Holly Huey.

Applied Scientist @ Adobe | Experimental Psychology, PhD | Improving GenAI technologies to enhance human creativity

At Adobe, I'm an Applied Scientist working to improve GenAI image & video systems. I earned my PhD in Experimental Psychology from UC San Diego, under Dr. Judith Fan @ The Cognitive Tools Lab.

My research statistically compares human vs. model behavior to improve creative technologies related to: text-to-visual & visual-to-visual systems, semantic editing, object recognition, and user intentions

I specialize in how people create and interact with visual media by developing large-scale benchmarks to compare human vs. model behaviors. I do this by generating datasets of 10-100K human responses in order to systematically evaluate people's strategies for using and editing images and videos. This scale of data enables me to statistically measure improvements and regressions in GenAI models (e.g., text-to-image). I work with globally diverse professional photographers & editors, graphic designers, and video creators/editors to better understand human-AI collaboration.

I also collaborate closely with AI Ethics teams to ensure that models do not create or enable the creation of harmful content.

My disseration research focused on understanding how user intentions impacts visual production behavior and downstream interpretation of that content. I evaluated these questions by developing and analyzing large-scale drawing, diagram, and data visualization datasets. During my PhD, I also completed three internships with Adobe Research. My projects developed and improved parameters for video editing LLMs, automatic B-roll augmentation, and model reasoning.

research domains

VISUALIZATIONS

My dissertation research evaluated human visual content creation in the form of drawings, diagrams, and data visualizations to investigate how people interpret objects and scenes.

IMAGES

To systematically evaluate text-to-image model errors, much of my work involves developing behavioral tasks to crowdsource human annotators and generate quality scores for large-scale image datasets.

VIDEO

At Adobe Research, I interviewed video editors and directors to develop model parameters based on user needs and translated those insights into online experiments for quantitatively testing video editing models.
I have also published research in other domains: language, social pragmatics (theory of mind), navigation, causal perception, symbolic reasoning, and developmental psychology. See my publications below or my Google Scholar for papers in these areas.

Designed by Holly Huey. Please excuse any coding errors. I hand-coded this in my first year of grad school to teach myself web programming and am now paying for the sins of my poor coding practices from 2019...