by chriskanan on 1/16/25, 8:00 PM with 130 comments
by chriskanan on 1/16/25, 9:04 PM
I am Christopher Kanan, a professor and AI researcher at the University of Rochester with over 20 years of experience in artificial intelligence and deep learning. Previously, I led AI research and development at Paige, a medical AI company, where I worked on FDA-regulated AI systems for medical imaging. Based on this experience, I would like to provide feedback on the proposed export control regulations regarding compute thresholds for AI training, particularly models requiring 10^26 computational operations.
The current regulation seems misguided for several reasons. First, it assumes that scaling models automatically leads to something dangerous. This is a flawed assumption, as simply increasing model size and compute does not necessarily result in harmful capabilities. Second, the 10^26 operations threshold appears to be based on what may be required to train future large language models using today’s methods. However, future advances in algorithms and architectures could significantly reduce the computational demands for training such models. It is unlikely that AI progress will remain tied to inefficient transformer-based models trained on massive datasets. Lastly, many companies trying to scale large language models beyond systems like GPT-4 have hit diminishing returns, shifting their focus to test-time compute. This involves using more compute to "think" about responses during inference rather than in model training, and the regulation does not address this trend at all.
Even if future amendments try to address test-time compute, the proposed regulation seems premature. There are too many unknowns in future AI development to justify using a fixed compute-based threshold as a reliable indicator of potential risk. Instead of focusing on compute thresholds or model sizes, policymakers should focus on regulating specific high-risk AI applications, similar to how the FDA regulates AI software as a medical device. This approach targets the actual use of AI systems rather than their development, which is more aligned with addressing real-world risks.
Without careful refinement, these rules risk stifling innovation, especially for small companies and academic researchers, while leaving important developments unregulated. I urge policymakers to engage with industry and academic experts to refocus regulations on specific applications rather than broadly targeting compute usage. AI regulation must evolve with the field to remain effective and balanced.
---
Of course, I have no skin in the game since I barely have any compute available to me as an academic, but the proposed rules on compute just don't make any sense to me.
by chriskanan on 1/16/25, 8:05 PM
I personally think the regulation is misguided, as it assumes we won't identify better algorithms/architectures. There is no reason to assume that the level of compute leads to these problems.
Moreover, given the emphasis on test-time compute nowadays and that it seems like a lot of companies have hit a wall with performance gains with trying to scale LLMs at train-time, I especially think this regulation isn't especially meaningful.
by geuis on 1/16/25, 8:47 PM
by cube2222 on 1/16/25, 9:10 PM
And the second level, for some reason, includes (among others) a bunch of countries that would normally be seen as close US allies - e.g. some NATO countries (most of Central/Eastern Europe).
by intunderflow on 1/16/25, 8:37 PM
by mlfreeman on 1/16/25, 8:47 PM
I can't seem to find any information about that anywhere.
by wslh on 1/16/25, 9:15 PM
by clhodapp on 1/16/25, 9:20 PM
Usually, the US government tries not to do that.
by casebash on 1/17/25, 3:14 PM
If you think upcoming models aren't going to be very powerful, then you'll probably endorse business-as-usual policies such as rejecting any policy that isn't perfect or insisting on a high bar of evidence before regulating.
On the other hand, if you have a world model where AI is going to provide malicious actors with extremely powerful and dangerous technologies within the next few years, then instead of being radical, proposal like this start appearing extremely timid.
by 1vuio0pswjnm7 on 1/17/25, 7:55 PM
by veggieroll on 1/16/25, 9:27 PM
by PostOnce on 1/16/25, 9:56 PM
by resters on 1/16/25, 8:34 PM
Regardless of who is currently in the lead, China has its own GPUs and a lot of very smart people figuring out algorithmic and model design optimizations, so China will likely be in the lead more obviously within 1-2 years, both in hardware and model design.
This law is likely not going to be effective in its intended purpose, and it will prevent peaceful collaboration between US and Chinese firms, the kind that helps prevent war.
The US is moving toward a system where government controls and throttles technology and picks winners. We should all fight to stop this.
by djoldman on 1/16/25, 9:49 PM
https://www.federalregister.gov/documents/2025/01/15/2025-00...
by pjmlp on 1/16/25, 9:32 PM
by ChrisArchitect on 1/16/25, 9:42 PM
WH Executive Order Affecting Chips and AI Models
by United857 on 1/16/25, 9:48 PM
by miovoid on 1/17/25, 6:57 AM
by neilv on 1/16/25, 8:35 PM
There should be a federal regulation about that.
by saberience on 1/17/25, 12:48 PM
I don’t see how we can assume it will be enacted at all.
by chriskanan on 1/16/25, 9:08 PM