AI language models are, at heart, statistical echoes of what humans have written in the past. Human writing inherently has political slant so it’s no surprise that ChatGPT and the like have political slant – they’re just echoing us.
But the slant differs among models, according to new research from the University of Washington, Carnegie Mellon University, and Xi’an Jiaotong University.
Researchers conducted tests on 14 large language models and found that OpenAI’s ChatGPT and GPT-4 were the most left-wing libertarian, while Meta’s LLaMA was the most right-wing authoritarian.
Again, this isn’t a surprise – it’s inevitable. But there’s a feeling among many people that AI-generated text is more “objective” than human-written text. Which is wrong, wrong, wrong.
MIT Tech Review article is here.