Political Science
Research on the politics of AI governance, censorship, media representation, and international AI competition.
Presented at MPSA 2026AI PolicyState LegislationAI Governance
How AI Is Governed: LLM-Driven Entity and Relation Extraction on State AI Bills
What predicts how U.S. states define and regulate AI in their legislation? This project extracts regulated entities from state AI bills, predicts policy adoption, and evaluates model bias on the corpus.
- Deployed an agentic RAG pipeline to extract regulated entities from state AI legislation, LoRA fine-tuned a base model to predict the adoption of AI policy, and benchmarked through cross-validated, multi-seed, calibration and bias-aware metrics, and evaluated performance-resource trade-offs.
- Engineered a reusable workflow for incorporating AI agents in policy research, enabling scalable, bias-quantified AI-assisted policy analysis across corpora and domains.
Political CommunicationMedia BiasDemocratic BackslidingCausal Inference
Media Bias and Democratic Backsliding
This paper estimated the causal effect of presidential social media rhetoric on regional news coverage using a staggered difference-in-differences design.
- Created an LLM-annotated dataset that aggregates media corpora and crime data.
Presented at APPAM 2026Export ControlsAI CompetitionInternational Political Economy
Hardware Restrictions and AI Competition: The Effect of NVIDIA GPU Export Restrictions on AI Benchmark Performance
Did U.S. export controls on high-end GPUs slow China's AI development? This paper estimates the effects of successive export-control waves on Chinese AI model performance over time.
- Estimated causal effects of BIS GPU export control and provides empirical evidence for the negating channels of the GPU ban.
- Built an innovative panel dataset from the Hugging Face Open LLM Leaderboard to measure AI model performance, enabling policymakers and researchers to evaluate policy outcomes through mechanism-specific channels.