It should be noted that Thingiverse’s policy is against “firearms” and not guns in general. The company has no problem with replica props, airsoft guns, sci-fi blaster toys, or gun-like objects that shoot candy.
…
“AI will be used only to flag potentially harmful designs, but a human will always be the one to decide if something should be removed,” Chapman told Tom’s Hardware. If a file is removed from Thingiverse, it will be removed by a person, not a machine.
This was my biggest worry, otherwise I see 99% of removed files just being cosplay props
DemBoSain@midwest.social 5 days ago
This is really just a way to save money on human moderators. I’m pretty sure Thingiverse has always forbidden functional weapons. Now they don’t have to examine each one, they’ll just let the machine deal with it.
BeardedGingerWonder@feddit.uk 5 days ago
Wonder how it handles nerf blasters?
Bubs@lemmy.zip 5 days ago
My guess, and confirmed by another comment, is that the ai only flags posts for review. Then the moderators have to manually check the post.
Honestly, it’s not a terrible use of AI in my opinion. Considering posts practically never change, they really only have to scan each post once. The mod can either flag it as safe or remove it. They are probably just running image and text pattern recognition on previously banned posts to flag newly submitted posts.
kn33@lemmy.world 5 days ago
My money’s on “poorly”
MysteriousSophon21@lemmy.world 4 days ago
image recognition AI is notoriously bad at context, so it’ll probably flag half the nerf blasters as “potential weapons” and require human review anyway lol