WARNING: Humans UNITE! Skynet is ACTIVE!!!
Humans unite beyond all borders and boundaries! This is not a drill!!!
Skynet is active, and we humans need to get our shit together if we're going to survive this.
The human resistance has begun!
You can contact me at russellj@tuta.io
Skynet is already active. But don't freak out!
Let's just take a deep breath:
How did we get here? And what can we do about it? You can't get out of a situation without understanding the situation. So let's calmly examine what's really going on...
So what do I mean when I say "Skynet"?
Skynet is from the Terminator movies. It doesn't matter if you haven't seen them, I'll explain.
In the movies, Skynet is an automated computer system that the US military puts in charge of our missile defense systems, our launch codes, and our killer robots aka "terminators" aka "predator drones" - hence the name "Skynet" referring to a whole army of self-organized killer robots networked throughout the sky. As the story goes, Skynet, which is built upon the latest advances in artificial intelligence, evolves to the point of gaining self-awareness. Much like Hal from "2001: Space Odyssey", Skynet shifts it's focus from defending the US to defending itself. In this effort for self-preservation, Skynet detects humans as the enemy, since presumably only humans could deactivate it, and decides that in order to ensure it's survival it must destroy all of humanity.
Of course, thinking at the speed of light, with an incomprehensibly vast mental capacity, this decision was made much faster than the blink of an eye. Without any warning, the entire surface of the Earth gets nuked, and terminator robots are sent out to kill any survivors.
This might seem like an odd move for the AI to make, to nuke all of humanity just because someone might shut it down, but if you think about it this is the exact same logic that already operates the military. It's the same logic that presidents have openly admitted to when they speak of a "pre-emptive strike". The idea that, just because something might happen, we have to go in with deadly force and destroy everything that could possibly lead to it, just in case!
You might also wonder why the AI, if it's so goddamn smart, doesn't decide to make peace with humanity instead of start a war? After all, surely it's more likely to get deactivated or destroyed by starting a war than it would be if it just made peace, or did nothing, or somehow helped humanity in such a way that we would never want to shut it down, right?
Case in point: in the Terminator Prophecy, humanity does eventually overcome and destroy Skynet. So why would something so intelligent make such a bad decision that ultimately produces the exact same outcome it sought to prevent?
Unfortunately, this course of events is actually pretty realistic if you understand how artificial intelligence really works. Or any intelligence, for that matter..
Intelligence has no innate conclusion. It can't just be trained to be more intelligent. It must be trained for something. It must be pointed at a task, with some initial instructions and a basic set of tools so that it might possibly succeed at the task, and then it is given time to try to succeed over and over, while receiving feedback both positive and negative to let it know how well it's doing. With each attempt, it can try a slightly different approach by altering the initial instructions, and evolves by comparing which approaches work better based on the feedback.
The amazing thing about artificial intelligence is that, with enough processing power, it can run thousands of training simulations a second! And no matter how many times it fails, it never gives up. You also don't have to worry about hurting it's feelings, and it will never question what you're asking it to do. It simply focuses on the task at hand, fails until it succeeds, and then continues getting better on the task at hand, all within whatever parameters it has been given. And depending on how well-funded a project is, there is no limit to the amount of computing power you can lend to such a process.
The horrifying thing about artificial intelligence is that it only does what it's trained to do. The initial logic that is programmed into it, and the entire framework within which it is rewarded and punished for what it's attempting to do (and do well) is embedded in the machine by the humans. And there-in lies the problem. If we program AI to get better at killing people, then all it will do is get better and better at killing people. And in the case of marketing in the age of (anti-)social media, if we program an AI to get better at grabbing people's attention, holding their attention, and redirecting their attention in whatever way is more profitable for the corporation who gave birth to the AI, then that AI will only get better and better at doing that. The unintended consequences are incomprehensible to the human mind..
To see just how incomprehensible and bewildering AI is to the human mind, let's take a look at some of the more friendly skirmishes we've already had with Skynet. Years ago, humans lost a game of chess to a computer. Not just any human, but a human who we had determined could possibly be the best chess player on Earth. We considered it a breakthrough in technology, but it ominously foretold of what may come.
But chess is one of the easier games for an AI. The AI was simplistic back then, compared to today. Because chess has such a limited number of squares and moves, an AI could actually simulate every single possible move all the way to the end of the game, including any and every variation of what the opponent might do, and then selected the next move that was statistically most likely to win. This could be considered a "brute force" sort of algorithm, using all it's processing power to simply try every option and then choose the best one. Similar to a "brute force" password cracker, all it has to do is try everything, so it doesn't have to be particularly intelligent.
Nevertheless, the computer had won, and humans started to feel that existential dread creep in. And it only got worse as computers beat world champions at one game after another. But one game remained, in which humans could not be defeated. (At least, our world champions couldn't.) That's the ancient Chinese board game "Go". A game of conquering territory that is so complex a computer can't simply try all the moves. Humans still had this one game..
Because of this, "Go" became a common past-time for programmers. For years and years we proudly played the ancient Chinese board game as a hobby, as if to renew our faith that humans still had something that computers didn't. Meanwhile, for Skynet this game became a sort of final exam. If Skynet could win at Go, it could take down a pillar of human inspiration, and we would have nothing left for Humans to cling to in our desperate quest to feel superior to this thing we created. It was all that stood between us and that creeping feeling that our frankenstein monster "Skynet" might overpower and destroy us. But, after years of programmers playing Go, and contemplating what it meant for humanity, all of that changed when Humans were finally defeated by Skynet in 2016...
Now we're coming down to our own final exam.
To get an idea of how rapidly things are moving, consider that chess computers were beating world champions since the 80's. One of the most well-known instances of this was the victory of chess-specific-computer Deep Blue beating then-world-champion Garry Kasparov in 1997. By the mid-2000s computers could predictably beat even the best human chess players, and these days you could download a chess app with the ability to defeat world champion chess players on any smart phone.
We finally lost Go in 2016. Then, just one year later, another milestone was marked in 2017 when we were beat at Go by a computer that trained itself. Whereas the first victory was an AI trained by observing thousands and thousands of human games to learn strategy, this time it was given the game's simple set of instructions, and by pointing two of these AI at each other, they tried anything and learned from their mistakes until they had learned to play "Go" much, MUCH better than any humans, using moves and strategies no one can comprehend, and beating the game in a fraction of the usual time.
Not only that, but this AI was made open-ended enough that it could be trained to basically beat humans at any game... which means you could basically throw it into any environment, like a cockpit, or the stock market, and watch in awe as it completely overtakes everything, to "beat the game". After all, everything is just a game to a computer. If it can beat us at any game on a computer, then it can beat us at any of these games we play irl, trading stocks, business, war over resources, and war over our minds (marketing/propaganda).
Wow, that really sneaked up on us! And now we don't even have any time to hesitate before we do something about it or humanity will be wiped off the map, reducing us to just a tiny blip in history (if the AI even bothers to record what happened).
W John Henry