Unpopular review time. Despite the beautiful cover and glowing reviews, Sea of Rust was decidedly anticlimactic.
Set in a post-apocalyptic future where humanity has been eliminated, it is centered on Brittle, an autonomous artificial intelligence, and her effort to survive. Brittle is one of the last Caregiver models of AI, originally created to care for humans. When out in the Sea of Rust, scavenging parts from dying AIs, she is attacked by another AI, setting in motion a complex chain of events.
Most of the story is told in first person, although when Brittle relates the general events that led to the evolution of artificial intelligence and the war with humanity, she often takes a third-person historian approach. I found the story generally well told and interesting, although there are some occasional time shifts when I found myself thinking the transition was a bit abrupt. Most of the time, the past eithers provides perspective on Brittle’s existence or that of the AIs.
The problem for me is perhaps that I read a bit too much fantasy and sci-fi, and the characterization never really coalesced for me. For about 80% of the time, I mentally characterizedd the AIs as ‘somewhat dysfunctional human character,’ with maybe 20% of the time believing I was reading about non-human intelligences. Contrast that with Children of Time, in which Tchaikovsky was able to create a race of believable, intelligent spiders that did not feel human in their thinking, or Chambers’ A Closed and Common Orbit that centered on an AI trying to figure out an identity. Rust’s AIs almost universally felt like humans with dysfunctional emotional issues. Didn’t work for me. It didn’t impede me from reading, by any means, but it spoke more to my apocalypse issues than issues of AI or self-determination. There’s a twist at the end that hardly seemed a twist at all, and further confused me about its very existence.
“And most systems weren’t top-of-the-line when it came to security, instead running on mainstream driverless systems yanked out of any old car, modified only with a standard widely used manual drive code written twenty-five years back. And this was no exception. The code had eccentricities, and few bots knew enough about them to bother debugging them. If you fucked with the things enough internally, you could force a reset that would give manual control over to the driver, without the need for a password.”
Paragraphs like this left me wondering: why didn’t AIs write new code instead of using 25 year old code that had bugs? Why is an AI saying ‘fucked’ to describe a very human habit of ‘ineffectively screwing around? It reads like “Mad Max,” not computer AI.
From reviews and descriptions, I was a bit afraid this might head into ponderous ‘philosophical’ discussions, musing on what it means to choose one’s identity. It does do that, but in a very accessible and generally interesting way–although again, it didn’t feel alien as much as ’emotionally stunted anti-hero.’ There’s a great deal of action and gunfights–at one point, I found myself wondering about bullet factories–to distract the reader from the thorny character and determination issues. There’s also a bizarre one page discussion about what it means to be ‘God’ in this new order, which seemed oddly out of place. I am almost positive that it would have driven my philosophical friends bananas with logical leaps and assumptions.
Of all of it, I found I was most interested in the backstory, the development of the AIs and the leap into self-direction. I also found myself wondering about the current ability of AIs to manage the parts and power they needed, issues that seemed largely ‘hand-wavey’ for the importance they play. Not bad, certainly, but underwhelming in terms of character and world-building. Honestly, it feels light in actual science, and more like “The Terminator Grows a Heart.” In short: this is Hollywood Sci-Fi. But now it can head back to the library.
Two and a half RAM chips, rounding up because I’m human.