CSE Colloquium: Co-Active AI-Assisted Programming

Presenter: Nadia Polikarpova, UCSD

Abstract:

AI-assisted programming has rapidly moved from novelty to default. Today, most developers use AI coding tools, and increasingly rely on agentic systems capable of making multi-step design and implementation decisions with minimal human guidance. While these systems boost productivity, they also introduce new risks: developers may disengage from the reasoning behind generated code, leading to shallow understanding, loss of ownership, and what is increasingly described as cognitive debt.

In this talk, I argue that AI-driven software development must be co-active: humans and AI should remain continuously engaged in a shared process of understanding and decision-making. I will present two complementary research directions toward this goal. The first focuses on observability—helping developers understand, validate, and trace the behavior of AI-generated code. The second focuses on controllability—making AI decisions explicit, persistent, and steerable. Together, these ideas restore programmer agency while maintaining the productivity gains of AI-assisted development.

Bio:

Nadia Polikarpova is an associate professor at UC San Diego, and a member of the Programming Systems group. She received her Ph.D. in Computer Science from ETH Zurich in 2014, and then spent some time as a postdoctoral researcher at MIT. Nadia’s research interests are at the intersection of programming languages, AI, human-computer interaction, and social computing.

Hosted by: Professor Nikos Tziavelis

Location: Engineering 2, Room E2-180 (*Refreshments such as coffee, tea, pastries, and fresh fruit will be available.)

Zoom: https://ucsc.zoom.us/j/93445911992?pwd=YkJ2TQtF79h0PcNXbEcpZLbpK0coiY.1&jst=3