Artificial Intelligence: Savior, Antichrist, or Hyperbole?
By
Duane Thresher, Ph.D. September 11, 2018
Artificial intelligence is going to save humanity! Artificial
intelligence is going to destroy humanity! Artificial
intelligence is going to sell magazines and get research
funding! Which is it? And what is it?
Artificial Intelligence (AI) has been in the news a lot
recently, including from
my alma
mater MIT, where I got a BS in Electrical Engineering and
Computer Science and where I used to hang out at the AI Lab
(which was funded by the military). I later wrote parts of
some of the most sophisticated computer programs in existence
— climate models (for example, see
NCAR's
climate model and search for "Thresher") — and AI is
just programming (which will make it another victim of
IT
incompetence).
AI has been most notoriously in the news as the war of words
between Tesla's Elon Musk and Facebook's Mark Zuckerberg.
Watching IT incompetent Musk and
IT
incompetent Zuckerberg arguing about AI reminded me of the
saying "those of you who think you know it all are very
annoying to those of us who do".
But what is AI? The public, who pays for the research so gets
to decide what it is, thinks of AI as a computer that is
smarter than humans. Like HAL in the 1968 movie 2001: A Space
Odyssey (for each letter in HAL, go to the letter after it in
the alphabet and the result is IBM). Or Colossus in the 1970
movie. Or The Terminator in the 1984 movie. These led to the
perception of AI as Antichrist, cleverly trying to kill us.
This doomsday scenario is then played upon by media hounds
like Elon "AI-as-Antichrist" Musk just to get attention,
particularly investors' attention (when you plow through money
without making any, you need constant investment).
Those whose fantasy is to be saviors and who want to be
perceived as intelligent themselves — although usually
IT incompetent, like Mark "AI-as-Savior" Zuckerberg —
and researchers who want funding, insist that AI will save
humanity by saving us from our stupidity, as if AI wasn't
created by humans and not destined to have all our problems,
which may be inherent (God created Man in his own image and
Godel's Theorem says all mathematical systems, like logical
thought, have contradictions).
Google
is very much into AI as Savior, including its research at MIT
(IT incompetent MIT President Reif is a big fan of AI for the
"public good", but doesn't say who gets to decide that), but
Google's use of AI for mass privacy invasions and censorship
would make it more of AI as Antichrist. Same for
Facebook.
Interestingly, in George Orwell's 1984, the privacy-invading
censoring Big Brother could be a computer, and while
introducing its Macintosh computers in the 1980's, Apple's
award-winning commercials were about smashing Big Brother,
suggested to be IBM.
The media,
IT
incompetent as it is, portrays AI as Savior or Antichrist,
whichever sells more ads, and always uses Hyperbole, with a
capital H.
Robots are often thought of as AI, even though robots these
days are dumb. A robotic vacuum cleaner is not intelligent
just because it can maneuver around furniture; any mouse can
do better. The Terminator came from the future, 2029, which
is just 10 years away now and no robot even remotely as
advanced yet exists. In Star Trek: The Next Generation, which
takes place in the 2360's, the good robot Data is portrayed as
being uniquely ahead of his time but still not really humanly
intelligent. And he has an evil twin.
Expert systems are often thought of as AI. But expert systems
know only about one subject, like IBM's chess computer Deep
Blue (programming a computer with an encyclopedia so it can
win Jeopardy like IBM's computer Watson does not make it an
expert on even one subject). At best that makes them idiot
savants. Moreover, these expert systems are not at all
humanly intelligent in their interaction with humans. If
you've ever typed a question into Google Search, you know how
fruitless and frustrating that can be, just like using Amazon
Echo/Alexa or speech recognition programs or language
translation programs or ...
In 1950 Alan Turing proposed the Turing Test, which is if you
conversed blindly with something, about any subject, and you
can't tell if that something is a human or a computer, then if
it is really a computer, it has passed the Turing Test and is
AI. Turing predicted that by 2000 a computer would exist that
could fool 30% of the public, but to date no computer has
passed the Turing Test to the degree the research-paying
public would consider AI.
Also in 1950, when asked how long before there would be
commercial fusion (clean) nuclear power (not the commercial
fission (dirty) nuclear power we have today), scientists,
including those at MIT, answered "50 years". In 2000, when
they were asked the same question, they gave the same answer.
(And the DeLorean car's Mr. Fusion power source in the 1985
movie Back to the Future is from 2015.) AI is exactly the
same as fusion in always being 50 years in the future.
Moreover, fusion is easier than AI.
I've lived long enough to see every new technology,
particularly computers, be touted as the solution to the
"education problem" and seen every one of them fail. In fact,
computers are often
the problem in education these
days. AI is just more of the same.
Same for every other field.
In fact, AI has had multiple heydays since 1950. We just
happen to be in one now. I can't wait until this one is over,
although I'll probably be around for the next, unless, as has
already happened to others, I'm killed by a Tesla or Google
self-driving smart car (I'd rather take my chances with a New
York cabbie, which says a lot).
Thus, any intelligent being, artificial or human, has to
conclude that AI is just Hyperbole, with a capital H, not
Savior or Antichrist.