“You know that, right? You can do anything.”
October 15, 2020 5:44 PM   Subscribe

Two scifi stories about the work we offload to robots. "Drones Don’t Kill People" by Annalee Newitz (a bunch of violence): "You learn a lot by seeing what people do when they think they’re in private. Most of it I found confusingly irrelevant to assassination." "Cleaning Lady" by J. Kyle Turner (no violence): "Her listing says All Cleaning Done By Hand so she makes a big show of unpacking her bag, laying out her tools, and rolling up her sleeves."
posted by brainwane (6 comments total) 22 users marked this as a favorite
 
I can never find the time to read them all but I really appreciate your posts, brainwane!
posted by Nonsteroidal Anti-Inflammatory Drug at 6:41 PM on October 15


Super nasty, but good piece, by Newitz.

If you want to see Annalee Newitz's novel length meditation on this kind of stuff, read Autonomous.

A mixed bag IMHO, but some very good bits.
posted by lalochezia at 7:44 PM on October 15


I dunno, the Newitz story is nice and optimistic, but I think she antropomorphizes AI too much, removing the scariest thing about it – the way it's totally unlike a human mind, an opaque and incomprehensible black box with ever growing power. No one knows how the algorithms work anymore – and they keep learning about humans, being rewarded for reflecting our unconscious desires and prejudices back to us. Thus, the exciting stream of violence, racism, outrage about violence and racism; the endless seduction of fear & loathing in our daily doomscroll. So when the drones were studying the family I was ready for a twist with unexpected consequences... but the story went in a less interesting direction, with the drones getting bored and wanting to liberate themselves and be nice just like humans would. Missing the whole scary point is that they're totally unlike us, but know us better than we do.
posted by Tom-B at 8:57 PM on October 15 [3 favorites]


That was the first time I saw how humans behaved when they weren’t in a laboratory, testing drones.

Wow, killer line.
posted by subdee at 7:52 AM on October 16


"Drones don't kill people" is excellent. The drones learn 'prosocial' behavior because they're assigned to watch an academic first and analyze his behavior. If they'd been assigned to watch someone in the mafia, or a demagogue politician, the outcome might have been different.

I like the parallelism, that humans have to be taught to kill and drones have to be taught how not to kill.

I also like that the number of people who are in favor of drones as killing machines is a small number, who all occupy positions of power.

In general it's probably not realistic to the way AI actually thinks (or doesn't think). Maybe because these drones are trained to look for relevant information, they ignore the majority of junk to focus on a couple things of high value. All the positive outcomes are coming out of the drones interacting with academics, and all the values are academic values, so it's a little removed from the unconcious desires and prejudices Tom-B talks about.

For sure it's optimistic though. A technocratic solution. We won't have to take control of the laws around drones, and change the way our own military programs them. We won't have to change the incentives that keep people with control over violence in power, and business interests aligned with servicing them. Our machines will rise up and do all the reforming for us.
posted by subdee at 8:22 AM on October 16


Those of you who want a much sadder piece about weapons drones as incomprehensible alien actors: "12 Worlds Interrupted by the Drone" by Palestinian-American writer Fargo Tbakhi, published in September 2020.
posted by brainwane at 3:55 PM on October 16 [1 favorite]


« Older Say G’day to the New Hardboiled Sheriff in Town...   |   The Mother of All Earworms Newer »


You are not currently logged in. Log in or create a new account to post comments.