Embed Notice

@ciggysmokebringer @neonsnake I don't think that human modes of cognition and non-human modes of cognition are completely separate spheres that have nothing in common, so this isn't news to me. We evolved from a common ancestor, for heaven's sake! But I definitely think it's absurd to attribute capabilities to animals that they clearly don't have, or to insist that there is not a vast separation in human cognitive abilities and (self-)consciousness from most other animals, a gulf that is composed both of quantitative changes and qualitative ones, as we have many entire capabilities that most other animals don't have. That's the sort of blindness that really frustrates me about how people talk about large language models — they claim that LLMs are sentient or conscious or intelligent when they don't even have the basic neurological structures and designs and reward functions to be so, it just so happens that they can string a semicoherent group of words together in the right conditions and have an architecture that is vaguely like the basic architecture that our brains also share. It's like claiming that an arcade machine and a supercomputer are basically the capable of the same things because they both use logic gates.

And I'd like to point out that I'm not in principle closed to being proven wrong on this, because I don't really have anything against the idea that there may be animals that are equivalent to humans in cognition and intelligence and consciousness or whatever. I'm not committed to only seeing humans as moral subjects and/or moral agents, or thinking we're special and only we have "souls" or whatever. I'd be totally willing to grant personhood to any animal you like if you can show me that has the qualitative capacities to have it. It's just that I've done a fairly significant amount of research into stuff like Koko and Noam Chipskey and all the other animals that people claim have achieved that kind of thing and I'm just deeply unconvinced.