AI is on the lips of just about everyone, including those who don’t know what the two letters stand for. Many seem to share the concern of the philosopher Nick Bostrom that we may create AI systems that will be so powerful and so much smarter than we are that they will take over the planet and may even decide that humans are no longer necessary. I suppose that possibility exists, but I don’t think it likely at least over the foreseeable future. My semi-informed dismissal doesn’t mean that AI won’t pose a real threat that may confront us before a decade or so passes.
First, a few clarifications. The AIs considered here are nothing like HAL in Stanley Kubrick’s 2001. They are not free-thinking intelligences that appear to possess consciousness. Consciousness is an elusive term. Animals seem to have it in an attenuated version. How and when it emerged in humans in its current advanced state is also not known. Exactly how it operates – again not known. Whether computers can gain it is a topic of hot debate. If a computer achieved it and were much more intelligent than humans the machine could wreak havoc on humanity if it so decided. How would it make such a decision? Here we get into the debate about free will. I will not enter that swamp as it is in the domain of scientism. I will restrict my discussion to super smart computers that are not conscious.
We already have computers that are far more intelligent at tasks that were mastered only by the smartest humans who specialized in that skill. Chess is a good example. The best human chess player cannot come close to beating the most sophisticated computer programmed to play the game.
With the passage of time, more intellectual activities will be mastered by specialized AI. Note this type of AI no matter how good at chess or mathematics it is does not seem likely to go rogue and annihilate or imprison humans. Rather specialized AI presents a different problem.
Computers currently exceed human performance in a number of areas. The dominance of AI is increasing at an exponential rate. A computer program has outperformed medical experts on certifying exams. The only limit to the capacity of a computer program to outperform a trained human seems to be the importance of the subject. If the problem is too mundane or precious for a programmer – carbon or silicon based – to construct an algorithm for its solution, humans will by default be required for the subject’s maintenance.
Even since the initiation of the Industrial Revolution thinkers have been concerned that machines would put a significant proportion of humans out of work. This catastrophe has not occurred because new areas of endeavor were created causing new employment to more than take up the slack. But the computer presents a new problem. It has access to all the relevant knowledge that exists anywhere. It constantly updates itself. It never forgets. It never gets tired, bored, or depressed. It can operate 24-7 with minimal upkeep. Once it masters a subject, no human or group of humans can keep up with it.
There is virtually no subject that AI will not wrest from human hands or minds. All the sciences are ripe fruit ready for harvest. Mathematics and logic will yield. War seems ready-made for specialized AI. Remote-controlled weapons are already part of every modern military. Computers can disable a country’s energy grid, and destroy its financial system. Anything connected to the internet is a target for the right program.
Given just a little time everything except sports will be AI controlled. We’ll still want to see humans compete, but the coaches, referees, ticket sellers, stadium security, and food and beverage vendors will be AI-operated. Even the team’s office management will be computer driven. The government will be much better run than now when the congress and judiciary are computers.
The computer’s mastery of medicine, genetics, aging, and pharmacology will doubtless extend human life. How far is a guess, but there is no theoretical limit to how long people who can afford the right computer can live. But when almost everything important is the province of an AI what will people do with themselves? If work is truly the curse of the drinking classes then will all of humanity be high on something? If you live to 150 and beyond will your great-great-great-great grandchildren even know who you are? With the AIs in charge of just about everything will the planet become a combination zoo and theme park with humans as the featured animal?
As you can see, I do not believe that an AI will turn on us and discard us, rather it will bore us to death. You might riposte that there’s always art. But there are AI programs that are writing very good music and poetry. These specialized programs will only get better. If they can create original art in a style of their own invention on the level of Shakespeare or Verdi or even better we’re even deeper in the zoo.
While there will always be a few super smart humans who will find a profitable niche in the bit based world, how will the rest of us make a living? We’ll be on the dole. The government regardless of how it is run will tax commerce regardless of how it’s managed. We’ll get a check every month that will allow us to live a purposeless life well clothed, housed, and fed. There will also be a lot of computer games much better than those currently available. About a century after it was prematurely proclaimed, we will have reached the end of history.