There's no such thing as "true" general intelligence. There's just a bunch of specific cognitive traits that happen to (usually) be positively correlated with each other. Some proxies are more indicative than others (in the sense that getting high scores on them consistently correlate with doing well on other proxies), and that's about the best you can hope for.
Within the human range of intelligence and domains we're interested in, IQ is decent, so are standardized test scores, so (after adjusting for a few things like age and location of origin) is income, so is vocabulary, so (to a lesser degree) is perception of intelligence by peers, and so forth.
I am not asking about ‘true’ general intelligence? Or whatever that implies.
If your not sure, I am asking regarding the term commonly called ‘general intelligence’, or sometimes also known as ‘general mental ability factor’ or ‘g-factor’, in mainstream academic papers. Such as those found in pedagogy, memetics, genetics, etc…
See: https://scholar.google.com/scholar?hl=en&as_sdt=0%252C5&q=“general+intelligence”&btnG=
Where many many thousands of researchers over the last few decades are referring to this.
Here is a direct quote by a pretty well kno...
And the only practical way to realize this, that I can think of now, is by predicting the largest stock markets such as the NYSE, via some kind of options trading, many many many times within say a calendar year, and then showing their average rate of their returns is significantly above random chance.
The threshold for doing this isn't being above average relative to human individuals, it's being close to the top relative to specialized institutions. That can occasionally be achievable, but usually it isn't.
Well I agree it is a much higher bar than just ‘above average’, yet it still seems like the easiest way of delivering a credible proof that can’t be second guessed somehow. (That I can think of, hence the post)
Since ‘cheating’ at this would also mean that the person somehow has gained insider information for a calendar year that was above and beyond what the same ‘specialized institutions’ could obtain.
Which is so vanishingly unlikely that I think pretty much everyone (>99% of readers) would accept the results as the bonafide truth.
But it probably is limited only to literal geniuses and above as a practical mechanism.
No one that I know based on first hand information, otherwise I probably would have included them as an example.
Most likely due the fact that people who succeeded at such have little reason to advertise it beyond a small circle.
Standardized tests work, within the range they're testing for. You don't need to overthink that part. If you want to make people's intelligence more legible and more provable, what you have is more of a social and logistical issue: how do you convince people to publish their test scores, get people to care about those scores, and ensure that the scores they publish are real and not the result of cheating?
Which tests are you referring to and how do they exactly measure general intelligence?
(And not say IQ or how much the test taker crammed…)
Proving to whom, and to what degree of credence, and avoiding what correlated measures? Standardized tests, like the SAT, are perfectly fine at showing "above average", for most purposes. They're not very fine-grained, so it's hard to tell "very slightly above average", and not very accurate for outliers, so can't distinguish between 90th and 99th percentile with much certainty. But they're just fine for showing "very likely above average" for a lot of uses.
I'm not sure there exists a formal operational definition of "general intelligence", there's no direct measurement possible. Still, if "above average or not above average" is the criteria, most correlates are usable.
Since I’m asking LW readers I imagine the default ‘degree of credence’ for any proof is something that the vast majority of LW readers will accept as the actual, bonafide, truth and are willing to acknowledge this when presented with it.
And predicting the future on a global scale, successfully, repeatedly, and precisely, has no correlated measures, if we assume precognition is impossible.
So we can conveniently sidestep this issue. Hence why I mentioned it…
I'm not sure there exists a formal operational definition of "general intelligence", there's no direct measurement possible
Then how can anyone prove, in the future, whether an AGI exists, or not?
Oh. Your model of LW readers is very different from mine - I doubt there exists anything that "the vast majority" will accept as "actual, bonafide truth". In fact, words like actual and bonafide are likely to confuse most of us, and we'll want clarification.
Then how can anyone prove, in the future, whether an AGI exists, or not?
I don't think anyone can (and I don't think they'll have to - it'll either be self-evident or irrelevant). But you said you didn't want to extrapolate to that anyway.
So then what is the issue you want to discuss?
Neither of us can do more than offer our guesses and opinions on these two points.
Many talk about IQ in many places, in all kinds of contexts and situations, and IQ can be ‘provable’ to a certain degree of reliability and consistency using methods such as Raven’s progressive matrices.
But how is the general intelligence of the individual proven to any degree of reliability/consistency/validity/etc…?
The only way I can think of is successfully predicting the future in a way and publicly sharing the predictions before and after, many many many times in a row. And in such a way that no interested parties could realistically help them fake it.
And the only practical way to realize this, that I can think of now, is by predicting the largest stock markets such as the NYSE, via some kind of options trading, many many many times within say a calendar year, and then showing their average rate of their returns is significantly above random chance.
And even this is kind of iffy, since it would require sharing most of their trades publicly, along with the possibility of ‘cheating’ with insider information.
Has anyone thought about the exact methods that are feasible?
Edit: Any such methods would probably also apply to AI but I don’t want to extrapolate too far.