TechFlow news: OpenAI has announced a collaboration with former Google CEO Eric Schmidt to launch a $10 million "Superalignment Rapid Grants" program supporting technical research aimed at ensuring the alignment and safety of superintelligent AI systems. This grant initiative focuses on addressing the alignment and safety challenges posed by future superhuman AI systems—systems that could bring immense benefits but also significant risks. Current AI alignment techniques, such as reinforcement learning from human feedback (RLHF), may not be sufficient to handle the complexity and creativity of superhuman AI. Research areas covered by the grants include generalization from weak to strong supervision, interpretability, and scalable oversight.
The grant program will provide funding ranging from $100,000 to $2 million for academic labs, nonprofit organizations, and individual researchers. For graduate students, OpenAI is also offering a one-year $150,000 Superalignment Fellowship, consisting of a $75,000 stipend and $75,000 in compute and research support. Applicants are not required to have prior experience in alignment research, and first-time researchers in the field are encouraged to apply. The application deadline is February 18.
