Summary
Keng-chi Chang is an Assistant Professor of Quantitative Social Science who studies how AI systems reshape information access, political communication, and behavior at scale, blending multimodal ML, causal inference, and large-scale digital trace analysis. He holds a PhD in Computational Social Science and an MS in Computer Science from UC San Diego and has published in PNAS, Journal of Communication, and NeurIPS workshops, with coverage in The Economist. His technical work spans PyTorch-based VLMs, LoRA fine-tuning, LLM interpretability, and production ML pipelines that processed over 1M social media images. Recent projects audit short-form video recommendation algorithms and develop methods to detect visual treatments in AI-generated content, reflecting a rare combination of platform algorithm auditing and experimental survey design. With 11 years of experience across academia and industry, he brings both rigorous causal thinking and production engineering to questions of AI safety, trust, and moderation.
11 years of coding experience
University of California, San Diego
Visiting Graduate Student, Economics, Visiting Graduate Student, Economics at University of Wisconsin-Madison
Bachelor of Arts - BA, Economics, Bachelor of Arts - BA, Economics at National Taiwan University
English, Chinese