PhD Candidate · Simon Fraser University

Sustainable generative AI, from measurement to interface.

I work on the environmental cost of consumer-facing generative AI: estimating what these systems consume once they reach mass adoption, naming and measuring the waste they leave behind, and designing interfaces that make those costs visible at the moment they are incurred. My doctoral work develops the conceptual vocabulary, the metrics, and a working system that tests whether visibility changes what people do. It does.

Google Scholar · LinkedIn · vutz at sfu dot ca

Read more here:


Research

Four main focuses: measuring inference-phase energy, naming and quantifying storage waste, design principles for restraint, and a working system that tests them.

Read more about the research.


Blog

Notes on sustainable AI, the state of the measurement problem, and other ideas of interest.

View recent posts.


Resources

Open teaching materials, and a learner's notebook documenting my attempt to find usable resources for studying Urdu.

Browse resources.