overview
My AI research draws on my background in Human-Computer Interaction, taking a user-centered perspective on artificial intelligence. This includes developing novel techniques for human-AI interaction, harnessing the power of AI for socially beneficial uses (including AI for accessibility systems that augment users' ablities), and studies relating to ethics and fairness topics, including the impact of AI on society. As of 2026, I am focused on establishing the new field of Post-AGI Interaction Research, the aim of which is to develop sound methods for understanding the range of ways in which people might interact with advanced levels of Artificial General Intelligence.
research projects
AI for Accessibility
- Perspectives on Alt Text for AI-Generated Images [ CHI 2024 paper]
- Context Matters for Image Descriptions for Accessibility [ EMNLP 2022 paper ]
- Automatic Structure Extraction in Slide Decks for Non-Visual Access [ CHI 2023 paper ]
- SpeakFaster project [ CHI 2023 Case Study ]
- LaMPost writing assistant for people with dyslexia [ ASSETS 2022 paper ] | [ CACM article ]
- Designing an Online Infrastructure for AI Data Collection from People with Disabilities [ FaaCT 2021 paper ]
- Sign Language Recognition, Generation, and Translation [ ASSETS 2019 paper ]
- Combining Image Recognition and Eye Gaze to Support Conversation for People with ALS [ DIS 2017 paper ]
AI Ethics and Fairness
- The Ethics of Advanced AI Assistants [ DeepMind technical report ]
- Beyond Safety: Toward a Value-Sensitive Approach to the Design of AI Systems [ NeurIps 2022 workshop paper ]
- Understanding the Representation of Age in AI Data Sets [ AIES 2021 paper ]
- Designing Disaggregated Evaluations of AI Systems [ AIES 2021 paper ]
- AI & Accessibility: A Discussion of Ethical Considerations [ CACM article ]
- Disability, Bias, and AI [ AI Now whitepaper ]
- Toward Fairness in AI for People with Disabilities [ ASSETS 2019 Workshop ]
Interaction with Generative AI
- Generative Ghosts [ CHI 2025, arXiv pre-print ]
- Evaluating Anthropomorphic Behaviors [ arXiv pre-print ]
- Prompting Considered Harmful [ Communications of the ACM ]
- Interactive Alignment [ arXiv pre-print ]
- Bi-directional Alignment [ arXiv pre-print ]
- Can Language Models Use Forecasting Strategies? [ arXiv pre-print, June 2024 ]
- Generative Agents [ UIST 2023 paper ] [ Nov. 2024 arXiv pre-print of synthetic social science study ]
- Social Simulacra [ UIST 2022 paper ]
- Generative AI for Scientific Acceleration [ arXiv pre-print ]
- LaMPost: A Generative Writing Assistant for People with Dyslexia [ ASSETS 2022 paper ]
- LaMDA [ arXiv pre-print ]
- The Design Space of Generative Models [ NeurIps 2022 workshop paper ]
- Context-Aware Abbreviation Expansion using Large Language Models [ NAACL 2022 paper ]
Human-AGI Interaction
- Levels of AGI [ ICML 2024, arXiv pre-print ]
- AGI is Coming... Is HCI Ready? [UIST 2023 Visions Keynote]
- HCI for AGI [ Interactions 2025 ]
- Characterizing Jaggedness of AI Models [ preprint, Jan. 2026 ]
- A Cognitive Framework for Measuring AGI Progress [ tech report, March 2026 ]