GVPT Professor Jóhanna Birnir, GVPT Ph.D. alumnus, Henry David Overos, current GVPT doctoral candidates, Ojashwi Pathak, Harriet Goers, Jordan Gouws-Dewar, and their colleagues recently published a new PNAS Nexus paper entitled "Coding with the machines: machine-assisted coding of rare event data."

Birnir and her team compared GPT with other machine-learning models and trained human coders to code data on protests by various ethnic groups. The study found that while some machine-learning models performed better than others, it may not be wise to solely rely on human coders for specific projects, even with intensive training. The findings suggest that using AI models like GPT in conjunction with human coders could enhance confidence in research findings. This research has promising implications for the potential of AI in assisting social change by providing more timely and accurate data analysis.

Read more here: Study: Can AI Reliably Accelerate Research?