Intelligence cannot be boiled down to a single number, but it can be boiled down to about five numbers. If you gave somebody who was unable to grasp high levels of abstraction inhuman processing speed (say, 15 standard deviations), then this qualitative difference would not make a quantitative difference (so FSIQ scores aren’t equal, the subtests which result in the final score are important, too). Also, the intelligence distribution is a bit weird than our models suggest, which is why geniuses can be so far apart from eachother. Nikola Tesla could visualize his designs in 3D space, just how many bits of information could this visual working memory of his contain? The standard deviation of working memory items is roughly 1, so even 50 items ‘should’ only appear in one in 10^400 people.
I don’t believe much in super-intelligence, but I have recently had the horrifying thought that most plans have a threshold on required intelligence, and that further intelligence doesn’t make much of a difference (In short, that scaling is unreasonably effective). This means that an AI with 140~ IQ and 100000 times more actions per second than a human, could take over the world, even if I could beat it at some IQ test subsets. That competitive RTS game players aim to improve their APS (actions per second) speaks in favor of this idea.
Is there any reason you make Humman so obviously wrong and dislikable? If any reader believes something that Humman does, I think they will feel offended and close the page before they to read the counter-arguments to the beliefs he represents.
Ideally, by the way, we’d gatekeep things like AI research in a manner that people like Humman will be barred from entering. There are classes in topics like security and biology which are good at destroying many of the naive arguments Humman represents in the post.
Finally, beliefs like “everyone is born with the same amount of stat points” are spread even in Harvard (e.g. Howard Gardner’s theory of multiple intelligences) and it’s downstream from the social instincts which guard against outliers. The hard sciences are getting less hard by the year because of ideological, political and social dynamics, and these are directly responsible for the poor general understanding of intelligence. Human beings fail to understand AGI because they anthropomorphize it (they consider their own humanity axioms that all systems are bound by).
Intelligence cannot be boiled down to a single number, but it can be boiled down to about five numbers. If you gave somebody who was unable to grasp high levels of abstraction inhuman processing speed (say, 15 standard deviations), then this qualitative difference would not make a quantitative difference (so FSIQ scores aren’t equal, the subtests which result in the final score are important, too). Also, the intelligence distribution is a bit weird than our models suggest, which is why geniuses can be so far apart from eachother. Nikola Tesla could visualize his designs in 3D space, just how many bits of information could this visual working memory of his contain? The standard deviation of working memory items is roughly 1, so even 50 items ‘should’ only appear in one in 10^400 people.
I don’t believe much in super-intelligence, but I have recently had the horrifying thought that most plans have a threshold on required intelligence, and that further intelligence doesn’t make much of a difference (In short, that scaling is unreasonably effective). This means that an AI with 140~ IQ and 100000 times more actions per second than a human, could take over the world, even if I could beat it at some IQ test subsets. That competitive RTS game players aim to improve their APS (actions per second) speaks in favor of this idea.
Is there any reason you make Humman so obviously wrong and dislikable? If any reader believes something that Humman does, I think they will feel offended and close the page before they to read the counter-arguments to the beliefs he represents.
Ideally, by the way, we’d gatekeep things like AI research in a manner that people like Humman will be barred from entering. There are classes in topics like security and biology which are good at destroying many of the naive arguments Humman represents in the post.
Finally, beliefs like “everyone is born with the same amount of stat points” are spread even in Harvard (e.g. Howard Gardner’s theory of multiple intelligences) and it’s downstream from the social instincts which guard against outliers. The hard sciences are getting less hard by the year because of ideological, political and social dynamics, and these are directly responsible for the poor general understanding of intelligence. Human beings fail to understand AGI because they anthropomorphize it (they consider their own humanity axioms that all systems are bound by).