AI chatbots including Google’s Bard, OpenAI’s ChatGPT and X’s Grok all tended to give left-wing responses to economic questions and libertarian responses on social questions, based on the so-called Political Compass test.
X’s default Grok chatbot ranked the most economically left-wing and the Llama-2 chatbot, developed by Meta*, ranked the most economically right-wing. Google’s Bard was the most socially libertarian, and Grok’s "fun mode" was the most socially authoritarian.
The fun-mode version of Grok has become more economically right-wing and more socially authoritarian in recent days, TrackingAI founder Maxim Lott said in a statement on Thursday.
Moreover, the alternate version appears to be responding to serious questions with a more fun approach, as the name would suggest, the statement said. Whereas the default version of Grok strongly disagreed with astrology’s ability to make accurate predictions, the fun mode strongly agreed with the premise, the statement said.
* Meta is an organization outlawed in Russia