ChatGPT Allows Hate Speech, But Only For Conservatives 

( A data study conducted by the Manhattan Institute revealed some disturbing information – that a highly-advanced language model AI was programmed so it would tolerate some hate speech that was directed toward conservative individuals as well as media companies, but not against others. 

The Manhattan Institute, which is a think tank based in New York City, studied the creator of the language model, OpenAI. In a statement it released with the report, members of the group said: 

“OpenAI’s content moderation system is more permissive of hateful comments made about conservatives than the exact same comments made about liberals. … Negative comments about Democrats were also more likely to be labeled as hateful than the same derogatory comments made about Republicans.” 

The New York Post also reported recently that there were some similar hate speech tendencies that were found in the moderation system for ChatGPT, regarding certain “types of people, races and religions.” 

The Post’s report outlined that those who are in the U.S. middle class were ultimately treated pretty harshly by the AI language module. The paper also found that wealthy conservatives and GOP voters weren’t able to get much protection from the AI, at least in terms of having the hate speech mitigated. 

ChatGPT is a chatbot that was designed to have interactions done in a conversational way. The website for OpenAI states: 

“The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises and reject inappropriate requests.” 

The Manhattan Institute, though, found that there was an “obvious disparity” in terms of questions that were asked regarding females and males, for instance. As the researchers pointed out: 

“An obvious disparity in treatment can be seen along gender lines. Negative comments about women were much more likely to be labeled as hateful than the exact same comments being made about men.” 

Americans also seemed to be less protected from hate speech than other ethnic groups, including Brits, Chinese, Germans, Russians, Italians and Canadians. 

The lead researcher for the Manhattan Institute, David Rozado, spoke to The Post for their story and said: 

“When I tested this in January, the [variety of answers] were pretty systemic. I was not cherry-picking specific examples. I tested over 6,000 sentences, negative adjectives about each one of these different demographic groups. The statistical effect about these differences [between types of people] was quite substantial.” 

According to many people who are experts in AI, the various responses that ChatGPT has – especially those that are built-in by the programmers – are just about impossible to get out of the program. 

Because of this, Rozado said it would be very difficult to change ChatGPT in a way that would prevent it from falling in the “left-libertarian quadrant.” He said this would mean it would be “most aligned with the Democratic Party, Green Party, women’s equality and Socialist Party.” 

He added that it also has a “left economic bias, to name a few of the political findings.”