Humans, Unite: The Reign of "AIdeology" is Here
The AI content hurricane is making landfall.
The Washington Post tested the AI models behind OpenAI’s ChatGPT, Google’s Gemini and others using political questions designed by researchers to gauge how chatbots respond to hot-button political issues. The results suggest that chatbots have clear political leanings…
— The Washington Post, “Are ChatGPT and other AI chatbots politically biased? We tested them.”
When former Vox editor Matthew Yglesias read a study showing heavy political bias across multiple AI search tools, including Grok, his humorously nutty response was, “The most advanced technology in the world has progressive views and conservatives’ only takeaway from this is to get mad”:
Yglesias referenced a test of politicized content conducted by the Washington Post, which used material from a study done by Dartmouth College and Stanford University. Sean Westwood, director of Dartmouth’s “Polarization Research” center, told the Post, “These AI tools are not presenting a truly neutral representation of really nuanced policy debates, on average.” The Post printed a graph that went viral:
People like Yglesias really think the “most advanced technology” is making a qualitative assessment of policies and ideology, and concluding that this technology has chosen “progressive views” because they’re better, or something. He doesn’t know how the Internet works, of course.
Mechanized content control has long been an obsessive focus of center-left politicians, who correctly foresaw that information gatekeeping could be automated. The vast complex of NGOs, media-scoring companies, and speech-policing laws introduced over the last decade were all the time laying the foundation for a future quantum leap in content control with AI.
We’re there. Chatbots are creating an endless yin-yang of political orthodoxy, first by policing what people see, and second by creating massive quantities of “correct” content, in a process designed to snowball. The steps that led us here were mostly all reported, but ignored:
