I am a second year PhD student at the University of Utah Kahlert School of Computing. I am a member of UtahNLP, where I am advised by Prof. Ana Marasović. I primarily study where and why AI models diverge from human behavior, and how to use these differences to understand the underlying mechanisms of how models reason better. I explore this problem through the development of benchmarks which test models performance on reasoning tasks compared to humans, and through interpretability methods, to translate model behavior into insights that people can understand.

From January to May 2026, I was a research intern at Handshake in San Francisco, where I worked on predicting the outcomes of LLM agent trajectories before they finish.

Outside of research, I spend most of my time rock climbing, running and skiing in the mountains around Salt Lake City. You can check out what I’m up to in climbing on 8a, and running on Strava.

Publications

Divergent Paths: Learning an Outcome-Predictive Representation of LLM Agent Trajectories
Alexander Gill, Guram Gogia, Francisco Guzmán
Under review, AAAI 2027

The Hard Part Comes After Search: Benchmarking Web Agents on Synthesizing, Organizing, and Displaying Knowledge
Alexander Gill, Md Farhan Ishmam, Xuyen Nguyen, Neha Bhat, Parker Henry DeYoung, Fateme Hashemi Chaleshtori, Nathan Stringham, Kenneth Marino*, Ana Marasović*
EMNLP Findings 2026

What Has Been Lost with Synthetic Evaluation?
Alexander Gill, Abhilasha Ravichander, Ana Marasović
EMNLP Findings 2025

On Evaluating Explanation Utility for Human-AI Decision Making in NLP
Fateme Hashemi Chaleshtori, Atreya Ghosal, Alexander Gill, Purbid Bambroo, Ana Marasović
EMNLP Findings 2024