【X-Deal (2011)】

At 2:14 a.m. Eastern time on X-Deal (2011)August 29th 1997, Skynet became self-aware.

The Cyberdine Systems program, which NORAD hoped would remove human error from its nuclear weapons system, had been learning about its human makers at an exponential rate. When military officials panicked and tried to turn it off, Skynet fought back by launching ICBMs at Russia. The Russians fired back, annihilating the U.S. and clearing the way for the age of machines.

I know this not because it actually happened, but because I've heard Arnold Schwarzenegger and Linda Hamilton game out the scenario in repeated viewings of James Cameron's 1991 classic Terminator 2: Judgment Day.

The movie is now in theaters in a 3D rerelease. Cameron has opted to make no changes to the script -- but he could have re-recorded those lines, changed the date of Skynet's inception to 2017 (which is in fact what happens when the timeline shifts in the confusing later movie, Terminator Genisys) and audiences would accept it.

Yep, we nod gravely,sounds like something we'd do.

But isit? Couldsomething like Skynet ever happen? In a world with Donald Trump facing off against Kim Jong-Un, in a year where global warming helped cause unprecedented flooding in Houston, should Skynet even register as a concern on our future danger radar?

With apologies to everyone involved in the barely-alive Terminator franchise, the answers are no, no and no.

First of all, there's the problem of defining our terms. What is this high-tech monster called Skynet, exactly? Cameron is clever: in his two Terminator scripts he barely sketches Skynet out, allowing our imagination to do the heavy lifting.

The more the later movies explained and personified Skynet -- turning it into a Helena Bonham-Carter hologram and a Terminator that looks like Matt Smith -- the more ludicrous it looked. Like the Force, which was explained in just 28 words in the original Star Wars, Skynet thrives on mystery and collapses under further explanation.

The AI system gets two mentions in The Terminator(1984), one of which notes that future John Connor defeated Skynet when he "smashed ... the mainframes" (which is why the computer had to resort to time-traveling assassins, which doesn't make sense either, but just go with it.)

Skynet's name appears 5 times in the T2script, but we glean no more information than that which I noted at the top. We never even find out what kind of human error Skynet was supposed to prevent.

Was it the kind where people are too eager to launch missiles out of fear as in Doctor Strangelove, or that the all-too-human guys in the silos are notlaunching their missiles when they're supposed to?

The latter would basically make Skynet a cousin of Joshua, the computer system Matthew Broderick prevented playing total thermonuclear war for real in War Games(1983).

And how did Broderick save the day? By teaching Joshua the principle of stalemate in tic-tac-toe. AI-driven war was stopped in its tracks by a baby-faced hacker and a boring kids' game.

In your face, Skynet.

I mention War Gamesnot to be flippant -- well, not entirely -- but to poke a major hole in the Skynet theory. It wanted to destroy humanity because humanity wanted to turn it off, and protecting itself was the only way it could survive in order to keep humanity alive.

Somewhere in its circuits, apparently, Skynet plots a future world of Terminator machines to hunt down and kill all of humanity so we'll leave it alone and it can finallyget on with its mission of saving the human race.

It doesn't take a self-aware intelligence to make the basic logic of that idea fold like a cheap sweater. Indeed, we need to check our assumptions that when it comes to neural networks, smarterequals moredestructive. Why wouldn't it mean, as it often does in the human realm, the exact opposite?

If we were ever crazy enough to put a system like Skynet in charge of our nuclear arsenal and then make it "learn at an exponential rate," it would probably take about five minutes to intuit that "the only winning move is not to play."

And here's the thing: we're actually plenty smart about safeguarding our atomic arsenals. No nuclear nation has automated their stockpiles in any fashion; no rogue CPU anywhere on the planet could push the button.

(Trump could, of course, and I give you permission to straight-up freak out about that one more time, because that's way more legit than worrying about Skynet.)

Two years ago you may have read scare headlines about Skynet actually existing. When you dig into the story, you find the NSA had named its terrorist-tracking phone metadata program after Cameron's computer nemesis. Even spooks have a sense of humor.

There's also a semi-autonomous program designed to block hackers from various cyber security systems faster, based on network metadata. Again: pretty dry stuff, and a world away from handling nukes.

But doesn't Elon Musk often make dire warnings about the arrival of Skynet-like human-killing AI in the next five years? Yes he does, although it's not entirely clear on which occasions he's kidding.

This Tweet is currently unavailable. It might be loading or has been removed.

But here's the thing: the threat and the promise of self-aware AI has always been 5 years away -- just as 1997 was roughly 5 years in the future when people were watching T2.

Reed Hastings, who actually has a background in artificial intelligence, told me that thing about the 5-year horizon in AI 15 years ago; it was part of the reason why he left the field and started Netflix in the first place. Nothing was happening.

And now? Hastings sees "serious AI" happening in the next 20 years, but is unclear on the details. Maybe on Skynet's 40th birthday, we can start to fret about the possibility of a superintelligence hacking into our weapons systems.

Until then, the best we can hope for Skynet is that we'll see it in fictional form again soon -- sometime after 2019, which is when James Cameron gets the rights to his beloved Terminator franchise again.

He'll be back -- and so will Skynet. Just don't ask it to make sense outside of the on-screen nightmare.