I for one welcome our new robot overlords

Fantastic Adventures 1941 Jan cover A common theme in science fiction is the idea that robots will take over our world. This doesn't seem very likely, but suppose it were to happen. What would the first steps look like? How would we first begin to notice that it was happening?

In most if not all of these visions of the future, the machines are as intelligent as humans. Sometimes it's one intelligence, sometimes it's a robot society, sometimes it incorporates human intelligence in one form or another. But I don't know that I've seen a vision of the future in which machines are essentially in control, but without any human-like intelligence anywhere. If that sounds weird, well, maybe it is, but remember that smart as individual humans may sometimes be, en masse we are not noticeably more intelligent than, say, raccoons, apples, or escherischia coli. Looking at things in evolutionary terms, it's not at all clear that we must necessarily be "in control" in any meaningful sense. One could even argue that machines have already enlisted us to build more machines, to our own detriment. From sweatshops that force humans into machinelike repetitive labour, to mining that turns habitable land into moonscapes in order to build more machines, to factories that pour out long-lasting poisons while making machines, it sometimes seems like the machines are already running our lives. Of course, these are in some sense human systems, though they are not necessarily under the control of any individual human.

But what about more traditional machine domination? Well, machines are already in control of huge segments of our economy: algorithmic trading accounts for much of the stock market behaviour (73% of the volume of US trades, for example). These algorithms are each written by humans (though there are certainly many layers of abstraction underlying the code, so that it's possible that no single human understands a given algorithm), but their details are closely-held secrets, and since they are written by a number of competing interests, it's fair to say that their interactions are neither intended nor even understood by any human. If this unsettles you, join the club. But the efforts of various humans to change this have been stymied, possibly by the influx of large amounts of money of unclear origin into politics.

That's just the market, though, and you can argue that humans have been driven by uncontrolled market forces since long before the Dutch tulip mania. What about the classic robots-on-a-rampage? Well, many or even most of you reading this have a machine in front of you with a camera and microphone which can watch you, often without you noticing. It's been years since most people could be confident that their computer wasn't doing anything they didn't expect or didn't want. For the most part, though, our computers have rather limited ways of affecting our physical environment. They can order unwanted items, I suppose, but otherwise they're mostly limited to making noises and displaying things.

My Prusa Mendel RepRap (with RepRapped Filament Spool) Fortunately for the science-fictionness of the present, more and more people now have robots in their home. They vacuum, they clean pools, they — unnervingly — mow lawns (an autonomous machine wandering freely out of doors and equipped with a sharp rotating blade; what could go wrong?). There are an increasing number of sessile robots that can make a tremendous variety of objects, including copies of themselves.

Flickr - The U.S. Army - iRobot PackBot Those are all modest household robots, though. We could throw them out, if we had to, and none of them (with the possible exception of the lawn mower) is particularly alarming. Fortunately, again, for the we-are-living-in-the-future, there are an increasing number of military robots. Some of them are for surveillance (there are even civilian flying surveillance robots), and some for things like bomb disposal. But there are also more alarming robots out there. The United States is waging a war in (and to some extent, against) Pakistan using primarily flying killer robots. They use them in combat and for assassinations (including of American citizens). Still, they're only killing people on the command of a human operator, so that shouldn't be too alarming. Except that the command centre — a building full of pasty-faced military nerds who fly these robots — has been infected by a computer virus. In fact, they've had the same virus for three years now, and they haven't been able to get rid of it in spite of heroic efforts. They don't know quite what the virus is supposed to do, but if it is intended to somehow hijack the robots, it wouldn't be the first virus intended to do such a thing.

Predator and Hellfire
In summary: machines control much of our economy, much of which is devoted to building more machines. We have machines watching and listening, and increasingly, free to move about or make more machines, in our homes. We also have a large number of armed military robots killing people. None of these is reliably under human control. So what would a machine takeover look like? And would we notice?

ETA: Not only are the killer robots virus-infested, the US military doesn't necessarily know what hardware is inside them. -A


mvc said...

wow, with a virus on those flying killer robots, none of us are safe. thanks for scaring the hell out of me.

Anne Archibald said...

Well, so far, the only observed effect is to log all keystrokes. Which sort of suggests a more Wikileaks kind of break-in: not exactly benign, but more interested in embarrassment than damage. The fact remains, though, that these computer-controlled killing machines can be infected by viruses and there's nothing their handlers can do about it. So how long until a hostile faction gets a nasty one through?