AI, or Artificial Intelligence

Mary Shelley’s Frankenstein story was not created out of thin air.  Indeed, her novel’s subtitle is The Modern Prometheus.  Prometheus, a Titan of Greek mythology, brought fire to humanity and enraged the gods who knew that humans would eventually challenge their supremacy.  Dr. Frankenstein creates a creature from human body parts, and the creature demands his right to be treated as a human.  In Blade Runner (1982), replicants (bioengineered humanoids) demand their rights to be treated as humans.  In Short Circuit (1986), a military robot convinces the film’s humans that he has human consciousness (or a soul) when he laughs at a joke.  We have always thought that consciousness was a peculiarly human trait not shared by our animal cousins (despite the fact that science has no clear and agreed upon definition of “consciousness”); we have wised up lately.  Lest I forget, the origin story of all these stories is Genesis’s Garden of Eden, where our ancestors are told not to eat of the Tree of Knowledge of Good and Evil, “For God doth know that in the day ye eat thereof, then your eyes shall be opened, and ye shall be as gods, knowing good and evil” (KJV, Genesis 3:5).

AI, and robots, are the next steps in the evolution of human thought.  In the list above, only the comedy, Short Circuit, ends well.  Nevertheless, AI – and robotics – has always been the Holy Grail of computing.  Every super-gifted mind in today’s world wants to be one of those who finds/discovers/creates that Holy Grail.  Despite (perhaps because of) all the warning signs that litter the roads of history.

Technology.  Technology is making us more productive.  And less smart.

Atomic Energy (Albert Einstein, J. Robert Oppenheimer) is on its own neither good nor ill.  Its first uses destroyed the lives of a few hundred thousand human beings.  But the continued use of atomic energy is the only credible interim energy source on the way to worldwide solar power.  So, it will be with AI and robotics: for good and evil.

In medicine, we have little doubt that AI will have a profound impact on the treatment and cure of heart disease, cancer and Alzheimer’s.  But who will have access to these scientific miracles?  Science won’t decide that; politicians (or The People) will decide that.  If we cure these diseases and give access to everyone, the population of old people will explode, with some predictable and some unpredictable consequences.  If we don’t give access to all, there will be blood on the streets and nationwide rioting, the end of civil society.  If there isn’t, we will deserve what we get.

In industry, computers have made us more “productive,” and cost us millions – if not tens of millions – of jobs, sometimes replacing them with other jobs.  AI will only speed up the transition toward a jobless world.  How many of us will survive lives of total leisure?

An older view held that Neanderthals vanished from the planet, not because Homo Sapiens had bigger brains, but because Homo Sapiens had more aggressive brains (my unsupported inference).  Genetic research has shown that many modern humans carry Neanderthal DNA.

Extrapolating from this past, when a superior (smarter, more aggressive) species meets an inferior species, the tendency is for the inferior species to … disappear.  Or to integrate.  Integration is more hopeful, but only if its alternative is human species extinction.

Bill Joy, a co-founder of Sun Microsystems, wrote a prescient article, in mid-2000 – 25 years ago! – for Wired magazine, entitled Why the Future Doesn’t Need Us.  Beyond thinking about the article’s title, I recommend that you read it.  Its message is the same as the messages that wise men (and women) have warned us about for millennia, those mentioned above.  Then you can ask any AI, “who else has written of AI's danger to humanity's survival?”, or should we "pull the plug"?

Before we quit the subject, we must talk about AI’s need for “data centers.”  In the old days, large corporations’ computers ran on its building’s electricity.  In the near future, a single low-moderate size data center will consume the same electricity as 80,000 homes.  Or 800,000,000 iPhones.  It doesn’t matter who will pay for this extra electricity, because YOU will pay for it in the end.  With electricity bills two and three times what they are now.  And you and I will pay for it in another way: with an explosion in pollution and global warming. 

Data centers are grossly inefficient.  Like killing a fly with a machine gun.  If we wanted to learn about Chilean dictator Augusto Pinochet, we might use our iPhone to Google his name or go directly to Wikipedia.  Virtually no iPhone energy but 5 watts of human energy.  Asking AI to do the same thing might cost 5 watts of data center energy (searching the entire Internet) and maybe 3 watts of human energy.  We seem to have bought into the idea that paying a machine to think for us and consuming less of our own human energy is a good idea.  So, we don’t practice thinking, and we gain weight at the same time.  What a grand bargain that is!  The fact that it takes so much electricity to solve simple problems is the very definition of inefficiency (not the machine’s fault) and proof that Artificial Intelligence is not very Intelligent.  And that we are even stupider for wanting it, its cost be damned!

Every champion of AI is aware of AI’s potential as an existential threat to our survival.  Nevertheless, they all push on.  “We need to beat China!  Better to die at the hands of OUR robots than China’s robots!”

What do I think?  It’s a race between our own assault on our planet and its life forms (global warming, the Sixth Extinction, etc.), and AI/robotics; “which will finish us first?”

The only possible road to survival is a well-informed democracy, one where the people choose its leaders wisely.  I can’t speak for Europe’s path, for China’s path, for India’s path, for Africa’s path; as for us, we are NOT traveling that path.

 

Add: Reference Hinton: Pull the Plug; list of players to slow down or stop; unintended (but predictable) consequences of data centers