I first posted on this topic, regarding Ludwig Wittgenstein's view of language, over ten years ago. My ideas have evolved quite a bit since then, and I decided that I should update what I had to say on the original post. In that post, I outlined how a lion might at least understand some human symbolic language if it were given the right exposure. This was based on the fact that some dogs do understand at least one symbolic action: the meaning of pointing. Similarly, domestic cats also understand some symbolic communications made by humans. Therefore, if cats and dogs could speak, we would definitely be able to understand them to some extent. I didn't mention it at the time, but dolphins, whales, elephants and other mammals have their own languages which, with further study, we might be able to understand. As far as I know, Wittgenstein never owned a pet.
Wittgenstein would be technically correct in the sense that all species have their individual evolutionary paths, so their priorities might be completely different from ours and unintelligible to us. However, the languages of organisms are probably all survival-related, hence it would not necessarily be difficult to decipher them. Many animals make sounds indicating the presence of dangers in their immediate environment that would be easy for us to understand through observation. Similarly, it wouldn't be hard to figure out calls for help. The fact that we have complex symbolic language is the result of our development of bipedal gait, our close cooperation with each other over thousands of years, and the changes that occurred in our brain function that became encoded in our genes and permitted us to survive as a species. I would say that Wittgenstein didn't realize that language starts as a kind of noise-making that has survival benefits, and, if you look at various noise-making species, I don't think that it is necessarily difficult to interpret a noise-making behavior if you examine it in the context in which it is made.
I also mentioned Noam Chomsky, who has said that we don't understand how language works in the human brain. In the context of AI, he thought that we could never teach AI how to think, because we don't know how we think. This was before the development of large language models, which are currently at the forefront of AI. I don't study the research in these fields, but it appears to me that, although words, grammar and usage did develop in human brains, it isn't necessary to have a human brain or human learning techniques to learn languages. In fact, it is probably likely that all of our ideas could be expressed without human brains or human language. I think that the messiness of neural development, as described by Robert Sapolsky, indicates that languages that perform better than ours could probably be developed. That is because our languages developed through the rather haphazard process of natural selection. In natural selection, what counts is not necessarily the development of the best possible system, and it all boils down to whether each successive generation survives or dies. This is why I expect AGI to be developed before long, and I see no reason why it would have to function much like the human brain.
Wittgenstein, I now think, received far more credit than he deserved as a philosopher. Because of his personality and his initial acceptance by English academics, he never had to argue or publish his positions after he received his job at Cambridge. It sounds to me as if most of his career involved lectures with no discussion, at which his devotees simply took notes. That was certainly a dream job!
No comments:
Post a Comment
Comments are moderated in order to remove spam.