OpenAI, the artificial intelligence research lab, has been actively working on addressing the challenges posed by superintelligent AI systems. The organization formed the Superalignment team in July to focus on developing ways to steer, regulate, and govern AI systems with intelligence surpassing that of humans. OpenAI co-founder and chief scientist Ilya Sutskever leads the Superalignment team, emphasizing the importance of aligning superhuman AI systems to ensure safety.
- Superalignment Team’s Mission: The Superalignment team’s mission is to build governance and control frameworks applicable to future powerful AI systems. The team aims to address the challenges of aligning models that exceed human intelligence.
- Challenges with Superintelligent AI: Collin Burns from the Superalignment team noted that while current models can be aligned to human-level intelligence, aligning models that surpass human intelligence is less obvious. The team is exploring methods to guide advanced AI models, such as using a less sophisticated AI model to guide a more advanced one.
- Analogy of Weak-Strong Model: The Superalignment team’s approach involves using a weaker, less-sophisticated AI model (e.g., GPT-2) as a stand-in for human supervisors to guide a more advanced, sophisticated model (e.g., GPT-4). This analogy is intended to help prove superalignment hypotheses.
- Crowdsourcing Ideas: OpenAI is launching a $10 million grant program to support technical research on superintelligent alignment. The grant program will reserve tranches for academic labs, nonprofits, individual researchers, and graduate students. OpenAI plans to host an academic conference on superalignment in early 2025 to share and promote research in this area.
- Involvement of Eric Schmidt: A portion of the funding for the grant program will come from Eric Schmidt, former Google CEO and chairman. Schmidt, an active AI investor, is known for expressing concerns about dangerous AI systems. While some see his involvement as AI doomerism or commercial interest, Schmidt emphasizes the importance of ensuring AI alignment with human values.
- Commitment to Open Sharing: OpenAI assures that its research, including code, and the work of others who receive grants and prizes from OpenAI will be shared publicly. The commitment aligns with OpenAI’s mission of contributing to the safety of AI models across various labs.
OpenAI’s efforts reflect a proactive approach to addressing the potential challenges associated with superintelligent AI systems, including safety and alignment issues. The organization emphasizes collaboration, transparency, and public sharing of research to benefit humanity.
🤖🧠🌐 #OpenAI #AIAlignment #SuperintelligentAI #AIResearch #GrantProgram 💡
Check out the latest news in our Global News section
Stay updated on environmental data and insights by following KI Data on Twitter