alt Hacker News> The obvious problem: you cannot reliably detect firearms from geometry alone.
The obvious problem with this argument is that in just the medium term, world-model style AI will get good at this task, but having big brother pre-approve every print will still be bad.
Replies
I think it's still not a viable problem to solve.
What happens if you print the handle on a different printer, and print it with an attachment which works as an ice-cream scoop?
Or how about you actually print an ice-cream scoop, and then stop the print halfway to just take the handle, and do the same for several other innocent looking parts which are carefully modelled to fit together after printing individually. There are just so many ways to get around any measures they could put in place.
How? The printer only ever retrieves G code for individual parts without any knowledge of what they are going to be assembled into. There is no viable way to solve this classification problem on this kind of incomplete data, is there?