Recursive, a new artificial intelligence research company founded by former leaders from OpenAI, Google DeepMind, Meta AI, Salesforce AI, and Uber AI, announced it is emerging from stealth with a mission to develop self-improving AI systems capable of conducting experiments to safely improve themselves through automated scientific discovery.
The company revealed it has raised $650 million in a heavily oversubscribed funding round at a valuation of $4.65 billion. The financing round was led by GV and Greycroft, with participation from AMD Ventures and NVIDIA.
According to Recursive, the company believes the fastest path to superintelligence will come from AI systems that can recursively improve themselves through open-ended algorithms designed to generate continuous innovation. The company believes human intelligence emerged through open-ended evolutionary and cultural processes that continually built on previous discoveries.
Recursive said that current scientific discoveries in artificial intelligence are still largely driven by human researchers, but the company sees a broader trend in machine learning in which increasing compute and data replace hand-designed methods with AI-driven approaches. The company plans to first focus on AI systems that improve AI itself before eventually expanding those capabilities into broader scientific disciplines.
The company emphasized that safety will remain a core priority as it develops recursively self-improving AI systems. Recursive said it aims to maximize the benefits of advanced AI systems for humanity while reducing associated risks.
Recursive’s founding team includes researchers and entrepreneurs who previously helped create AI research organizations at Salesforce and Uber and led teams at OpenAI, DeepMind, Google Brain, and Meta. The company also noted that several team members have previously founded successful startups, including companies that achieved unicorn status or were acquired by major technology companies such as Salesforce, Meta, and Uber.
The company said its team, which currently consists of more than 25 employees and continues to grow, has contributed to advancements across several AI research areas including open-ended algorithms, quality diversity algorithms, AI-generating algorithms, self-improving coding agents, automated red teaming, prompt engineering automation, foundational world models, vision transformers, retrieval-augmented generation, and AI scientist systems.
Recursive said it is actively recruiting researchers and engineers interested in advancing safe, open-ended, recursively self-improving AI systems. The company currently operates offices in San Francisco and London.