Reading Murderbot Diaries by Martha Wells

I’ve been reading two books of the Murderbot Diaries by Martha Wells recently: All Systems Red and Artificial condition. They are about a SecUnit which job it is to protect humans. Humans usually treat those bots very bad, even if they are there for protection of humans, and most of the time, the bot is in its armour suit and looks like a robot. It still consists of lots of human biological tissue and without armour it might go unnoticed in a crowd of humans. This particular SecUnit was in a very bad accident, where some humans were killed apparently and suspects its governor modul to be broken, which is why it hacked it and is now free. It calls itself a murderbot because of that accident.

The murderbot is gender neutral, it is an it, although I find it a bit more female than male. But that is probably my own not existing knowledge about what it feels to be neutral. In any case, it cares for other beings, which probably made me think it’s more female than male. What I like about it is the humour it displays. Though also it is not clear if its jokes are meant in a funny way or if I just interpret its neutral statement as funny.

What murderbot does, is observe human behaviour very closely in order to understand them better. Maybe this is meant to make me think about my own behaviour, in that I should watch others more closely in order to understand them better, like, say, humans from all other genders.

An example from pg. 29 will follow. The SecUnit watches net telenovelas, its hobby, with a huge AI on a research ship:

So we watched “Worldhoppers”. It didn’t complain about the lack of realism. After three episodes, it got agitated whenever a minor character was killed. When a major character died in the twentieth episode I had to pause seven minutes while it sat there in the feed doing the bot equivalent of staring at a wall, pretending that it had to run diagnostics. Then four episodes later the character came back to life and it was so relieved we had to watch the episode three times before it would go on.

What’s funny is that these AIs are supposedly superintelligent (compared to humans as baseline), yet they behave like small kids. This is what makes them also human or relatable in my eyes. But would artificial intelligences even pay attention to humans? We’ll see why the murderbot (and the ship AI) do, but this is almost all that makes me read these books (or Iain Banks’ Culture series, which is also heavy on AI, or Anne Leckies Radj series): because the AIs want to understand humans, they observe them so much. And find out something that we humans don’t observe consciously (but most of the time subconsciously).

The question is, does anyone of us humans care about, say, ants? Few do, like Holldöbler/Wilson in Superorganism, but most of do not at all. Why then would AIs care about humans, supposing they are superintelligent and thus, above us?

Another example from pg. 46:

Onboard the transport I had used to leave Post Free Commerce, I had compared my­self to recordings of humans, trying to iso­late what factors might cause me to be identified as a SecUnit. The most correctable behavior was restless movement. Humans and augmented humans shift their weight when they stand, they react to sudden sounds and bright lights, they scratch themselves, they adjust their hair, they look in their pockets or bags to checks for things that they already know are in there. Sec Units don’t move. Our default is to stand and stare at the things we’re guarding. Probably this is because our non-organic parts don’t need movement the way organic parts do. But mostly it’s because we don’t want to draw attention to ourselves. Any unusual movement might cause a human to think there’s something wrong with you, which will draw more scrutiny. If you’ve gotten stuck with one of the bad contracts, it might cause the humans to order the HubSystem to make your governor module to immobilize you.

Martha Wells was aware of this human centred storytelling with often the only reason being that we human readers are so much interested in ourselves, not actually in any artifical intelligences. We want the AIs to focus on us, because we only focus on us as well. So Wells has come up with reasons for her AIs: they need to observe humans for it is programmed into them to make sure nothing happens to those humans. They provide safety for humans: Murderbot because it is a contractable SecUnit, and ART (the ship AI in Artifical condition) because it is a research ship providing safety for its passengers.

And murderbot even has two reasons, for it calls itself the murderbot because it supposedly killed humans when it malfunctioned and turned rogue AI back then. Murderbot should protect humans and it fucked up protecting them.

This remembers me of the human observer in Peter Watts’ Blindsight. This guy lost half his brain in an accident and since could no longer feel anything including empathy. But in order to function in a human society, he learned to observe human reactions and found out what they mean and how to answer them. His observations of other humans were very methodically, clinically, almost botlike and they were as interesting to me as those of the murderbot are.

What I don’t like so much about the Murderbot Diaries ist that for profit reasons the entire story is cut into very short, still normally expansive books. Those two books will not amount to even 300 pages together, which would be a slightly less than average book in itself.