Higgenbotham wrote:
> A computer valuing equity is like a computer valuing a
> prostitute. All the computer can do is guess what Joe might be
> willing to pay for a prostitute because the computer has no use
> for a prostitute. It is Joe who actually determines the value of
> the prostitute because he has a use for one. If computers were the
> only entities in the market for prostitutes, the value of a
> prostitute would go to zero.
I think the way to understand this is to separate the goal
from the implementation.
Suppose you have an AI computer that knows how to find and
kill people. It won't know whom to kill unless if has a specific
goal.
So if the goal is "kill anyone," then it will kill anyone.
If the goal is to "kill prostitutes," then it will figure out
who the prostitutes are, and it will kill them.
If the Americans set the goal to "kill jihadists," then it
will figure out from the internet who the jihadists are, and
it will go out and kill them.
If the Chinese set the goal to "kill Americans," then it will
figure out where the Americans are, and go out and kill them.
So you have exactly the same AI software in all four cases, but with
different goals.
As I wrote previously, computers have been able to value equity for
decades according to various algorithms. This is a pretty well
understood these days. So, for example, one could set a goal for an
AI computer to do a historical analysis of equity values to determine
the best algorithm for evaluating equities. And then the results
could be used to predict equity values in the future, and the system
will in time do a better job at it than the best humans.