Artificial Intelligence Pedagogy Must Incorporate Interdisciplinary Exposure to Ensure Safety
Why Training in Economics, Political Science, and Other Fields Is Essential for AI Researchers to Responsibly Anticipate the Societal Consequences of Their Research
I have noticed that, when computer scientists are asked the potential impacts of artificial intelligence, they rarely use frameworks from economics, political science, sociology, etc. Instead, they will give answers roughly comparable to what someone with no academic training in the relevant fields would give, e.g., ‘AI models will be able to replace receptionists at paper companies’, etc.
Prima facie, I would expect exceptionally well-trained computer scientists to be able to situate their advancements into the broader ecosystem of our world, analysing, for example, how artificial general intelligence would impact the environment using a framework informed by political-science-theoretic models of green energy production, e.g., ‘AGI models would cause less environmental degradation in a world where states are able to engage in popularly disfavourable, but expert-supported, activities—fission production in this case.’
My guess is that this derives from the currently dominant pedagogical models we have from American-led education [1], where students are siloed into their discipline, with limited exposure to other fields. Without my informal education, for example, I would personally not be able to situate my research in artificial general intelligence within a macroeconomic framework [2].
I am very inclined towards the current trend for more interdisciplinary work. More specifically, I would strongly encourage computer science departments to require early-stage undergraduates interested in artificial intelligence to take coursework in economics and political science, focusing primarily on the fundamentals of macroeconomics and the fundamentals of democratic institutions, respectively. Do we really want our leading computer scientists to lack an understanding of how their research impacts our broader social ecosystem? Do we not want Dr Frankenstein to understand how his Adam would affect the townspeople?
[1]: Levine, Emily J. 2024. ‘Research & Teaching: Lasting Union or House Divided?’ American Academy of Arts & Sciences, 30 May 2024. https://amacad.org/publication/daedalus/research-teaching-lasting-union-or-house-divided.
[2]: Though I emphasize that my ability to do so is a fraction of what someone with doctoral training in macroeconomics would be able to accomplish.
Subscribed