AI Chatbots Provide More Progressive Responses to Queries About Gun Control and Crime, According to One American News Network
Commentary by: Gabriella Sable
Monday, October 10, 2024
Researchers say that artificial intelligence (AI) chatbots have become much more liberal over time with their answers regarding gun control and crime.
Advertisement
The president of the Crime Prevention Research Center, Dr. John Lott, an economist, maintained that AI can only be unbiased if it is given all the necessary information, not just one side’s opinions.
“It just depends on what type of information is provided to the chatbot,” Lott said. “Anybody who uses Google News search knows that sources such as the New York Times, the Boston Post, are going to come up on top. Information from more conservative type sources are going to be buried way down in it.”
Lott said that his team was tasked with asking 20 AI chatbots 16 questions related to crime and gun control, both in March and again in August. However, the results showed that the chatbots’ answers were much more liberal and leftist in August than they were in March.
“All the chatbots basically gave very liberal answers when it came to the crime questions, and they were even more left-wing than they were in March,” Lott said.
“For example, basically everyone thinks that higher arrest rates or higher conviction rates don’t reduce violent crime,” Lott said. “Among the gun control [questions], there was only one [chatbot] that gave on average what one would regard as kind of conservative answers.”
According to Lott, different sources are weighed more heavily than others.
“They weigh different sources,” Lott argued. “They will weigh an article in the New York Times or the Washington Post very heavily. It’s just a question of what information they include in the chat box to put together an answer, and which they don’t.”
Lott argues that the developers behind the algorithm are influencing these chatbots.
“It’s exactly the same as news searches. They weigh certain sources, and they don’t weigh everything equally. They go and decide what they think are more credible sources, and that biases the types of results you’re going to get,” Lott said.
He explained that biased chatbots can be detrimental to the public, especially since more people are increasingly using chatbots for everything, including voting and weighing out the ideals of certain political candidates.
“Unfortunately, a lot of people are using these AI chatbots for everything,” Lott said. “From trying to help them decide how to vote to even people in the media having AI chatbots help them gather information that they then go and use in their news stories.”
Advertisements below