3 ms·

> We will probably soon build the electronic computer which will, from the perspective of the humans communicating with him, indeed behave "as a conscious perso

> We will probably soon build the electronic computer which will, from the perspective of the humans communicating with him, indeed behave "as a conscious person": soon we won't need you to write the answer you wrote, the computer will be able to make even better one. This is wrong. We are absolutely nowhere near anything even remotely resembling this.

acqq 8y ago
No. We already have a complete nematode implemented as a computer program. We already have a realistic simulation of the living creature. Twenty years ago the current output of Google Translate was a pure science fiction, compared to what it was possible then. Today it already often translates better than a lot of humans knowing both languages. Had anyone seen the output of Google Translate of today some 20 years ago, nobody would have believed it was produced by the machine. Note I don't say that it's always good. It's not. But it can produce amazing outputs already, which can't be recognized to be produced by a computer. Also read about AlphaZero: https://en.wikipedia.org/wiki/AlphaZero#Training https://en.wikipedia.org/wiki/AlphaZero#Training "AlphaZero was trained solely via self-play, using 5,000 first-generation TPUs to generate the games and 64 second-generation TPUs to train the neural networks. In parallel, the in-training AlphaZero was periodically matched against its benchmark (Stockfish, elmo, or AlphaGo Zero) in brief one-second-per-move games to determine how well the training was progressing. DeepMind judged that AlphaZero's performance exceeded the benchmark after around four hours of training for Stockfish, two hours for elmo, and eight hours for AlphaGo Zero." Apparently, the computer learned chess better than the long developed programs by only playing the game with itself for four hours! And the programs are already better than humans there.
- root_axis 8y ago
  > We already have a complete nematode implemented as a computer program. We already have a realistic simulation of the living creature. It is not, by any stretch of the imagination, "a complete nematode". It is very much a simple simulation that isn't at all close to the actual complexity of a living organism, even a worm. We can't even simulate a bacteria with the kind of fidelity required to claim that it approaches the complexity of a living organism. > Twenty years ago the current output of Google Translate was a pure science fiction I think you're giving google translate way too much credit. Compared to 20 years ago, yes, much has improved, but compared to a human native speaker it is pitiful. > Also read about AlphaZero: I am aware of Alpha Zero, it is an impressive but narrowly applicable use of machine learning that doesn't at all represent anything nearing the cognitive complexity of an animal. It is a complex achievement of applied statistics, not an intelligence. The "world" of Alpha go is a very strictly defined finite universe where the only possibilities are the clean and simple rules of the go board. Yes, the combinatorial explosion of potential moves is impractical to brute force, but ultimately the rules of go are so simple and discretely defined that its worthless as a comparison to a living intelligence.
- kamaal 8y ago
  >>Apparently, the computer learned chess better than the long developed programs by only playing the game with itself for four hours! https://www.theatlantic.com/magazine/archive/2013/11/the-man-who-would-teach-machines-to-think/309529/ https://www.theatlantic.com/magazine/archive/2013/11/the-man... Basically the machine isn't exactly thinking. It simulates a model of thinking. There lies the biggest difference.
  - lisper 8y ago
    Congratulations, you have re-discovered Searle's Chinese Room argument.

acqq 8y ago

No. We already have a complete nematode implemented as a computer program. We already have a realistic simulation of the living creature. Twenty years ago the current output of Google Translate was a pure science fiction, compared to what it was possible then. Today it already often translates better than a lot of humans knowing both languages. Had anyone seen the output of Google Translate of today some 20 years ago, nobody would have believed it was produced by the machine. Note I don't say that it's always good. It's not. But it can produce amazing outputs already, which can't be recognized to be produced by a computer. Also read about AlphaZero: https://en.wikipedia.org/wiki/AlphaZero#Training https://en.wikipedia.org/wiki/AlphaZero#Training "AlphaZero was trained solely via self-play, using 5,000 first-generation TPUs to generate the games and 64 second-generation TPUs to train the neural networks. In parallel, the in-training AlphaZero was periodically matched against its benchmark (Stockfish, elmo, or AlphaGo Zero) in brief one-second-per-move games to determine how well the training was progressing. DeepMind judged that AlphaZero's performance exceeded the benchmark after around four hours of training for Stockfish, two hours for elmo, and eight hours for AlphaGo Zero." Apparently, the computer learned chess better than the long developed programs by only playing the game with itself for four hours! And the programs are already better than humans there.

root_axis 8y ago
> We already have a complete nematode implemented as a computer program. We already have a realistic simulation of the living creature. It is not, by any stretch of the imagination, "a complete nematode". It is very much a simple simulation that isn't at all close to the actual complexity of a living organism, even a worm. We can't even simulate a bacteria with the kind of fidelity required to claim that it approaches the complexity of a living organism. > Twenty years ago the current output of Google Translate was a pure science fiction I think you're giving google translate way too much credit. Compared to 20 years ago, yes, much has improved, but compared to a human native speaker it is pitiful. > Also read about AlphaZero: I am aware of Alpha Zero, it is an impressive but narrowly applicable use of machine learning that doesn't at all represent anything nearing the cognitive complexity of an animal. It is a complex achievement of applied statistics, not an intelligence. The "world" of Alpha go is a very strictly defined finite universe where the only possibilities are the clean and simple rules of the go board. Yes, the combinatorial explosion of potential moves is impractical to brute force, but ultimately the rules of go are so simple and discretely defined that its worthless as a comparison to a living intelligence.
kamaal 8y ago
>>Apparently, the computer learned chess better than the long developed programs by only playing the game with itself for four hours! https://www.theatlantic.com/magazine/archive/2013/11/the-man-who-would-teach-machines-to-think/309529/ https://www.theatlantic.com/magazine/archive/2013/11/the-man... Basically the machine isn't exactly thinking. It simulates a model of thinking. There lies the biggest difference.
- lisper 8y ago
  Congratulations, you have re-discovered Searle's Chinese Room argument.