we assign to conscious   [04:49] beings on the other hand is not directly a  reaction to input. And Hoel says that the most   obvious reason why LLMs are not conscious  is that they can’t learn from new input,   [05:01] though that has not stopped quite a few  people from becoming senior management. This leads to a strange possibility. You see,  large language models work in two stages. The   first is the training in which they get  fed a lot of input. Once this is done,   [05:17] you freeze the fully trained LLM and  use it as an input output machine.   These are the apps you sign up  for. Yes, they now use memory,   but this is externally latched on as context to  the prompt, it does not actually update the model. [05:32] If Hoel is right, then maybe large  language models are not conscious   when we use them. But maybe they  are conscious during training. Personally I don’t think this is a relevant  distinction. Imagine you have a human being   [05:47] unable to form new memories, you wouldn’t conclude  that therefore they are unconscious, would you? That said, to me the question of whether  an LLM is conscious or not makes no sense,   [06:00] the question is how we quantify its consciousness.  This is why I give all these papers a 9 out of 10   on the bullshit meter. They fail at the relevant  question: if you can’t measure it, what are you   [06:14] even talking about? Though I have to admit if  Claude has thoughts it does not say out loud. This   already puts it ahead of most people on social  media. Thanks for watching. See you tomorrow.