How can we delegate the world’s hardest questions to machine and human experts?
Abstract: Machine learning will transform any area of life that has abundant data and easily measurable objectives. But will it improve how we think, reflect, and do research? We can take small steps in this direction using language models like GPT3. I'll first explain language models from basics, then show Elicit, the GPT3-based tool we're building to incrementally automate the work researchers do.
Bio: Andreas is co-founder of Ought, a non-profit research lab that builds machine learning tools for research and thinking. Before Ought, Andreas was a postdoctoral researcher in Noah Goodman's Computational Cognitive Science lab at Stanford. He co-created WebPPL, a programming language for probabilistic machine learning. He holds a Ph.D. in Cognitive Science from MIT, where he worked in Josh Tenenbaum's group.
Speaker:
Andreas Stuhlmüller, ought.org
Sponsored by SAS
21 авг 2024