Garry Tan, the head of Y Combinator's G Sigma Lab, is calling for smaller, American open-weight artificial intelligence labs to adopt similar training methods used by their counterparts in China.
Tan wants these US-based labs to replicate the type of data-driven approaches that have proven successful in Chinese AI research. This means using more data from everyday life and incorporating human judgment into the AI development process. The goal is to create a robust set of open-weight options for the United States, which are not being replicated by many other countries.
By adopting these techniques, US-based labs can improve their chances of success in a market dominated by Chinese companies like Baidu and Tencent. Tan has been vocal about the importance of supporting domestic AI innovation, and he believes that this approach will help to level the playing field for American tech startups. This could lead to more opportunities for growth and development in the US, as well as increased competitiveness in key areas such as autonomous vehicles and facial recognition technology.