Ought

Ought is a non-profit research lab dedicated to developing mechanisms that delegate high-quality reasoning to advanced machine learning systems. The organization spun off Elicit as an independent public benefit corporation, which has successfully raised $9 million in venture funding. Ought has developed tools such as the Interactive Composition Explorer (ICE) and the Factored Cognition Primer, and focuses on process-based supervision for language models. The team operates across various locations including the Bay Area, Austin, New York, and Oristà, and is funded by grants from organizations like Open Philanthropy, Jaan Tallin, and the Future of Life Institute, along with support from individuals in the effective altruism and longtermism communities. Ought’s mission is to scale up good reasoning with the help of machine learning, aiming to make it useful for open-ended questions while avoiding alignment risks posed by end-to-end optimization. Elicit, designed to assist with literature reviews, does not automate the process entirely but enhances it by interacting with multiple ML APIs for different tasks. The effectiveness of Elicit depends on the quality of the underlying research, and users are encouraged to double-check its work and consider its limitations. The development of Elicit balances clarity and accessibility with comprehensiveness and nuance, and is influenced by the latest advancements in machine learning technology.