Re:NTIA Solicits Comments on Open-Weight AI Models
by convexstrictly on Saturday February 24, 2024 @06:27PM
(#39496094)
The comments will inform the drafting of regulations on open weight models under the Biden executive order on AI using his powers under the Defense Production Act.
The NTIA website says "Link to the Request for Comment is below. This will be updated with a link to the Federal Register once formally published, along with the final due date, which will be 30 days after formal publication and any other changes made in the final published version."
Re:NTIA Solicits Comments on Open-Weight AI Models
by jph00 on Saturday February 24, 2024 @07:07PM
(#39496303)
Please folks, do take the time to submit a carefully written comment to this, because if legislation is introduced to limit or ban open models, then it's going to cause huge pain for society at large -- possibly as far as undoing the very principles of the Enlightenment. And don't think it couldn't happen; there's a reason that these comments are being sought, which is that regulations of open models are being seriously considered.
I spent months researching and writing an analysis of AI regulation that covered open models in some detail, so please read this if you're interested in learning more:
Re:NTIA Solicits Comments on Open-Weight AI Models
by convexstrictly on Saturday February 24, 2024 @07:34PM
(#39496479)
Jeremy makes compelling arguments. Here are some more mundane corollaries:
It is only a matter of time before you and your company are affected by the pending regulations. In the future, almost all software products will be using AI models, in the same way that most software products use the Internet today, whereas they did not in the 1990s.
Imagine you had to license Oracle software because MySQL or PostgresSQL could not offer certain capabilities, or are less capable because of regulation.
Now also imagine that your products have to agree with the political world view of either Sundar Pichai or Elon Musk. And if you need capabilities only present in one of the commercial alternatives, you don't even that that choice.
Re:NTIA Solicits Comments on Open-Weight AI Models
by RcouF1uZ4gsC on Saturday February 24, 2024 @07:47PM
(#39496565)
Also once this door opens, maybe they start regulating the hardware since it is easier than the software.
Maybe all ML training and inference hardware can only be sold to “certified” vendors (for safety’s sake, of course). After all, since you can rely on the cloud, why would any one need on-premise, potentially dangerous ML inference.
Fact Sheet: https://www.whitehouse.gov/briefing-room/statements-releases...
Full Details: https://www.whitehouse.gov/briefing-room/presidential-action...
https://www.ntia.gov/federal-register-notice/2024/dual-use-f...
The NTIA website says "Link to the Request for Comment is below. This will be updated with a link to the Federal Register once formally published, along with the final due date, which will be 30 days after formal publication and any other changes made in the final published version."
https://www.regulations.gov/document/NTIA-2023-0009-0001
I spent months researching and writing an analysis of AI regulation that covered open models in some detail, so please read this if you're interested in learning more:
https://www.fast.ai/posts/2023-11-07-dislightenment.html
It is only a matter of time before you and your company are affected by the pending regulations. In the future, almost all software products will be using AI models, in the same way that most software products use the Internet today, whereas they did not in the 1990s.
Imagine you had to license Oracle software because MySQL or PostgresSQL could not offer certain capabilities, or are less capable because of regulation.
Now also imagine that your products have to agree with the political world view of either Sundar Pichai or Elon Musk. And if you need capabilities only present in one of the commercial alternatives, you don't even that that choice.
Maybe all ML training and inference hardware can only be sold to “certified” vendors (for safety’s sake, of course). After all, since you can rely on the cloud, why would any one need on-premise, potentially dangerous ML inference.