3 ms·

If AI can barely replace car drivers or truck drivers, I think developers will unfortunately have to keep writing code. Of course AI made progress in the langu

by jokoon 2y ago

If AI can barely replace car drivers or truck drivers, I think developers will unfortunately have to keep writing code. Of course AI made progress in the language realm with chatGPT, but I don't see any advance in the realm of consciousness, general AI, emotions, etc. There was some recent, small progress where the brain of a fly was mapped. But I don't see transverse research made between cognition, psychology, neuroscience and AI, which is terrible. Science doesn't even have any relevant insight on a good definition of intelligence that could be transposed into an algorithm. There is also quantum biology, that argues that many biological systems use quantum mechanics. Maybe quantum computing will offer new horizons with additional computing power but it will maybe take 20 more years.

gpderetta 2y ago
We have little reason to believe that intellectual jobs are harder to automate than physical jobs. Consider the Moravec's paradox (https://en.wikipedia.org/wiki/Moravec%27s_paradox https://en.wikipedia.org/wiki/Moravec%27s_paradox).
- snakeyjake 2y ago
  It depends. Many "intellectual" jobs build things that have never existed before and I have never seen an automation for something that has never existed before. I have also not yet seen a model create something new that was not already contained within the dataset it was trained on. An astronaut on a horse is only superficially original and does not count as "new". Of course, I am not omniscient so that may have already occurred but there is so much bullshittery in the "AI" (it's not I) space that any claim that it has occurred should be quarantined into the "likely a lie" zone of of one's brain until there is overwhelming evidence in the amount needed to convince an astronaut who has been in orbit and gazed upon the Earth that the planet is flat.
Lichtso 2y ago
> Maybe quantum computing will offer new horizons with additional computing power When somebody starts talking about quantum computing outside cryptography I immediately know that they have no idea what they are talking about. That is because they think quantum computing is a faster general purpose computer, which it definitely is not. Instead it has its own complexity classes [0] and most of the problems we care about in general purpose computing seem to not be in there. This to me strongly suggests that biological brains would not gain any advantage by using quantum computing and are thus very unlikely to do so. [0]: https://en.wikipedia.org/wiki/Quantum_complexity_theory https://en.wikipedia.org/wiki/Quantum_complexity_theory

by jokoon 2y ago

gpderetta 2y ago

We have little reason to believe that intellectual jobs are harder to automate than physical jobs. Consider the Moravec's paradox (https://en.wikipedia.org/wiki/Moravec%27s_paradox https://en.wikipedia.org/wiki/Moravec%27s_paradox).

snakeyjake 2y ago
It depends. Many "intellectual" jobs build things that have never existed before and I have never seen an automation for something that has never existed before. I have also not yet seen a model create something new that was not already contained within the dataset it was trained on. An astronaut on a horse is only superficially original and does not count as "new". Of course, I am not omniscient so that may have already occurred but there is so much bullshittery in the "AI" (it's not I) space that any claim that it has occurred should be quarantined into the "likely a lie" zone of of one's brain until there is overwhelming evidence in the amount needed to convince an astronaut who has been in orbit and gazed upon the Earth that the planet is flat.

Lichtso 2y ago

> Maybe quantum computing will offer new horizons with additional computing power When somebody starts talking about quantum computing outside cryptography I immediately know that they have no idea what they are talking about. That is because they think quantum computing is a faster general purpose computer, which it definitely is not. Instead it has its own complexity classes [0] and most of the problems we care about in general purpose computing seem to not be in there. This to me strongly suggests that biological brains would not gain any advantage by using quantum computing and are thus very unlikely to do so. [0]: https://en.wikipedia.org/wiki/Quantum_complexity_theory https://en.wikipedia.org/wiki/Quantum_complexity_theory