Sen. Josh Hawley’s “Separation of American Artificial Intelligence Capabilities from China Law” was introduced on January 29th and proposes major restrictions on the exchange of AI research technology between the United States and China. The law could have far-reaching implications for institutions conducting generative AI research, such as UC Berkeley.
The bill seeks to prevent the import and export of AI technology to China, and directly affects the way US researchers, including on campus, work with Chinese companies. These entities are broadly defined in the bill as labs, universities, or other organizations affiliated with Chinese institutions that could affect various academic partnerships.
Ritwik Gupta, a PhD campus AI candidate, shared insights into the potential impact of the bill, noting that its implications are much broader than the public’s online response suggests.
The proposed limitations could reconstruct how campus researchers participate in the global AI community. Gupta said the bill could affect the use of popular global AI development tools, such as Pytorch, a library used by many machine learning researchers.
He added that using these tools could be limited under the bill.
“Let’s say a Chinese with concerns Chinese people provide some code for popular projects like Pytorch and commit to Github. Then I’m on the other side of the American world, Clone the storage and lower those changes. I imported the Chinese AI code. Is it illegal? Is it going to prison now and paying a million dollar fine as the bill says? ” Gupta said.
There are also potential impacts, including doctoral degrees. A campus student who participated in a Chinese undergraduate institution. According to Gupta, some campus doctoral students come from such institutions, and in many cases their first papers have been co-authored with professors at universities in China. He added that this will lead to major losses in the AI ​​research community on campus.
Furthermore, the bill, which is presented as protecting the American industry, could potentially lead to damaging consequences in the development of AI models, Gupta said.
“What Senator Holy doesn’t realize is that the US is dependent on China, and China relies on the US to advance with AI. AI itself is stateless and the development of this technology You can’t arbitrarily split into two worlds because it actually leads to worse outcomes for everyone. And the first thing that will be affected by this is American companies,” Gupta says. I did.
However, the outcome of the proposed bill is still uncertain.
Despite the serious restrictions presented in the bill, Gupta expressed skepticism about its passing due to its broad language and the possibility of opposition.
“At this moment, I’m going to continue to share my work openly,” Gupta said. “We’ll reconsider that strategy when the bill is nearing passage. Until then, we’ll urge senators to reconsider this ultimately harmful strategy.”
