Flapping Airplanes is a frontier data efficiency lab conducting foundational artificial intelligence research. Currently operating in stealth, the organization aims to develop models that can achieve human-level reasoning while learning from far less data than today’s systems. It estimates that humans are 100,000 to 1,000,000 times more sample-efficient than existing models and believes that closing this gap will require substantial new ideas rather than incremental changes to current techniques.

The lab takes a long-term, open-ended approach, pursuing radical research without focusing on a single technical method or industry. Its stated priority is paradigm-shifting research, and it is not currently seeking to commercialize its work. In the future, advances in data efficiency could create value across enterprise applications, robotics, trading, and scientific discovery. The team brings together experienced research scientists and younger researchers with achievements including international Olympiad medals. Flapping Airplanes has raised $180 million and is backed by Google Ventures, Sequoia Capital, Index Ventures, and other investors, with advisors and angels including Chris Ré, Andrej Karpathy, and Jeff Dean.