Monday, December 26, 2011

Daemon and Freedom, Daniel Suarez

Daniel Suarez's Daemon and Freedom tell a single story about an artificial intelligence, (the "Daemon" of the title) created by a game designer, that becomes extremely powerful in the real economy through violence and by hacking software systems throughout the world. In the end its creator offers to relinquish that power, but by then agents of various governments have shown how ruthless they can be in trying to defeat it, and a representative of the people being protected says it should continue running things. The story is occasionally violent, and includes several explicit misogynistic and sadistic scenes that I had trouble reading. The rest of the story is interesting enough on a couple of levels that I kept going.

It's a fast-paced story involving MMORPGs, augmented reality systems, and an AI created by a wealthy software gaming entrepreneur. The story (and its characters) presumes that the world's economy is controlled by a few powerful, wealthy, and unscrupulous tycoons who pull the strings of the big governments and are getting out of control. The AI makes a preemptive attack and attempts to take over in order to prevent an eventuality which is never really made clear. Instead, we see the US and other governments' secret agencies and militaries attempt to strike back at a system they don't understand, and which is decentralized and has infiltrated most of the world's computer systems. This allows it to watch its opponents and strike back in completely unexpected ways. It also invests heavily in a fleet of autonomous weaponized vehicles that it deploys very effectively.

The depiction of disaffected people from many backgrounds being recruited into a hidden network that deploys them on unexplained tasks that they willingly take on is disquieting. It's clear that many people who join have made thinly justified assumptions that the network's objectives are consistent with their values, but others do it because they're desperate for a job or a sense of belonging, and are willing to ignore their moral qualms about what they're doing--building, testing, delivering obviously dangerous weapons, or worse. It's plausible that many people would go along with an AI that could get this far, and the thing that keeps this from being possible now is that neither AI nor conventional intelligence augmented with modern tools give anyone enough power to pull it off. Who knows how soon that will change, though.

I thought the violence was over the top, and not really necessary to the story. I disagree with the world view that predominates here and says that malevolent actors control signifiant aspects of our economy and will go to extreme lengths to maintain their influence. I found the depictions of technology (other than the killing machines) reasonably plausible, and won't be surprised to see augmented reality, consensual shared overlays, and large-scale real-time cooperation. I expect the cooperation to be much less centrally controlled, and much more either spontaneous, or polycentric. And my strong expectations about AI are that it will arise gradually and in many places. It is very unlikely that one person (much less likely even than one independent group) will make a breakthrough that will enable them to take over. This particular question has been much debated, and there isn't clear agreement on how it will play out, but I firmly agree with Robin Hanson's position that a "hard takeoff" under the control of a single entity isn't a likely scenario. That one person could pull it off with help from at most two colleagues (as in the story) is completely unbelievable.