by Jonathan Woon
21 February 2024
Once we clear through the market-driven hype around AI, I see ChatGPT—and other increasingly powerful generative AI tools—as indeed truly transformative technologies. Where some see a threat, I see enormous potential for good. How did I get here?
As with many colleagues in academia, the threat to academic integrity was obvious. In early 2023, after ChatGPT was first released to the public, I gave it a try and found that it could write a perfect job description and compellingly answer the kinds of questions I might ask on a political science exam—much more polished and cogent than many of the blue books I’ve read. Other than a few fantastic, highly informative presentations in our governance councils, I didn’t think too much more about it for months.
Fast forward to July. I attended a workshop in which Betsy Barre (Wake Forest Center for the Advancement of Teaching) demonstrated how ChatGPT could—with an extensive, carefully crafted prompt—write a complete curriculum for a philosophy major, with learning outcomes, credit requirements, and advanced electives. To my surprise, those electives included Asian and Comparative Philosophy, Feminist Philosophy, and Environmental Philosophy, which I did not expect given my preconceptions about a philosophy major. Impressive!
Curiosity piqued, I devoured Ethan Mollick’s Substack newsletter, and I was intrigued by the possibility of using ChatGPT as a writing tutor rather than a ghost writer. Having learned more about prompt crafting, I began experimenting with interactive uses:
- As an interactive encyclopedia to re-learn long-forgotten topics and explore new ideas: Explain the allegory of the cave. How is it relevant to understanding politics today? What is the metaphor about foxes and hedgehogs? How has this been elaborated in contemporary philosophy and psychology?
- As an interactive dictionary and thesaurus to help me find the right word or understand nuances of usage: Fill in the following sentence with a word like transactional or instrumental: [sentence with blank]. Provide 5 words, explain their meaning, and provide suggestions about appropriate usage.
- As a discussion partner to brainstorm, refine ideas, generate acronyms (GAINS). Almost as good as talking to colleagues and collaborators. Definitely opens up a lot more possibilities than talking to myself—my AI partner is trained on a huge amount of data and traverses a network of topics at lightning speed.
More recently, I’ve experimented with using ChatGPT for coding and data analysis, also relying heavily on interactive use to accomplish the following:
- To update my website from a sad mid-2000s HTML site to a sleek, modern Jekyll-based one.
- To quickly make visualizations: “Make two density plots, one for the 100th Congress and another for the 110th Congress. For each Congress, use separate densities for each party overlayed on the same plot region. Make Democrats (blue) and Republicans (red). Use transparency so that you can see where the densities overlap.” Far easier to give natural language instructions than to figure out the correct R/Python/Stata syntax.
- To create simple web apps: an interactive version of the congressional density plots, an app to demonstrate an analytical model of ideological candidate competition, and an app to visualize the implications of risk preferences on voting decisions. Bespoke tools for teaching and research can be built in a few hours with high level conceptualization, natural language, and modicum of coding knowledge.
I started 2023 with a limited imagination for how generative AI could be used by scholars and educators in positive ways. Through exposure to models of creative and innovative uses, combined with the insights of many of our colleagues here at Pitt (Kirill Kiselyov, Annette Vee, Alison Langmead among them) and through trial and error exploration, I began 2024 with optimism and excitement about the power of AI to boost productivity, creativity, and communication. To be sure, this excitement is tempered by serious concerns about cheating, bias, intellectual property, confabulation, misinformation and various other downside risks we haven’t yet imagined. Yet, as MIT economist David Autor explained on the Planet Money podcast, there is also great potential for AI to be a tool for equity.
I am incredibly excited that we’re launching the Dietrich GAINS project. By establishing a framework through which graduate faculty and students can share advice and wisdom, we aim to cultivate a community of practice that not only advances AI literacy but also champions consideration of how artificial intelligence can be responsibly used to enhance research and pedagogical practice across disciplines.