2 ms·

I feel like a broken record on this topic lately, but I strongly believe that training ML models on copyrighted works should be legal. It is clear to anyone th

I feel like a broken record on this topic lately, but I strongly believe that training ML models on copyrighted works should be legal. It is clear to anyone that understands this tech that it is not simply "memorizing" or "copying" the training data, even if they can be coaxed into doing this for certain inputs (in the current iteration of the tools). Ultimately, I think the problem of reproducing certain popular works or code snippets will be solved. One interesting direction here are the tools of information theory and differential privacy. e.g) proving that certain training inputs cannot be recovered from the weights, or that there is a threshold of how much information can be gleaned from a single input. It is easy to imagine (because its nearly there already) a future version of StableDiffusion (or CoPilot) which provably compresses all training data beyond any possibility of recovery, and yet still produces extremely convincing results which disrupt the creative professions of art and programming. Until we get to that point, it feels like the only consistent and sensible place to apply regulations is with the model end user. When I use CoPilot, I accept the small (honestly overblown) risk that _maybe_ I won't have the license to use some small snippet of code it spits out. But I'm happy to wear that responsibility, because the boost to productivity is so great that I dread a return to pre-CoPilot world. That is, a world where everyone keeps reinventing the same solution to simple problems over and over and over again.

kmeisthax 4y ago
Training being fair use is something I can buy into. As for actually using the model... personally I still find that unacceptable. Even if the risks are low, we have lots of works in the model, so the risk can still add up. The idea you have in your head of training without regurgitation is likely not possible. The underlying technology treats the training set as gospel: the system is trained to regurgitate first, and generalization is a happy accident. Likewise, we can't look into a model to check and see what it's memorized, nor can we trace back an output to a particular training example. Which has ethical implications with the way that AI companies crawl the web to get training data; such models almost certainly hold someone's personal information and there's no way to get it out of there aside from curating your training set to begin with.
- angusturner 4y ago
  I mean, in a sense yes the training set is gospel. But these systems are also (generally) tested against held out data. When you have to model 100TB of images with 4gb of weights there is no way this is possible without learning some kind of patterns and regularity that generalise outside the training set. Most generated items will be novel, and most training items will not be reproducible. It doesn’t seem radical to suggest that the copying issue will continue to recede as we get better models. And there are areas of research specifically concerned with _provably_ showing that you cannot identify what items a model was trained was. Lots of reasons to be optimistic in my view.

kmeisthax 4y ago

Training being fair use is something I can buy into. As for actually using the model... personally I still find that unacceptable. Even if the risks are low, we have lots of works in the model, so the risk can still add up. The idea you have in your head of training without regurgitation is likely not possible. The underlying technology treats the training set as gospel: the system is trained to regurgitate first, and generalization is a happy accident. Likewise, we can't look into a model to check and see what it's memorized, nor can we trace back an output to a particular training example. Which has ethical implications with the way that AI companies crawl the web to get training data; such models almost certainly hold someone's personal information and there's no way to get it out of there aside from curating your training set to begin with.

angusturner 4y ago
I mean, in a sense yes the training set is gospel. But these systems are also (generally) tested against held out data. When you have to model 100TB of images with 4gb of weights there is no way this is possible without learning some kind of patterns and regularity that generalise outside the training set. Most generated items will be novel, and most training items will not be reproducible. It doesn’t seem radical to suggest that the copying issue will continue to recede as we get better models. And there are areas of research specifically concerned with _provably_ showing that you cannot identify what items a model was trained was. Lots of reasons to be optimistic in my view.