#OpenWeight

www.scaleway.com

Changelog | Scaleway Documentation

Keep up to date with Scaleway's latest products and features!

en.killbait.com

Y Combinator's Garry Tan Advocates for U.S. Open-Weight AI Labs to Adopt Distillation Techniques on Frontier Models

Y Combinator CEO Garry Tan has expressed his stance on the use of distillation techniques by AI labs, particularly in the context of Chinese AI labs extracting knowledge from frontier model makers. Tan argues that U.S. AI labs should also be allowed to use distillation techniques on American frontier AI models, rather than having regulators intervene. He stated in an interview with CNBC that he would 'do nothing' regarding the issue, suggesting that there should be an 'American distillation regime.' Distillation, a process where a model is extensively prompted to learn how another model works and reasons, is commonly used by AI labs to train new models. Anthropic recently released a report alleging that Chinese labs are engaging in 'illicit distillation attacks,' using hidden identities and stolen credentials to distill models without permission. Anthropic CEO Dario Amodei has called for U.S. regulators to crack down on such practices. However, Tan disagrees with this approach. He clarifies that he is not advocating for American AI labs to use stolen credentials but rather to freely use distillation techniques through legitimate means. His argument is twofold: first, he believes it's an overreach for AI labs to dictate what their customers can do with the information their models share; second, he points out that proprietary AI labs themselves did not ask permission when they ingested vast amounts of human knowledge, including copyrighted material, to train their models. Tan emphasizes that access to intelligence trained on broad public data should be treated as a public good, rather than being locked behind restrictive terms of service. He envisions a balance between open-weight AI labs and frontier labs, where both can thrive. He warns that the true 'AI doomer scenario' is for all the power of frontier AI to be concentrated in a single proprietary provider, leading to a monopolistic situation that would be detrimental to innovation and competition. Tan, who describes himself as an avid AI user, believes that open-weight models provide freedom and access, while frontier labs drive innovation. He supports making frontier AI fundable and sustainable as a business model, ensuring that both open and proprietary models can coexist and benefit the AI ecosystem.

www.nytimes.com

Corporate America Is Getting Hooked on Open-Source A.I.

Companies like AT&T are increasingly using cheap, freely available artificial intelligence models over expensive ones from Anthropic and OpenAI.