Building superintelligent autonomous systems

Our goal is to build superintelligent autonomous systems. We believe that solving autonomous coding is the root node problem that will enable superintelligence more broadly. If you build a superintelligent autonomous coding system, all other verticals of computer-based work will follow.

Our research sits at the intersection of reinforcement learning and language models. Over the last decade, our team built some of the most capable AI systems, such as AlphaGo and Gemini, in both fields. Our research bet for building superintelligence is to improve the autonomous capabilities of language models with reinforcement learning. One way to think of our research agenda is by asking the question – how do we get a language model to exhibit the same level of autonomy on a computer as AlphaGo did at Go?