6 ms·

This McCarthy quote from the (linked-to) essay caught my eye: How the unconscious goes about its work is not so much poorly understood as not understood at all

This McCarthy quote from the (linked-to) essay caught my eye: How the unconscious goes about its work is not so much poorly understood as not understood at all. It is an area pretty much ignored by the artificial intelligence studies, which seem mostly devoted to analytics and to the question of whether the brain is like a computer. They have decided that it’s not Is that true? Who has decided the brain is not like a computer? That is news to me.

simonh 9y ago
It depends what is meant by computer. Certainly they don't have anything analogous to source code, registers and the operations that CPUs carry out on them, addressable memory, etc. Nurons are also different in many significant respects from the analogous elements in computational neural nets. However there are a lot of flat statements like this made in the article where I’m not entirely sure what the author meant or what grounds he had for saying it and this is one of them. Another is: "The invention of language was understood at once to be incredibly useful. Again, it seems to have spread through the species almost instantaneously." How can anyone possibly know that, especially if we have no real idea when it happened? Edit: Another thing that got my back up a bit were the disdainful references to 'influential persons', as though David Krakauer and Cormac McCarthy are renegade outsiders. There are some really good points here that I largely agree with. Most of what we do or know or think or even say is processed or generated by the unconcious mind. I know what I am going to write here and now, but I don't know it in linguistic form until I actually write it. I'm generating it in a non-English language representation in my brain and converting that written into English and only part of that pipeline is really conscious.
ppod 9y ago
I love McCarthy's work but this is nonsense. Current AI work is all about the relationship between differdntiable connectionist systems and algorithmic/symbolic computation. AI researchers are well aware of the parallels between this and subconscious/conscious human thought and things like Kahnemann's system1 vs system2 thinking. "Is the brain like a computer" is too ambiguous a question to take seriously. His article is interesting in parts but really arrogant and not as novel or anti-mainstream as he seems to think it is. It sounds like someone who spent a long time chatting with physicists or biologists and no time reading research in artificial intelligence, at least from the last decade.
thisisforyou 9y ago
I don't think there is any reason to believe that the brain is like a computer in any meaningful way. Both the brain and an electronic computer are able to operate on information, but beyond that the differences between the two far outweigh the similarities on pretty much every level. Even neural nets (that is, the machine learning algorithm) are just general, very simplified, models of how some parts of the brain are constructed. True, neural nets can do things similar to things that the brain can do (like describing images) but this is really only a facile similarity, to my understanding.

simonh 9y ago

It depends what is meant by computer. Certainly they don't have anything analogous to source code, registers and the operations that CPUs carry out on them, addressable memory, etc. Nurons are also different in many significant respects from the analogous elements in computational neural nets. However there are a lot of flat statements like this made in the article where I’m not entirely sure what the author meant or what grounds he had for saying it and this is one of them. Another is: "The invention of language was understood at once to be incredibly useful. Again, it seems to have spread through the species almost instantaneously." How can anyone possibly know that, especially if we have no real idea when it happened? Edit: Another thing that got my back up a bit were the disdainful references to 'influential persons', as though David Krakauer and Cormac McCarthy are renegade outsiders. There are some really good points here that I largely agree with. Most of what we do or know or think or even say is processed or generated by the unconcious mind. I know what I am going to write here and now, but I don't know it in linguistic form until I actually write it. I'm generating it in a non-English language representation in my brain and converting that written into English and only part of that pipeline is really conscious.

ppod 9y ago

I love McCarthy's work but this is nonsense. Current AI work is all about the relationship between differdntiable connectionist systems and algorithmic/symbolic computation. AI researchers are well aware of the parallels between this and subconscious/conscious human thought and things like Kahnemann's system1 vs system2 thinking. "Is the brain like a computer" is too ambiguous a question to take seriously. His article is interesting in parts but really arrogant and not as novel or anti-mainstream as he seems to think it is. It sounds like someone who spent a long time chatting with physicists or biologists and no time reading research in artificial intelligence, at least from the last decade.

thisisforyou 9y ago

I don't think there is any reason to believe that the brain is like a computer in any meaningful way. Both the brain and an electronic computer are able to operate on information, but beyond that the differences between the two far outweigh the similarities on pretty much every level. Even neural nets (that is, the machine learning algorithm) are just general, very simplified, models of how some parts of the brain are constructed. True, neural nets can do things similar to things that the brain can do (like describing images) but this is really only a facile similarity, to my understanding.