A study by David Rozado at Otago Polytechnic University found that AI language models lean towards progressive and democratic ideologies. The study showed that the training data influences the political leanings of AI models.
Key Points
AI language models lean towards progressive and democratic ideologies
Training data influences the political leanings of AI models
Pros
Provides insights into the political leanings of AI language models
Raises awareness about AI bias and its potential impact on public opinion
Cons
The study cannot definitively determine the source of political preferences in AI models