Ought

Ought is a nonprofit operator of a product-driven research lab intended to develop a machine learning system with delegated open-ended thinking, capable of automating and scaling open-ended reasoning. This helps users take any area of life and take the data of that area and turn it into measurable objectives. Ought was founded in 2017 by Andreas Stuhlmüller and Jungwon Byun and is based in San Francisco.

Ought's mission as a nonprofit organization is to develop machine learning systems with good reasoning and machine learning that is capable of helping users with thinking and reflection as much as it can with tasks with short-term, clear outcomes. This mission further goes beyond machine learning, helping or assisting people in automated tasks to develop systems capable of helping users get better at thinking in the short, medium, and long term.

Elicit is Ought's artificial intelligence assistant for researchers and academics that is free to use. Ought's plan for Elicit is to automate and scale open-ended reasoning with language models to make it capable of synthesizing evidence and arguments, designing research plans, and evaluating interventions. Elicit is built in a process-based way on an Interactive Composition Explorer (ICE). ICE is a Python library for compositional language model programs and trace visualizer for those programs.

Elicit's development began with automation literature reviews due to the discipline around synthesizing literature, offering an understanding of the status quo in order to continue to expand the research frontier and to help researchers with literature reviews. After literature reviews, Ought intends to expand the capabilities for Elicit to digest the full text of papers, extract evidence, judge methodological robustness, and help researchers with deeper evaluations. This can, in turn, support researchers in their own study design.