★★★★☆ Worth it just to see a realistic story about AI failure.
(Military SF) After corporal Cooper recovers from the disastrous battle of Kuching, he’s sure the sentient tank he worked with is still there somewhere, so he takes extended leave and goes to hunt for it. (3,623 words; Time: 12m)
"The Lori," by Fiona Moore [bio] (edited by Neil Clarke), appeared in Clarkesworld issue 167, published on August 1, 2020.
Mini-Review (click to view--possible spoilers)
Review: 2020.404 (A Word for Authors)
Pro: Although we don’t quite understand why Cooper is so attached to his old tank, the story make his obsession feel very real. It also does a nice job of gradually cluing us in as to what exactly happened and why the military lost its interest in sentient tanks.
From a technology perspective, this is a beautiful example of how real AI might fail. Kursk didn’t go off the rails because it “decided” to kill people or because it “overrode it’s own programming.” Nope. The problem was that the basic programming about protecting the property of the agribusinesses resulted in the tanks calculating that the property would be more valuable if the ecology were better protected, and so the tanks started killing the “wrong” people. When Cooper finds it, Kursk is spending its time protecting a single forest.
Cooper eventually learns this himself, but at terrible cost.
Con: There’s a word for what Cooper does, and it’s called “murder.” No, I’m not talking about him deactivating the tank; I’m talking about him killing the people he mistakes for “Reds.” There’s no military action there, and he has no rules of engagement, so everyone he kills is someone he murdered. The fact that he doesn’t seem terribly bothered by that, is hard to believe and makes him seem like a total psychopath. Yes, it bothers him enough to deactivate Kursk, but it’s on the order of, “Oops; that didn’t work.”
Other Reviews: Search Web
Fiona Moore Info: Interviews, Websites, ISFDB, FreeSFOnline
Follow RSR on Twitter, Facebook, RSS, or E-mail.
Pro: Although we don’t quite understand why Cooper is so attached to his old tank, the story make his obsession feel very real. It also does a nice job of gradually cluing us in as to what exactly happened and why the military lost its interest in sentient tanks.
From a technology perspective, this is a beautiful example of how real AI might fail. Kursk didn’t go off the rails because it “decided” to kill people or because it “overrode it’s own programming.” Nope. The problem was that the basic programming about protecting the property of the agribusinesses resulted in the tanks calculating that the property would be more valuable if the ecology were better protected, and so the tanks started killing the “wrong” people. When Cooper finds it, Kursk is spending its time protecting a single forest.
Cooper eventually learns this himself, but at terrible cost.
Con: There’s a word for what Cooper does, and it’s called “murder.” No, I’m not talking about him deactivating the tank; I’m talking about him killing the people he mistakes for “Reds.” There’s no military action there, and he has no rules of engagement, so everyone he kills is someone he murdered. The fact that he doesn’t seem terribly bothered by that, is hard to believe and makes him seem like a total psychopath. Yes, it bothers him enough to deactivate Kursk, but it’s on the order of, “Oops; that didn’t work.”
Other Reviews: Search Web
Fiona Moore Info: Interviews, Websites, ISFDB, FreeSFOnline
Follow RSR on Twitter, Facebook, RSS, or E-mail.
No comments (may contain spoilers):
Post a Comment (comment policy)