Claims Browser

Models that can use the most inference-compute will be able to push the limits of the hardest problems

"The model that can use the most inference-compute will be able to push the limits of the hardest problems."

Successfully achieving a coordinated AI slowdown between competing firms requires two key variables: transparency about technology development and ability to model other firms as trustworthy, rational actors

"The conclusion is that the two key variables in achieving stable outcomes are some level of transparency about technology development, as well as being able to model the other firms as trustworthy, rational actors."

Sincere disagreements about AI policy usually boil down to disagreements about the expected pace of progress and what future AIs will be able to do

"sincere disagreements about AI policy usually boil down to disagreements about the expected pace of progress, and what we expect future AIs will be able to do."

Public opposition to data centers is rising rapidly according to an Annenberg Public Policy Center poll

"this new poll from the Annenberg Public Policy Center is really striking. Opposition to data centers is rising fast"

Opposition to data centers crosses party lines

"it is a sentiment that crosses party lines"

The AI industry is experiencing a bubble with excessive funding from retirement and oil money

Steel industry has strict global standards enforced by corporations, governments, and scientists with independent verification - suggesting AI should have similar regulation

Open weights AI could be the difference between being free yeomen versus corporate serfs if AI becomes the linchpin of the future

Open weights AI removes the possibility of gatekeeping and enables criminals to commit crimes including hacking, child pornography, harassment, and terrorism

Google kept frontier models behind APIs for a few billion dollars in revenue instead of open-sourcing them

Google could have become the dominating force in AI by open-sourcing Gemini, Veo, and Nano models

Gemini team members may be interested in transitioning to work on open models or open-source AI

Treating APIs differently from open weights in AI regulation framework is good policy because they represent different layers of the stack with different risk profiles

Model weights are like raw materials (steel) that don't do anything on their own and should not be regulated the same way as APIs or applications

Regulating model weights would do the most damage to AI development because they are the foundational layer that everything else is built on top of

Anthropic and OpenAI are not colluding to slowdown AI development, but rather asking governments to build a regulatory framework that could mandate slowdowns at some point

Governments have the authority to regulate AI companies in various ways, including implementing the type of framework requested by AI labs

Antitrust law was built for things like price fixing, not AI development slowdown requests, and government has broad latitude with many exceptions in what to prosecute