Clem Delangue, Founder of Hugging Face defends open-source AI models as he discusses the necessity of open-source AI, preventing big tech companies from dominating the market, the benefits of collaboration and transparency, potential safety concerns, concentration of power in AI, differentiating between open and closed AI models, and the importance of transparency and openness in AI systems.
Open-source AI is necessary to prevent concentration of power in the hands of big tech companies.
Regulation should focus on specific domains and use cases to address safety risks of open-source AI.
Advocating and supporting open-source AI initiatives is crucial for continued innovation and competitiveness.
Deep dives
The Importance of Open-Source AI
Open-source AI is seen as a necessary enabler for smaller companies and startups to compete in the AI industry. It allows for the creation of millions of jobs and prevents power from being concentrated in the hands of a few technology companies. While some argue that open-source AI poses safety risks, the same risks also apply to closed-source systems. The approach should be domain-specific, applying limitations where necessary.
Defending Open-Source AI
The CEO of Hugging Face, an AI company, has been advocating for open-source AI in meetings with lawmakers. He emphasizes the importance of openness and transparency in AI, as well as the need for regulation that focuses on specific domains, use cases, and risks. Open-source AI enables collaboration and innovation, ensuring that power is not concentrated in the hands of a few technology companies.
Ensuring Competition and Innovation
The concentration of market power in AI is a concern, and the CEO stresses the need to prevent it. Encouraging open science and open-source AI is crucial for fostering competition and giving smaller companies a fair opportunity to compete with incumbents. Efforts such as releasing public datasets and shared compute infrastructure help level the playing field. However, implementing licenses for AI is seen as dangerous, as it could stifle innovation and further concentrate power.
Balancing Openness and Safety
Open-source AI is not a completely unregulated realm. Platforms like Hugging Face and GitHub have moderation measures in place to ensure safety. Limiting model distribution, using stage releases, and associating legal licenses to models and datasets help prevent misuse. While no system is foolproof, the existing legal system can handle illegal actions. Gradually increasing transparency and openness is seen as beneficial for maximizing public good.
Global Efforts in Open-Source AI
Open-source AI is gaining traction globally, with countries like France and the UAE investing in it. The CEO highlights the importance of the US maintaining its leadership in open-source AI, as it could impact its overall leadership in the AI field. Advocating and supporting open-source AI initiatives is crucial to ensure continued innovation and competitiveness.
Clem Delangue says open-source AI was vital to starting his company, Hugging Face. Now, he finds himself defending open AI models as Washington considers new regulations. On POLITICO Tech, Delangue explains why he views open-source AI as not only safe, but necessary to prevent big tech companies from gaining more market power.