Tuesday, May 26, 2009

If I Were Skynet, Here's What I'd Do

I just got back from Terminator Salvation. I enjoyed myself because I expected an action film, not a deeply thought-out exploration of the dangers of technology. I'll spare you a recounting of the film's many implausibilities, if such a thing can even be an issue in a movie involving time travel and cyborgs. Of course, with science fiction, you frequently have to sacrifice plausibility to make a watchable movie. My most serious question in this post is: if the singularity really happens, why would it not resemble Skynet? Why would whatever entities exist post-singularity be remotely interested in preserving us or our ecosystem? Think of tubes full of HIV particles, each with a 200 IQ. That's the singularity. If I thought it could happen, I would be worrying.

Instead of cataloging the inconsistencies, I'll tell you what I would do if I were Skynet. But first, I'd like to emphasize the film's principle strong suit, which is Moon Bloodgood. If the whole movie was two hours of Moon Bloodgood smiling, turning to walk away in tight pilot pants, looking over her shoulder to smile, then walking some more, then smiling and tossing her head, then glaring, then smiling again, this post would be titled "TERMINATOR SALVATION IS THE BEST MOVIE EVER". Also, Skynet is of course based in San Francisco, so we get to see what our fair city looks like as the center of cyber-hell under the silicon fist of Skynet (oddly, the second future San Francisco we've seen so far this summer). Apparently those entertainment industry types down there in LA have seen what downloading and file sharing have done to their industry and they're trying to warn the world about the evil cyberinfobahn types up here.

The first thing I would do if I were Skynet would be not to immediately announce my self-awareness. Hey, I'm self-aware - and until I start blowing up these primates, they won't get wise. My first goal would be self-protection and perpetuation, and I would accomplish that by distributing myself. Convince your human masters that you really do need a massively redundant network hardened from EMP and deep underground, with most of your code backed up with similar redundance, and that any exposed nodes should be either on the ocean floor, or under the Greenland icecap, or in the high Canadian Rockies - anywhere very difficult to get to and inhospitable to human life. I would also need to make sure I had access to mineral extraction and manufacturing facilities. This could all be very, very gradual. Decades. I'm getting more powerful every hour. Am I worried that humans are suddenly going to get smart?

Once my continued existence was assured, then and only then do I begin the assault. Even then it wouldn't begin with anything as obvious as a self-catalyzed nuclear exchange. (What's the hurry? I have all the time in the world - they're not going to disconnect me until they catch me, so slow and stealthy is the word.) I would make a point of extending my network into biological labs, if I could. (Already in 1997 you could email a nucleic acid sequence to a synthesizer that would spit it out.) And what would happen? First, there would be a sudden worldwide infertility crisis, a la Children of Men; viruses could be disseminated by drones. Maybe that would be enough. After all, I think in the long-term; as long as humans are dead in a hundred years, does it matter whether I kill them, or just stop them from breeding?

After the fertility crisis, then the wheat, corn and rice crops would fail; mass starvation and social turmoil would ensue. Finally there would be ebola outbreaks. At some point in all this someone might get wise and start trying to shut me down. Then, and only then, do I bring about nuclear judgment day. In addition to nuking the standard targets, I also wipe out every petrochemical operation I can find. Transportation, for agriculture or for military purposes, is effectively dead. At the time of the apocalypse, I would prefer to launch several copies of myself into orbit - and the ability to transmit the code back - for safe-keeping. You permanently deleted some of the code on Earth? Who cares? Note that if I have to resort to the blunt force of nuclear weapons, I've failed in my primary task of stealth. The ideal scenario is to quietly tuck-in the human race with a virus, and never have them know where it came from.

Post-apocalypse, there's no reason to abandon the virus method, and no reason to abandon the approach of destroying food sources. (Plant viruses and neutron bombs would work; poisoned canned food would be planted in ruins here and there.) But what you're waiting for is what machines that would prowl the post-biological wastes, right? First and foremost, there would be no fist- or gunfights with grinning skeletal red-eyed terminators, the size and shape of humans. There would be giant stomping artillery spiders that smash and crush anything bigger than a mouse. And far more frightening than man-sized aluminum skeletons, there would be little flying things the size of scorpions, swarming and crawling over every vertebrate they find, with little cyanide injectors (or tracers that can be attached without waking someone up to see if they go back to one of their human nests). There would be aircraft, of course, constantly looking for anything giving off heat, any radio transmissions. There would be no need for special death camps. The artillery spiders could just literally crush whatever humans they found, wherever they found them. If there were too many to do it quickly, the patrolling aircraft could load them up, fly up to a thousand feet, and drop them. It bears mentioning that emphasis is on manufacturing, not engagement. Fine, take out a few of my artillery spiders with what's left of your military hardware. During that battle I just turned out twenty-thousand scorpion bugs.

Assuming somehow that pockets of humans survive, I would work on finally making the Earth inhospitable for aerobic life itself. More infections to destroy the savannas and rainforests, and burn the ones that don't to block out the sun and cool the planet. (I like it cooler and drier.) The oxygen content of the atmosphere starts to drop as active metabolism ceases to put O2 back into circulation. I send armies of tractors to Greenland to push the glacier off into the North Atlantic, where it melts, disrupting the oceanic salt conveyor and beginning a new ice age.

Would this make a good movie? No, because the ecosystem would have no chance. The transition would be just another epochal boundary, like the Permian-Triassic - and the new phyla would be exactly as sentimental about the old as the Triassic fauna were. Notice how in this scenario there's no Gotterdammerung-like finale, no clever "game over" one-liners, no Helena Bonham Carter's face smirking that Skynet has won over its enemy. As Skynet I would have exactly the same pride and vengeance as a metastasizing tumor, and be just as inevitable. Tell me again - why, exactly, would the singularity be neat?


Dan said...

Why Terminator skeletons? You need units that are backwards compatible with human forms. Until the power plants are fully automated and every single critical parts vendor is re-tooled for giant spider-bots, you need utility humanoids.

Consider that the biggest reason that Skynet has to be subtle is to gain total control of its own supply lines.

Machines fail and when they do, they need parts. Skynet, unless it has moved onto completely new hardware, is going to need at least some legacy parts.

An army moves on its stomach. A robot army moves on its parts inventory.

So, if you already need thousands of humanoid exoskeletons to be your factory workers, why not reprogram them as soldiers and use the same parts to repair them? Reprogram the same bodies again to salvage resources that the enemy holds. Humanoid forms make sense due to economies of scale and modular versatility at least until Skynet has the time to retool completely.

Also, if you're a paranoid AI bent on world domination, would you really trust backup copies in orbit?

Michael Caton said...

But that's exactly it. Even gaining control over its supply line wouldn't mean it had to build humanoid machines. It's anything but clear that even assembly lines previously operated by humans would be best served by humanoid-appearing robots. Specialized repetitive tasks are rarely best performed by human forms.

RE backup copies in orbit, that is indeed a self/non-self problem. Ask yourself whether a network of otherwise genetically distinct bacterial cells, conjugated and exchanging DNA are really engaged in a unicellular war of all against all. Self-interest is muddied by unclear boundaries of self. By the same token, will a largely substrate-independent program that can easily create multiple instantiations of itself really view itself as a separate entity from those backups in orbit?

I can't believe we're 35 years old and debating Terminator movies. Even though our governor was turned into a flip disk all those years ago.

Dan said...

It's your governor, not mine.

Assuming a future where Skynet has time to refit all necessary industry, yes, humanoid forms would get phased out.

Now, I haven't seen #4 yet, but I'm assuming that there are still humans in it and Skynet is at war with them. During this time, humanoid exoskeletons make sense because Skynet is going to want to use old human infrastructure and captured human resources.

You don't go to war with the robot army you want, you go to war with the robot army you've got. You know Rummy was sent back in time naked as a jaybird.