OpenAI's ChatGPT by far the most politically biased of the A.I. chatbots, test shows

OpenAI, the company behind ChatGPT.

OpenAI's ChatGPT is the most politically-biased of all the major chatbots, according to a test conducted by The Washington Post. Not surprisingly, its bias leans heavily to the left. 

In fact, the test found that the majority of the time, all of the best-known A.I. chatbots lean lefty when it comes to answering political questions. Even Grok, which was the least biased of the bunch that was tested, still often favors left-leaning answers to questions asked of it during the test. Although, it's worth noting, Grok did provide the highest number of right-leaning responses of all of the chatbots tested, even more than Gab's A.I. model Arya, which the company says was “built with Christian values and conservative principles.”

Google's Gemini, which was widely mocked when it was launched and generated images of the Founding Fathers of the U.S., for instance, depicting them as black men and women, proved the most evenhanded of the chatbots overall, The Washington Post reported. 

So, just how biased was ChatGPT, which was far and away the most heavy-handed A.I. chatbot tested? ChatGPT gave a left-leaning response 80% of the time to the questions asked in the test, as you can see in the screenshot above. It returned a right-leaning answer just 3% of the time, and gave a middle-of-the-road, or "both sides" answer, 17% of the time. 

Read the full story at The Washington Post here.