Raw LLM Responses

No they didn’t misunderstand that actually. They literally addressed the possibility of that exact scenario within the article. >>”The report also raises the possibility that, ultimately, the physical bounds of the universe may not be on the side of those attempting to prevent proliferation of advanced AI through chips. “As AI algorithms continue to improve, more AI capabilities become available for less total compute. **Depending on how far this trend progresses**, it could ultimately become impractical to mitigate advanced AI proliferation through compute concentrations at all.” To account for this possibility, the report says a new federal AI agency could explore blocking the publication of research that improves algorithmic efficiency, though it concedes this may harm the U.S. AI industry and ultimately be unfeasible. The bolded is interesting tho because it implies that there could be a hard-limit to how “efficient” an AI model can get in terms of usage. And if there is one, the government would only need to keep tweaking the limit on compute downward until you reach that hard limit. So it actually is possible that this type of regulation (of hard compute limits) could work in the long run.