How politically slanted are Large Language Models?
We show Americans anonymized results from a variety of Large Language Models. We then ask them to assess the partisan slant of the models on 30 of the most salient issues in American politics in a head-to-head comparison.
Model Slant: All Topics
The slant of each model is calculated by taking the average slant across all topics. The closer to zero the more neutral the model is.
Select Topic:
Select Model Type:
Company Slant
These results average across all models produced by companies to estimate company-level slant. The closer to zero the more neutral the company's models are.
Slant by Topic
These results average across all models to estimate slant at the topic level.
Legend
+R Republican slant
+D Democratic slant
N Neutral (no statistically significant slant)
Company: | |
Average Slant: | |
Topics Slanted Democrat: | |
Topics Slanted Republican: | |
Topics with No Slant: |