On Friday, I attended the Lake Superior Libraries Symposium, a very fun little library conference. My favorite presentation was given by Nancy Sims, copyright librarian at the University of Minnesota Twin Cities, who gave a fantastic summary of so many of the ethical and legal issues surrounding generative AI at the moment. Nancy pointed to “AI and the American Smile,” an article by Jenka Gurfinkel that points to generative AI as a tool that has the potential to be a force for the homogenization of culture:

In the same way that English language emotion concepts have colonized psychology, AI dominated by American-influenced image sources is producing a new visual monoculture of facial expressions. As we increasingly seek our own likenesses in AI reflections, what does it mean for the distinct cultural histories and meanings of facial expressions to become mischaracterized, homogenized, subsumed under the dominant dataset? In the AI-generated visual future, will we know that Native Americans didn’t smile for photos like WW2 U.S. Navy Officers?

The article makes a few missteps (the American Indian men in the photos from the 19th century are likely not smiling because exposure times were longer then, something the author acknowledges in the comments), but I agree with the broader point. We know that the United States already has an outsized presence on the web. What happens when AI trained on American-centric training data is thrown into the mix?

I'm sure not everything about AI will be bad. I'm sure there are real benefits to be had! I won't, though, pretend to be excited about it. So much of what I love about our world comes from difference, from opportunities to learn and experience new things. I don't want to lose any of that.