As you know, one of the major themes of this blog revolves around the effect that design of technology and control of information has on society. I’m a big proponent of decentralization in computer networks and run a company that’s been working to achieve it, but at the same time I’m always interested in understanding the trade-offs between centralization and decentralization.
When it comes to drones, there seems to be an unprecedented combination. For the first time, autonomous robots with increasing intelligence will have ready access to public spaces. Self-driving cars are another example, but they are expensive to produce. But because of how cheap it will be to produce programmable drones, this can create a serious problem for society.
Our social systems aren’t designed to cope with AI, and many rely on inefficiency on the part of current actors, or “security through obscurity“. When a driver does a hit-and-run, we at least have a lead (the car, the human) and can set up some sort of retaliatory mechanism as a deterrent. Terrorism is largely a problem of technology (e.g. swords vs the gunpowder plot vs 9/11). When drones produced by no-name manufacturers are programmed and dispatched by anonymous individuals, they can wreak all kinds of havoc, without any repercussions. It takes 1 out of 10,000 rogue actors (whether nutcases, terror cells, other countries, etc.) to do something. If Amazon has its way and drones are legalized, whatever framework we adopt, what is to prevent a drone from taking off and doing damage anonymously? Even with zero malicious intent, an increase in heavy drone traffic raises chances of death by impact. But having anonymous drones equipped with bombs or things like this seems like a real danger. It seems to be an arms race that’s coming up faster than the AI arms race that lots of people are concerned about.
And yet there is no way to stop the progress of technology. This is one place where decentralization may not be such a good thing. What can we do?