MIT Working Group on Scholarly Content and Generative AI

As Generative AI evolves at a rapid pace, members of the MIT community are seeking guidance on the appropriate use of scholarly content in the training of GenAI models. Some scholars at MIT are concerned about use of their own content for GenAI, such as by tech companies who use content (with or without permission) to train commercial GenAI tools, while others are seeking to use scholarly content from various sources in the development of AI models as part of their research. MIT’s publishing entities (MIT Press, MIT Tech Review and the Sloan Management Review) and other stewards of MIT content (e.g. the Libraries and Open Learning) are wrestling with questions about whether and how to provide access to their content for GenAI training. Often, MIT’s publishing entities have priorities that differ from the rest of the Institute, in terms of a desire to monetize content that is otherwise being taken for free, without permission. Along with others across the Institute, publishers worry that GenAI is cannibalizing science journalism and scholarly publishing. These concerns are shared by third party publishers who frequently take this position–attempting to restrict access to content for training use–in contradiction to the Institute’s commitment to the open sharing of information and ideas, as embodied in the work of Open Learning and the Faculty Open Access Policy. Many at the Institute support the use of scholarly content in AI models, so that AI might be trained on trustworthy texts and ideas, rather than only on unvetted content from social media and the like. This misalignment is just one example of the multitude of different perspectives and competing priorities on campus.

To provide common understanding of the issues at stake and the current legal and technical landscape, and to articulate an initial set of resources, guidance and principles for MIT community members, the Provost, VPR and OGC are convening a working group on Scholarly Content and Generative AI. The working group will be co-chaired by Chris Bourg, Director of Libraries, and Claire Schneider, Counsel, OGC.

The group will address the following topics:

• Rights and resources for MIT authors whose scholarly content might be included in Generative AI training models.
• Legal, ethical and normative factors related to the use of MIT owned, licensed, and/or published content in GenAI training models.
• Rights, responsibilities, and guidelines for MIT scholars seeking to use scholarly content to train Generative AI models.

The group will produce the following outputs:

• A list of administrative offices that can assist with questions that pertain to the topics above.
• Initial white paper describing the issues and competing perspectives and priorities, as well as the current legal landscape, and providing a list of sources of information for the MIT community.
• A set of recommendations to guide MIT policy and practices related to the use of scholarly content for Generative AI/LLMs. We expect that any advice the working group provides to the MIT community will need to be flexible as the GenAI industry and legal landscape evolves.

• Chris Bourg, Director of Libraries (co-chair)
• Claire Schneider, Counsel, Office of the General Counsel (co-chair)
• Amy Brand, Director and Publisher, MIT Press
• Christopher Capozzola, Elting E. Morison Professor of History and Senior Associate Dean for Open Learning
• Catherine D’Ignazio, Associate Professor of Urban Science and Planning
• Dylan Hadfield-Menell, Bonnie and Marty (1964) Tenenbaum Career Development Assistant Professor of Electrical Engineering and Computer Science
• Graham Jones, Professor of Anthropology
• Yoon Kim, Assistant Professor of Electrical Engineering and Computer Science
• Ruth Lehmann, Professor of Biology and Director of the Whitehead Institute
• Meghan McCollum Fenno, Counsel, Office of the General Counsel, and Executive Director, Office of Strategic Alliances and Technology
• Susan S. Silbey, Leon and Anne Goldberg Professor of Humanities, Sociology and Anthropology, and Professor of Behavioral and Policy Sciences
• Eric So, Sloan Distinguished Professor of Global Economics and Management
• Lily L. Tsai, Ford Professor of Political Science
• Katie Zimmerman, Director of Copyright Strategy, MIT Libraries