Can the US and China verify an AI deal? Start-ups are starting to build the tools
There are growing calls for Washington and Beijing to agree to slow down development of artificial intelligence. But even if they wanted to do so, how could they verify that the other side would do the same? A group of researchers and start-ups is developing technology intended to make this a possibility, without relying on trust alone. Their proposed tools would allow third parties and…
The United States and China are considering agreements to limit the development of artificial intelligence (AI). However, verifying that the other country adheres to such an agreement poses a challenge. A group of researchers and start-ups is developing technology to enable third parties and governments to verify some claims about AI model training or usage without needing access to the AI models or infrastructure.
This technology, called "AI verification," is gaining attention ahead of a meeting between US President Donald Trump and Chinese President Xi Jinping, where AI is expected to be a topic of discussion. While the development of AI verification technology is still in its early stages, increasing interest from policymakers and industry is expected to accelerate progress.
One proposed verification mechanism is called "recomputation," which involves checking an AI system's outputs by running the same calculation again on a different system to ensure the same result is achieved. Another approach is using cryptographic techniques like "zero-knowledge proof," which allows parties to verify that a response to a query was provided by the exact AI model specified without revealing any other information.
Despite political and commercial barriers, both countries possess significant resources in AI development, and some experts believe that technical solutions to AI verification could help build early confidence in an agreement.
Written by urgent.news from South China Morning Post's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
