Robopocalypse

Rating: 3 stars
Length: Average (347 pages in trade paperback)
Publication: June 7, 2011 from Doubleday
Premise: Humans have been happily depending on robots for decades, but an artificial intelligence named Archos is determined to change the course of the world. Things start going wrong with small malfunctions, but when Zero Hour comes, no one is safe.
Warnings: fairly disturbing machine gore, nonconsensual body mutilation/modification
Recommendation: If you want to read about the robot apocalypse, go forth and read: this does a good job with both the machines and how the people recover from depending on them. It can be a bit uneven no the pacing and style counts, though, so I wouldn't recommend buying it new unless robots are really your thing. 

What makes this one genuinely unnerving:

The robots themselves form the natural centerpiece of the book, and they seem utterly convincing in most places; learning after I finished the book that Daniel H. Wilson has a PhD in robotics explained a lot. Each type of robot feels detailed and designed for a clear purpose. After the war has been going for a few years, that purpose is chosen by machines rather than people, and these later-stage robots feel more alien because human minds had no part in the creation process. Whether they're simple machines designed to seek out and destroy human body heat or complex humanoid soldiers tailored for rough terrain, they feel at home in their surroundings. The robots are methodical because they're programmed to be, and that ironclad background makes things like hints of emotion or a desire for free will stand out all the more. 

Robotic nature at its base is chilling from the start-- the story opens on a battered team of warriors digging out a cube full of data in the aftermath of the war. Archos, the artificial intelligence behind this war, has kept an incredible amount of data, some of it focused on the ever-defiant humans who it apparently couldn't help but see as heroes. This framing device works in part because it allows for very little certainty about who lives and who dies....and even when people do die, they come off as noble but necessary sacrifices along the way, as good people who died because that was what the struggle required of humanity. The story shifts between personal accounts, video transcripts, conversation logs, and even short snippets from Archos itself, and that opens up many avenues of telling along with a rich cast of characters. We're introduced to many of them early in the journey, when robots are just starting to malfunction and change-- simple dolls wake up, pacifying war machines go berserk on civilians, and even simple housekeeping robots can suddenly go for the nearest human. These slices of life from various witnesses make for a great sense of creeping menace-- watching them go berserk was a completely foreign experience, but every person who saw one of the attacks knew on some level that it wasn't an isolated incident. It's genuinely scary to watch a toy baby doll start threatening a girl's younger brother or a simple machine crush a man's face, and these early sections are where the book shines.

Many post-apocalyptic books focus on the struggle to find food and the fate of the warriors who end up controlling society, but this one zooms out to show snippets of the labor camps or the scientists trying to find a way to shut the artificial intelligence down. Some of the best of these revolve around urban demolition: ordinary people with no military training sometimes have the best idea. An ordinary construction worker notices that automated cars and domestic robots are killing all the humans they can find, and that many surviviors are fleeing to the country where the robots will have trouble with rough terrain. His solution is to blow up buildings and create rough terrain by destroying New York City, much of which he helped build; this moment, more than almost any other, calls back to the early warning that artificial intelligence ought to fear humans because they will do anything to survive. Even when they're mutilated and tortured by having parts of themselves cut off and replaced with machine parts, they ty above all else to unite with each other. Humans, interestingly enough, end up having to take on some machine characteristics voluntarily as well to survive: pausing for too long trying to save a friend from tiny robots will only kill the attempted savior as well, so humans who want to live often have to cut each other off from safety and even from simple connection. Many books focused on the struggle of humans against an external other choose to make essential human nature the key to victory, but Wilson creates a more nuanced portrait in which human and machine strengths are both necessary even if joining forces seems abhorrent.

The red pen:

Given how strong the science and robotics felt so much of the time, it was disappointing to see some aspects slump past the halfway mark of the book. Takeo Nomura, for example, starts out as a fascinating character: he's an absolute genius with machines and has fallen in love with his own female robot, who is designed to look like an old woman of about his old age. It's a sad-but-sweet dynamic, and he fixates on repairing her when she hurts him-- he loves her, but he's too intelligent to turn her on without solving the problem of what's been controlling the machines in the first place. When he finally does figure things out to some extent, the solution is hopelessly muddled and comes across more as a near-magical intervention than the clever science that it is, and that becomes something of a running problem. Archos is designing and building robots according to its own designs, and humans sometimes manage to extract bits of machinery and learn from them, but this process is glossed over after the more detailed explanations of how humans intended robots to work in the early sections. Even when one girl gains the ability to see machines, the view of how humans study robots when those robots become threats is clouded at best and absent at worst. This approach could have worked in an action-heavy novel, but Robopocalypse relies more on conceptual discussions and even allows some robots to gain freedom and independent thought; not fully addressing how that even happened weakens the structure of the book as a whole. 

One of the largest problems is the confusion over Archos's motivation-- we know that the AI went through fourteen iterations before managing to survive, escape its confinement, and plot the overthrow of humanity. The problem is that after that, its general direction is far less clear. We're shown that Archos wants to study the natural world and try to preserve it because it's bothered by how close humanity has come to destroying so many ecosystems; one scene with humans finding an enormous robot-run habitat opens up all sorts of interesting questions about why an artificial life is so intrigued with natural ones. Other conversations demonstrate that the AI doesn't want to entirely destroy human life, only open up space for its own people to live alongside biological organisms....but humans are dying off at an impressive rate, and many of the ones still living are being forced into work camps to help assemble more robots or to be experimental subjects. This gives the war an added darkness, but it means that even in the end it's hard to be entirely sure what Archos hoped to gain (or why seemingly all of the robots with free will choose to side with a race that has the most motivation to destroy them). The full reasoning of an AI that sees a microsecond as an impossibly long time is of course going to be impossible to comprehend, but there's not even much a hint to grasp here. 

The human characters ought to be as vivid as the robots, and they definitely are at first, but many of them just grow flatter as the story progresses. Lurker, a London hacker who learns something about Archos's existence before the rest of the world, gets better over the course of one stark scene and then shines in his next appearance, but the the narrator collecting all of these stories from the data cube is in some of the stories, and the heavy reflection about heroism tends to intrude on scenes that would pack enough of a punch on their own. When the narrator pauses to reflect that someone related to one character is halfway to being a hero already and that character has helped create an alliance, that's fair....except that several other major characters have sacrificed their lives to save their families or humanity as a whole and don't get the same accolades or level of personal detail. Everyone gets some sort of representation in the midst of action, but the jarring contrast of what we're told about these people versus what's shown that they do on the page disrupts the flow of the story. This evolved from a large-scale ensemble cast into a central core of characters, half of whom have almost no definitely personality, and the book suffered for it: it would have been better had it covered more ground and perhaps even some one-off characters with no relatives or friends being directly helped. This was meant to be humanity's fight, but it ended up being the fight of a small group with accent passages from some of the robots. Such a narrow focus damaged the emotional arc of the last few chapters, and that tipped this firmly into the category of "I wanted to like it more than I actually did."

The verdict: Robopocalypse absolutely shines in the technical aspects; I've rarely read robots and engineers that feel so convincing in their roles. Unfortunately, some of the science gets vague at roughly the time the characterization narrows and flattens, and that makes the first half of this book far better than the rest. I might check out some of Daniel H. Wilson's books in the future, but it doesn't feel urgent.

Prospects: There is no sequel to Robopocalypse and one would feel shoehorned in, but Wilson has written another futuristic novel called Amped, which explores the concept of enhanced humans.

Enjoyed this? Try:
~Syne Mitchell's Technogenesis deals with the question of human dependence on technology and artificial intelligence in an entirely different way, but the two books both excel at technical detail and sharp imagery.