There appears to be two distinct camps gathering steam as AI coding agents get increasingly adept at generating software on the basis of some light natural language specifications: those who want to maximally remove the ‘error-prone’ sentient bags of mostly water at the keyboard and those who stubbornly assert their own relevance as their AI-enhanced coding interfaces run off and make significant (but also often not-quite-right) contributions to the code base without even checking in. I fall into the latter camp, and in this talk, I’ll make the case for why the former camp is completely wrong—as long as we invest in designing the interfaces necessary to give our sentient water bag selves meaningful oversight and agency. Perhaps even more importantly, these interfaces might help preserve the meaning that brought us to this field in the first place: building up (rather than eroding) our understanding of the systems, languages, and software applications that fascinate us when our context allows for or demands it.

Elena Glassman is a the Stanley A. Marks and William H. Marks Assistant Professor at the Radcliffe Institute and a professor of computer science at the Harvard Paulson School of Engineering and Applied Sciences specializing in human-computer interaction. She designs, builds and evaluates systems for comprehending and interacting with intelligent systems and population-level structure in large code and data corpora. Prior to joining Harvard as an assistant professor, Glassman earned BS, MEng and PhD degrees from MIT EECS and served as a postdoctoral scholar at the University of California, Berkeley EECS department and Berkeley Institute for Design.