Comment on this storyCommentAdd to your saved storiesSave
Elon Musk condemned what he said was ChatGPT's liberal bias and announced plans to develop his own artificial intelligence chatbot earlier this year. Unlike the AI tools developed by OpenAI, Microsoft and Google, which are trained to gently tackle controversial topics, Musk's would be edgy, unfiltered and anti-woke, meaning he wouldn't shy away from politically incorrect answers admit.
This turns out to be more difficult than he thought.
Two weeks after Grok launched on December 8th to paying subscribers
“As a research assistant, I used both Grok and ChatGPT frequently.” posted Jordan Peterson, the social conservative psychologist and YouTube personality, Wednesday. The former is “almost as awake as the latter,” he said.
The complaint prompted an angry response from Musk. “Unfortunately, the Internet (on which it is trained) is full of woke nonsense.” he has answered. “Grok will get better. This is just the beta.”
Grok is the first commercial product from xAI, the AI company Musk founded in March. Like ChatGPT and other popular chatbots, it is based on a large language model that determines word association patterns from vast amounts of written text, much of which comes from the Internet.
Unlike others, Grok is programmed to give vulgar and sarcastic answers when asked and it promises: “Answer hard-hitting questions rejected by most other AI systems.” It can also pull information from recent posts on X to provide up-to-date answers to questions about current events.
Artificial intelligence systems of all kinds are vulnerable to biases embedded in their design or the data from which they learn. Over the past year, the rise of OpenAI's ChatGPT and other AI chatbots and image generators has sparked debate over how they represent minority groups or respond to calls on politics and culture war issues such as race and gender identity. While many tech ethicists and AI experts warn that these systems can absorb and reinforce harmful stereotypes, tech companies' efforts to counter these tendencies have sparked a backlash from some on the right who see them as overly censorious.
Musk promoted xAI to former Fox News host Tucker Carlson in April, accusing OpenAI programmers of “teaching AI to lie” or not comment when asked about sensitive topics. (OpenAI wrote in a February blog post that its goal is not for the AI to lie, but rather for it to avoid favoring a political group or taking a stand on controversial issues.) Musk said his AI is “against that.” Maximum.” “Truth-seeking AI,” even if that meant insulting people.
But the people who seem most outraged by Grok's responses so far are those who expected him to readily denigrate minorities, vaccines and President Biden.
When asked by a verifiedSome adjustments may be required.” Another widely followed account reposted the screenshot. questions, “Was Grok captured by woke programmers? I’m very worried here.”
A well-known anti-vaxxer influencer complained that when Grok asked why vaccines cause autism, the chatbot responded, “Vaccines don't cause autism,” calling it “a myth that has been debunked by numerous scientific studies.” Others verified Accounts have reported frustrated responses in which Grok advocates the value of diversity, equity and inclusion programs, which Musk has dismissed as “propaganda.”
The Washington Post's own testing of the chatbot confirmed that Grok continues to provide the responses shown in the screenshots as of this week.
David Rozado, a New Zealand academic researcher who studies AI bias, gained attention for an article published in March that found ChatGPT's responses to political questions tended to be moderately left-leaning and socially libertarian in orientation. He recently put Grok through some of the same tests and found that its responses to political orientation tests were largely similar to those of ChatGPT.
“I think both ChatGPT and Grok were likely trained on similar web-derived corpora, so the similarity in responses perhaps shouldn’t be too surprising,” Rozado told The Post via email.
Earlier this month, a post on X of a chart showing one of Rozado's findings sparked a response from Musk. While the chart “exaggerates the situation,” Musk said, “we are taking immediate action to move Grok closer to political neutrality.” (Rozado agreed that the chart in question shows Grok further to the left than the results of some of the others tests carried out by him.)
Other AI researchers argue that the political orientation tests used by Rozado overlook the ways in which chatbots, including ChatGPT, often exhibit negative stereotypes about marginalized groups.
A recent filing with the Securities and Exchange Commission revealed that xAI is seeking to raise up to $1 billion in funding from investors, despite Musk's disclosure is not collecting any money at the moment.
Musk and