Recursive self-improving superintelligence to automate knowledge discovery.

Human intelligence was created by the open-ended processes of Darwinian and cultural evolution. Both processes grow an archive of interestingly different discoveries, and each innovation builds on those that came before. These processes invented bodies, sight, simple reflexes, then reasoning, language, and science. They’ve taken us from the first replicating molecule to the moon. They have no ceiling and keep innovating forever.

The science of AI involves the same open-ended process of innovation. So far, these discoveries have been produced by human scientists.

But a clear trend in machine learning is that, as we get more compute and more data, hand-designed methods are replaced by AI-driven processes.

Designing endlessly self-improving systems

Recursive embraces the logical conclusion: the fastest path to superintelligence will be realized by AI that recursively improves itself, and does so via open-ended algorithms that drive endless innovation. We will first focus on the science of AI itself (by creating AI that improves AI), but the playbook we create will soon allow us to revolutionize every scientific discipline. The potential benefits for humanity of safely creating such an advance cannot be overstated.

Throughout, we will prioritize safety. We must make sure the system helps humanity flourish by maximizing the benefits while reducing risks.

Our team

Recursive’s co-founders are leading researchers and successful entrepreneurs. We have created the AI research labs at Salesforce and Uber, and led teams at OpenAI, DeepMind, Google Brain, and Meta. We have also founded many successful companies, including two unicorns and ones that were acquired by top tech companies, including Salesforce, Meta, and Uber.

Our team (over 25 and growing) are pioneers in many areas that are critical to creating recursive self-improvement. We have helped lead major advances in open-ended algorithms, quality diversity algorithms, AI-generating algorithms, self-improving coding agents, automated red teaming and capability discovery, prompt engineering and automations of it, generating learning challenges and environments, foundational world models, deep learning in NLP, vision transformers, retrieval-augmented generation, and AI scientists.

Select previous leadership roles

Build with us

We are confident it is time to massively scale up safe, open-ended, recursively self-improving AI. It’s the most exciting time in history to be an AI researcher, especially focused on these topics. The potential benefits to humanity are astronomical.

To achieve those goals, many research and engineering challenges remain to be solved. We are building the best team in the world to work on these fascinating challenges and opportunities. Our offices are in San Francisco and London.

If you're passionate about these topics, please contact us at talent@recursive.com.