Miles Brundage will lead two live Demo Q&A sessions on Societal Implications of Large Language Models at #werobot. The first session will be held at 11:00am on Friday, September 24th; the second session will be held at 2:30pm on Saturday, September 25th. We suggest viewing the recorded demo in advance of the Q&A.
Large language models, which learn to process and produce language by consuming gigabytes of text, raise a number of societal risks and opportunities. Large language models have raised concerns about disinformation, bias, privacy, and more, and have been applied to applications ranging from search to question answering to entertainment. This demo will show the performance of GPT-3, a state of the art language model debuted last year by OpenAI, and is intended to spark a discussion about the legal and broader societal implications of such systems. In addition to showing the performance of the system interactively with conference participants, this demo might also feature commentary from the demonstrators on legal issues in which they have conducted research such as intellectual property, and possible legal interventions to mitigate some risks associated with language models, such as bot disclosure laws.