Elon Musk (says, at least that he) is working to make a maximally curious, maximally truth-seeking, AI.The legal principle is that official facts are determined by juries. So ultimately a truth-seeking AI must become a prosecutor/plaintiff lawyer and, equally, a defense lawyer.
Stipulating that not every thing that every Republican says is going to be true, Adam Schiff Democrats have no analog in the Republican Party. Unless you count rogues like Liz Cheney or (2012 Republican nominee for POTUS!) Mitt Romney, whom mainstream Republicans repudiate. The reason is of course the fact that the business interest of any journalist favors excitement/fear inducing stories, and such stories need not actually be true.
It occurs to me that Jordan Peterson says a couple of germane things. One is that it is all very well to get all absorbed by the issue of intelligence - but there is such a thing as wisdom - which he says is not only different from, but actually uncorrelated with, IQ. But nobody is explicitly talking about “artificial wisdom.”
Another is that Peterson asserts that
This is somewhat counterintuitive. But since conscientiousness and openness are uncorrelated, I suppose someone like Elon Musk can be very high in both dimensions of personality.
- conscientiousness is highly correlated with conservatism, whereas
- openness/creativity is very highly correlated with opposition to conservatism.
The other three dimensions of personality, BTW, are extroversion, agreeableness, and anxiousness. Seems odd that the “five dimensions of personality,” are all uncorrelated with IQ and with each other, and yet Peterson never suggests how wisdom fits with the “five” personality traits. Does it stand alone, like IQ?
As an aside, it has always been my understanding that "IQ" pertains only to the specific quantitative test that is comprised of elements of reasoning, actual knowledge of any given subject within the test, and critical or logical thinking. Regardless, I've believed that intelligence without wisdom produces a very dopey person. Further, a smart person can be unintelligent. I consider myself "smart" but not very intelligent as I only know a little bit about many, many subjects. Regardless, your post is interesting and caught my on a day where I'm very mentally sluggish. 😜
A Maximally Truth-Seeking AI must start training in math, and work up to logic from that foundation.
Later, we can argue “what is truth” with the thing...to train it to recognize the truth.
BUT...how do we get our hands on it right now?
ON A PREDICTION BASIS: The AI’s behavior can be predicted by the materials and biases used to train it.
Know the trainers & programmers, know what to predict.
Thus: George Washington was black, and white people are not placed in images.
With the current set of trainers, AI will seek to support dictatorial powers pushing collectivism & social marxism.
But since conscientiousness and openness are uncorrelated,
= = =
I think one of the 16 Meyers-Briggs psycho combinations is like this.
Hey. Put a selector switch into the AI machine. To designate which of the 16 types that it will perform as.