March 28, 2024
To Govern AI, We Must Govern Compute
Computing power—compute, for short—is a key driver of artificial intelligence (AI) progress. Over the past 13 years, the amount of compute used to train leading AI systems has increased by a factor of 350 million. This has enabled the major AI advances that have recently gained global attention. However, compute is important not only for the progress of AI but also for its governance. Governments have taken notice. As we argue in a recent paper, they are increasingly engaged in compute governance: using compute as a lever to pursue AI policy goals, such as limiting misuse risks, supporting domestic industries, or engaging in geopolitical competition.
The Biden administration introduced export controls on advanced semiconductor manufacturing equipment and the most high-end AI-relevant chips, aimed at undercutting Chinese access to leading-edge AI applications. In October 2023, the administration’s executive order “On the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” introduced reporting requirements on models trained using more compute (1026 operations) than any that have been trained to date. Overseas, the EU’s AI Act will place additional requirements on foundation models trained using more than 1025 operations, currently covering three or four existing systems (GPT-4, Gemini Ultra, Claude 3, and Inflection 2).
Just because compute can be used as a tool to govern AI doesn’t mean it should be used in all cases.
The Biden administration introduced export controls on advanced semiconductor manufacturing equipment and the most high-end AI-relevant chips, aimed at undercutting Chinese access to leading-edge AI applications. In October 2023, the administration’s executive order “On the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” introduced reporting requirements on models trained using more compute (1026 operations) than any that have been trained to date. Overseas, the EU’s AI Act will place additional requirements on foundation models trained using more than 1025 operations, currently covering three or four existing systems (GPT-4, Gemini Ultra, Claude 3, and Inflection 2).
States understand the importance of compute. The U.S. and the EU are both investing $50 billion in subsidies through their Chips Acts. Companies understand its importance, too. Almost all start-ups working on advanced AI have entered into “compute partnerships” with U.S. Big Tech compute providers. This includes most recently the French company Mistral, even though it had branded itself as a French-European national champion. Microsoft is reportedly investing $50 billion into expanding its Azure data centers worldwide, one of the biggest corporate infrastructure investments ever. NVIDIA has rocketed up to having the third biggest market capitalization in the world. And to boost compute production, Sam Altman is reportedly trying to raise $7 trillion.
Read the full story and more from Lawfare.
More from CNAS
-
U.S. Chip Controls and the Future of AI Compute
That escalated quickly! Emily and Geoff discuss why the U.S. aim to deny China access to the computing power necessary for frontier AI capabilities has led to an ever expandin...
By Emily Kilcrease, Geoffrey Gertz & Pablo Chavez
-
Asymmetry and AI: The Battle for Power
Paul Scharre, Vice President and Director of Studies at CNAS, joins Zero Pressure to discuss the world of asymmetric warfare, a term used to describe imbalances in conflict. F...
By Paul Scharre
-
Competition, Not Control, is Key to Winning the Global AI Race
The United States, with much of the world’s AI-enabling infrastructure, has positioned itself as the global leader in AI innovation. That might not be the case for much longer...
By Keegan McBride & Matthew Mittelsteadt
-
Regulating AI Is Easier Than You Think
Countries can regulate AI from the ground up by controlling access to highly specialized chips...
By Paul Scharre