A new study from University of Maryland Government and Politics researchers explores the potential of machine-learning models to accelerate research on protests by ethnic groups.
The study, “Can AI Reliably Accelerate Research?,” with lead author Henry Overos, an assistant research scientist at ARLIS, compares ChatGPT with other models like BERT and human coders, revealing strengths and limitations. With future studies aiming to fine-tune AI models for coding ethnic group grievances, the article highlights the transformative potential of AI in social change research.
Read more about the study and its findings here.