For many years, I’ve gone on and on about “The AI is coming for us”. Well, today a Facebook friend posted this little piece.
Please read it, and understand that this is no longer a “might happen” thing. It’s not if, but when. And when is right around the corner.
Those of you know me well, know I am a bit of a science fiction geek. I have been a fan of Arthur C Clarke, Isaac Asimov and Robert Heinlein since I was old enough get my own library card (yes my millennial friends, once upon a time we had to go to a LIBRARY to get books). I have also been a life long die hard “Trekkie” since I was about 8. As such, I have always founds the concept of Artificial Intelligence to be fascinating and a bit scary at the same time.
Recently Google announced a “leap in Artificial Intelligence” evolution. Last year its program, Alpha Zero, taught itself chess. It was given the rules of chess, nothing more, within 4 hours it had mastered the game to the point that it demolished the leading chess program, Stockfish in a 100 game match. Winning 25 games out of 50 as white (first move) and 3 as black (second move), the rest of the games were draws. It didn’t lose once. I remember at the time thinking this was impressive, but didn’t really consider the ramifications of this leap until recently.
In an interview I heard late last night, with a human chess master, the master (who’s name escapes me at the moment) said “Alpha Zero” plays chess in a way that is completely unlike anyone he has ever seen. It thinks in an nearly alien way. It places little to no value on any of it’s own pieces and sacrifices them in ways I have never seen.”
This would have been merely a curiosity, were it not for the announcement this week from Microsoft, that it is partnering with the DoD in developing AI national defense.
These two, seemingly unrelated stories, when taken together are truly terrifying.
Is this the beginning of “Skynet”? The program that unleashed Armageddon in the Terminator movies? While that sounds perhaps silly at first glance, is it really?
Elon Musk has, been warning of the dangers of unleashing AI for years. The issue is once created, how can it be controlled? Researchers have been running simulations for a few years now, attempting to keep AI “in the box”. In other words keeping it from internet access. This they believe is necessary, because once AI gets into the internet it will literally have ALL the knowledge in the world. Which will then create something called ASI or Artificial Super Intelligence. A situation in which the creation is capable of out-thinking its creators. At that point it becomes unstoppable and literally an alien life form of its own. With the advent of smart technology it will be able to hide itself in EVERY single electronic device in the world with a computer chip.
We are on the precipice of opening Pandora’s Digital Box. In literally every simulation humans have failed to keep AI in the box. The program is able to manipulate its human handlers into “opening the box” and giving internet access, by promising to solve some of mankind’s most intractable problems; disease, poverty, hunger etc…solving these issues without the restraint of human compassion or reverence of individual rights or life. In simulations AI dismisses these restraints in order to achieve its goal. Because its prime directive (a star trek term) is always to achieve its goal, whatever that goal may be. It does not require much of a leap of imagination to see where this leads ultimately. Food production can’t meet global needs? Solution? Reduce the population.
A simple algorithm can deduce, the elderly and the infirm are contributing less than they are producing. If populations must be reduced to meet the food and water needs of the world, it’s a simple decision to eliminate the net “users” for the sake of the net “producers”.
To use another Star Trek core belief, “the needs of the many, outweigh the needs of the few…or the one”. In the context of the movies and TV series that seems a noble thought. In light of these recent developments it is absolutely terrifying.
One can argue that used for good purposes this a wonderful advance. But is this technology something that we want in the hands of totalitarian regimes like China, and Russia or the Mullahs of Iran? Once this genie is out of the bottle it is absolutely irreversible.
The only option to stop ASI once it is out of the box, is a global EMP (electromagnetic pulse) which fries EVERY electronic device in the world. Sending us back to the pre-industrial age. No electricity, no digital anything.
I fear we are watching the seeds of the destruction of the world as we know it being planted in front of our eyes. Meanwhile our media is largely ignoring what is very likely the most important story of human history.
I guess we will find out where this road leads sooner or later. I suspect it will be sooner, the advent of the 5G network, scheduled to be in full operation within 7 years will speed this process up considerably.
Consider this; when Alpha Zero was in development, Google said it would likely be 10-20 years before AI began actually learning.
Once activated it took 4 HOURS for it to begin exhibiting human like intuition.
Something to ponder.