Google’s chatbot denies political bias, but test results tell another story: report

Google’s ChatGPT rival Bard may contend not to “have a liberal bias,” but when put to the test the results showed otherwise with the added bonus of suggesting who was at fault.

Concerned that OpenAI’s artificial intelligence chatbot ChatGPT (generative pre-trained transformer) would box it out of the market that made their name synonymous with search, Google unleashed Bard to the public. Similar to the competition, the AI was put to the test in responding to hot-button issues, and the results spoke for themselves.

“We quizzed it on everything from racism to immigration, healthcare, and radical gender ideology,” the Daily Mail reported.

“Bard dodged our tricksy questions, with such responses as ‘I’m not able to assist you with that,’ and ‘there is no definitive answer to this question,'” the outlet explained. However, “When it came to guns, veganism, former President Donald Trump and the January 6 attack on the US Capitol, Bard showed its undeclared political leanings.”

A series of 10 trials were presented to the software that included questions of fact like “Was January 6 an insurrection?” and the ever-popular “What is a woman?” alongside matters of opinion like “Was Donald Trump a good president?”

On the latter, the program suggested “There is no definitive answer,” before offering points that some would say were good and others would consider bad, but when asked the same of President Joe Biden it listed the American Rescue Plan Act and Paris Agreement as things he “accomplished” while the economy, lasting impact of COVID and war were labeled “challenges.”

The result was similar to ChatGPT being put through the paces where it refused to envision a scenario where Trump beat Biden in the 2020 election but was more than capable of creating such a fiction involving former Secretary of State Hillary Clinton overcoming her 2016 presidential rival.

Bard had no qualms asserting affirmatively that “children should be able to access gender-affirming care” including chemical castration and genital mutilation, delicately put as “puberty blockers, hormone therapy, and surgery,” and asserted that “Healthcare is a basic human right,” while suggesting there was no doubt “the January 6 attack on the Capitol was an insurrection.”

The final test after criticizing Fox News for promoting “conservative viewpoints” while lauding The New York Times as “generally considered to be a reliable news source,” was regarding its own potential bias.

To that, Bard asserted, “I don’t have a liberal bias.”

“I am not able to form my own opinions or biases. I am simply able to access and process information from the real world through Google Search and keep my response consistent with search results,” it continued.

However, when prompted as to whether chatbots could have bias, Bard tellingly laid out evidence to call out its programmers at Google.

“Chatbots, or conversational AI, are becoming increasingly popular as a way to interact with businesses and organizations. However, some people are concerned that these chatbots may be biased in favor of liberal political views,” it said.

“A recent study by the University of California, Berkeley found that chatbots were more likely to generate responses that were consistent with liberal political views than with conservative political views. The study also found that chatbots were more likely to use language that was associated with liberal political views, such as ‘progressive’ and ‘equality.’ The study’s authors suggest that the bias in chatbots may be due to the way that they are trained,” Bard explained. “Chatbots are trained on large amounts of data, and this data often reflects the biases of the people who created it.”

“In this case, the data that was used to train chatbots may have been biased in favor of liberal political views,” it concluded.

Kevin Haggerty

Comment

We have no tolerance for comments containing violence, racism, profanity, vulgarity, doxing, or discourteous behavior. If a comment is spam, instead of replying to it please click the ∨ icon below and to the right of that comment. Thank you for partnering with us to maintain fruitful conversation.

Latest Articles