3 ms·
Anybody who knows what the pigeonhole principle is should know that a lot of these fears are complete bunk.
by alexb_ 4y ago
Anybody who knows what the pigeonhole principle is should know that a lot of these fears are complete bunk.
- cedilla 4y agoI know what the pigeonhole principle is. Could you elaborate how it leads to what fears to be bunk how?
- alexb_ 4y agoThe biggest fears are "AI is theft, the model just remembers what you give it. The proof is in the fact that you can reproduce what it gets". When in reality, a model that is smaller than the training data literally cannot reproduce all of the images it's trained on.
- alkonaut 4y agoThe argument isn't that it can reproduce ALL images, but that it might occasionally reproduce one image. And if that is devastating to e.g. the economic feasibility of a product built on the model - then that's an existential threat to that business model. To make a really dumb example: I can make a black box model that I push in 1000 bestseller novels through, that has the total "storage size" of 1 novel. The pigeonhole principle says my model can't possibly contain all 1000 novels. If you ask it (anything) it will respond with that one novel, verbatim. It never reproduces anything else. Does it now matter whether I trained it on 1000 novels? Does it matter that my "black box" is just that, a literal black cardboard box containing just a copy of a novel?
- pessimizer 4y agoActually a really great example. Wouldn't be surprised if some enthusiastic future corporate-friendly circuit court ruling accidentally made this a legal copyright-removing machine.
- tsukikage 4y agoDoes that really follow, though? I mean, completely true, it is absolutely the case that there will be some inputs that cannot be losslessly compressed. But that hardly seems relevant here: * being unable to recover many or most images does not imply that you will be unable to recover any at all. Being able to recover a nonzero proportion is still a problem even if the proportion is small. * the existence of artifacts in the recovered images is not in itself sufficient to prevent legal claims, and yet ignoring them is plenty sufficient for increasing the proportion of inputs that can be considered recovered
- alexb_ 4y agoMost of the reaction you see from this are to the tune of "The AI people lied to us, the model DOES remember all of the images you train it on, here's proof in that you can get the images back!". When in reality, you absolutely cannot store data to reproduce every image it's trained on because the model isn't big enough to hold even a fraction of what it's trained on.
- tsukikage 4y agoSome. Some images back. I've not seen anyone claim anything like what you suggest. Do you have an example? I mean, it literally says "some" right there in the title, and if you click through and actually read the paper you see it's a tiny proportion.
- ForestCritter 4y agohttps://www.cbc.ca/news/canada/sudbury/ai-generated-art-consent-1.6722981 https://www.cbc.ca/news/canada/sudbury/ai-generated-art-cons... Here is a real world example.
- tsukikage 4y agoThat article, much like the OP, is about someone that found a few specific images were recoverable. I see no claim there that all images are. Am I missing something?