• 0 Posts
  • 38 Comments
Joined 1 year ago
cake
Cake day: June 14th, 2023

help-circle

  • Keep in mind, there’s a lot more that goes into the wildly superior Dutch infrastructure than just which streets get bike lanes along them. Roads are also fundamentally deferential to pedestrian and bike traffic there. As in, crossings are elevated above the road, which has a surprisingly effective impact on the culture and behaviors around modes.

    The other issue with bike infrastructure in MA is also the reason we see pockets of it done so well-- cities here have way more power than in other states. It’s a lot easier to get this stuff implemented here with purely local efforts than, say, in WI or wherever. We also just straight up generally love biking here. It’s still totally baffling to me how that fucking gremlin thought he could run for mayor of Somerville on an explicitly antibiking platform. I also wish we could get our shit together in Medford on this, but my city is so god damned broke, I don’t think we could afford the paint for slapping down lines.






  • Friendly reminder that this is person is an idiot and this is a deeply incorrect explanation of the Jewish religious understanding of jews’ relationship to God. In talmudic tradition, it is more or less correct to say that jews are understood to be in a contractual relationship with God.

    I suspect this person has a bad relationship with their ex, and has decided to weave a cloak of straws to justify their shitty relationship and the community and familial support their ex receives to their personal disadvantage. It’s a weirdly common vector of effectively unintentional antisemitism. That, or they might just be dumb.







  • The ultimate issue is that the models don’t encode the training data in any way that we historically have considered infringement of copyright. This is true for both transformer architectures (gpt) and diffusion ones (most image generators). From a lay perspective, it’s probably good and relatively accurate for our purposes to imagine the models themselves as enormous nets that learn vague, muddled, impressions of multiple portions of multiple pieces of the training data at arbitrary locations within the net. Now, this may still have IP implications for the outputs and here music copyright is pretty instructive, albeit very case-by-case. If a piece is too “inspired” by a particular previous work, even if it is not explicit copying it may still be regarded as infringement of copyright. But, like I said, this is very case specific and precedent cuts both ways on it.



  • This is actually an effective measure when you sit down to actually think about this from a policy perspective. Right now, the biggest issue with AI generated content for the corporate side is that there is no IP right in the generated content. Private enterprise generally doesn’t like distributing content that it doesn’t have ability to exercise complete control over. However, distributing generated content without marking it as generated reduces that risk outlay potentially enough to make the value calculus swing in favor of its use. People will just assume there are rights in the material. Now, if you force this sort of marking, that heavily alters the calculus.

    Now people will say wah wah wah no way to really enforce. People will lie. Etc. But that’s true for MOST of our IP laws. Nevertheless, they prove effective at accomplishing many of their intents. The majority of private businesses are not going to intentionally violate regulatory laws of they can help it and, when they do, it’s more often than not because they think they’ve found a loophole but were wrong. And yes, that’s even accounting for and understanding that there are many examples of illegal corporate activity.