Frontier AI Research
Pushing reliability and capability with real-world evaluation and new paradigms.
- Investigate the fundamental limits of today's AI systems and explore new paradigms that unlock greater capability and reliability.
- Build real-world task-specific RL training environments, dataflows, benchmarks, and evaluations to track human-in-the-loop outcomes.
- Explore the potential of AI to accelerate breakthroughs across science, biology, climate, and beyond.