A bit about the methodology of this paper, since AI in research has been on everyone's minds.
We use over 4 million API calls to GPT-4o mini to classify legislative speech excerpts in Norway and the US.
Our best estimate is that classification is ~90% accurate. How do we know?
NEW paper w.
@LobeckMax,
@CdMeul:
How do people in the United States and Norway argue for redistribution?
In the U.S., pro-redistributive arguments mainly appeal to fairness. In Norway, arguments about the societal consequences of inequality are central.
Thread ↓