Dear Colleagues,
Last winter, I wrote to announce that faculty-led working groups would be appointed to develop frameworks and make proposals for the University’s approaches to computational and statistical thinking and the emerging concepts and tools of machine learning and artificial intelligence (AI). Following that announcement, the provost and I, in close consultation with the Committee of the Council, appointed three working groups: one on topics related to instruction; one to identify strategic opportunities to advance world-class scholarship across diverse areas of interwoven application and development of these tools; and one to advise on possible organizational structures that will ensure that the foundational disciplines of statistics, mathematics, and computing are configured optimally for this university to make the strongest contributions. I write today to update you on each of these important developments and to preview next steps.
First, the advisory group on AI and education has issued its report. I believe it is thoughtful and worthy of your careful consideration. At the highest level, the report affirms that our educational mission is one driven by human expertise, human engagement, and human interaction. From that premise, the report then urges faculty to construct pathways for students that will guide them to engage with AI and large language models (LLMs) in ways that are ethical, skeptical, and ambitious. I am writing to students today to reinforce these points. If our community of instructors follows the recommendations of this working group, faculty will develop course policies about AI use that hold for their individual courses and that also resonate with broader curricular offerings. These efforts follow discussions at local levels among those who are responsible for Core sequences and degree programs. That approach allows for both customization and consistency. The report also recommends that the lenses of computational and statistical thought be explicitly addressed in the Core curriculum of the College and that opportunities to become conversant in the underlying concepts and tools will be made available to students from any field of study. The implication is that to be educated today also means having an understanding of how a great thinker might frame problems creatively with these perspectives and how to be skeptical of assertions in substantive ways. To promote pedagogical innovation on these topics, we have supported twelve projects in a wide variety of instructional contexts that seek to either expand and leverage machine learning and AI use or to limit it deliberately. The net effect of this approach is that we collectively can help our students become excellent at both thinking with machines and at thinking without them.
The second faculty-led advisory group worked to help us identify a set of cross-disciplinary projects to support that will advance the interests of faculty in distinctive UChicago fashion. We received more than 40 proposals for such projects and today we announce ten that are going forward. Together, they leverage many of UChicago's existing core strengths and target opportunities where the University can establish itself as a distinctive leader in this highly competitive field. Here is the list:
Culture and Creativity in an AI-Empowered Society
AI & Human-Environment History
Human Machine Creativity
AI for Resilient and Adaptive Societies
Jobs and Prosperity Impacts of AI
AI Innovation in Markets and Governance
AI Models of Climate and Sustainable Growth
Learning the Rules of Life and the Universe
New Forms of Socio-Cognitive AI
Science Labs that Only AI Can Build
From Cells to Organs With AI
AI in the Service of Therapeutics
AI-Driven Cancer Drug Discovery
AI-Based Biological Design
Their scope is substantial and reflects the broad applicability of the new tools to aid in understanding and discovery. These projects will help inform our work to secure resources for similar activities across the University in the years ahead.
The third group was asked to consider the path forward for computational and data sciences and AI, including the possibility of a new academic unit. That group has engaged faculty across campus. Among other questions, they have assessed the intellectual case for computational, mathematical, statistical, and data sciences, and AI as fields that together will yield deep conceptual breakthroughs for a long period of time; have deliberated on the advantages and disadvantages of forming a new decanal unit for advancing scholarship in these fields; and have been challenged to think about how fields can be organized in order to leverage the distinct features of the UChicago environment and lead in scholarship and education. The group is completing a final report that will be made available and will inform discussions with the Council of the Senate and other faculty groups in the months to come.
I conclude this note to you with some reflections of my own on the nature and scope of the developments in machine learning and artificial intelligence. I see the advent of these new concepts and tools as a major development that presages further advances and that will have profound implications for creators of knowledge and understanding for decades to come.
Faculty leading in the understanding of financial markets, providing new means of developing understanding of ancient cultures, and enhancing human creativity all see opportunities to ask questions in new ways that suggest there are even deeper ways of thinking that will be created in the years ahead, not just superficial use of the latest tools. In areas where I am active as a scholar, I can see this quite clearly.
In the early 1900s, the physicist Ernest Rutherford famously quipped words to the effect that there are only two branches of science: physics and stamp collecting. More than one hundred years later, he would likely be unsettled by the confidence of the hierarchy he was asserting. Today, the use of large datasets, stamp collecting, if you will, increasingly sits alongside the construction of equations to model the physical world as a complement to discovery. Models often oversimplify, while rich datasets studied with machine learning and other AI-based methods can uncover previously indiscernible correlations that challenge and inform models with nuance and relevance. The same can be said of pretty much every discipline that seeks to model an aspect of the world with equations, economics being a clear example. The power of the new modes of thought extends to include anything that can be sensed or represented symbolically. Indeed, as the list above of just a few of the projects our faculty are advancing shows, computational and statistical thinking tools are continuing to develop, and these new approaches are deeply attached to nearly every part of knowledge creation and understanding from the humanities to the social sciences to natural science writ large and to the professions represented here.
While we have seen some early demonstrations of the power of these new methods, they are still very far from being fully elaborated. It seems clear that deep conceptual advances are still to come. This—to me—suggests that we should grapple with the educational, research, and organizational developments that are commensurate with the opportunities. The University of Chicago, justifiably famous for being a pacesetter in how to think, has a great calling to organize itself to promote excellence in how to think with machines as well as to maintain its strengths in how to think without them. I look forward to this year's many advances and deliberations on these matters as we chart our course.
All the best,
Paul
--
Paul Alivisatos
President