There are ways of controlling the new technology—but first we have to stop mythologizing it.
This article articulates what I have been thinking - so called AI has been with us for decades and giving it a new name is sensationalism.
Comments appreciated.
The article is interesting, but does not precisely describe the problems
with the sensational MISUNDERSTANDINGS among American citizens,
as to what artificial intelligence is, or what its dangers are.
For those who are not familiar with the historical development of artificial
intelligence, I include these points...
-- AI is a sub discipline of Computer Science
-- AI is not defined by sci-fi movies or books about "artificial intelligence"
-- Computer Science has proven a large number of computationally
intractable problems. Many of these, are involved in complex
human problem solving. No matter how fast computers become,
these problems will remain intractable, at the size of variables that
represent real life.
-- AI is almost all implemented in software programs
-- AI accomplishments are seriously constrained by problems that
Computer Science has proven to be computationally intractable
-- AI implemented using artificial software "neurons" (such as neural
nets) DO NOT really emulate the complexities of biological neurons
in the human brain
-- AI is the emulation of complex human problem solving, not the
automation of simple human jobs, nor the computation of answers that
human beings cannot do (such as computing 100 million mathematical
answers a second)
-- AI is NOT artificial life, or the creation of artificial "consciousness"
-- Deep learning nets are actually in the dumbest class of AI algorithms
-- AI is not the Tom Cruise "Don't think -- just do!" approach. The black
box in Tom Cruises jet, can still precisely describe what the pilot DID.
But the neural net algorithm approach, does not have semantics attached to
specific memory paths, and so neural nets CANNOT be explained in ways
that a human being would understand. This entire class of AI algorithms
is described as "sub-logical", and because answers cannot be carefully
traced, there is a builtin danger of not being able to demonstrate why
the software answer should be considered valid or trustworthy.
-- "Machine learning" is not particularly AI. Computing a mean average,
is a form of machine learning. But it is not complex human problem solving.
-- Most software programs do not have the ability to reason about their
own logical approach to problem solving, and change it, as a human being does.
-- No AI tools that I know of, have a builtin Moral-Ethical model that constrains
what they can do. The hard sciences do not have the vocabulary to describe
virtues or vices, righteous behavior or evil behavior. Because of this, AI tools
are a TERRIFIC danger, as they will be used by immoral, evil criminal gangs.
This is just a beginning of current misunderstanding, about "artificial
intelligence".
Do you REALLY think that politicians running around in Congress, who
believe the ridiculous conspiracy theories of Tucker Carlson, have the intelligence to
pass legislation that will practically stop the horrendous abuse of AI tools?
The emergence of (even dumb) AI tools, is like the emergence of the
atomic bomb. But in the case of artificial intelligence, we do not have
intelligent scientists like Einstein and Oppenheimer agonizing over what
will happen, when this technology is gotten and used to destroy entire nations.
We have a bunch of politician-Barbies in congress, who use a lot of language
that may make it seem that they know what artificial intelligence is, who
actually have no idea what artificial intelligence is.