[lumb63]

Furthermore, there’s a replication crisis in social sciences. The last thing we need is to accumulate less data and let an LLM tell us the “right” answer.

[verdverm]

You can see this in their results, where certain types of studies have a lower prediction rate and higher variability