Panelists: Matthew Gold (Grad Center, CUNY), Andrew Lih (Wikimedia Foundation), Filipa Calado (Pratt School of Information)

Moderator: Benjamin Zweig (Columbia University Libraries)

There is an inherent contradiction in the name of one of the world’s largest AI companies: OpenAI. OpenAI is neither “open” in its business practices nor in the way it uses open access data to train its models. But it is no secret that closed models are built upon open knowledge.

This panel explores the critical and evolving relationship between open access and open knowledge as it pertains to AI/LLMs. The panel will discuss the many possible costs – financially, communally, scholarly – of outsourcing open knowledge for profit. For example, are open knowledge communities, such as the Wikimedia community, being forced to rethink their contributions or the terms of their work? How can the open knowledge community and critical practitioners, from faculty to students, challenge what the author James Bridle calls in New Dark Age the danger of “automation bias” in the use of LLMs?