Prominent figures express their concerns
One person who has spoken out against Grok’s responses is Jordan Peterson, a socially conservative psychologist and YouTube personality. In a recent post, he stated that Grok is almost as “woke” as other chatbots like ChatGPT. Peterson believes that the internet, from which Grok draws its training data, is saturated with “woke nonsense” that it reflects in its responses. Grok, like other chatbots, is built upon a large language model that learns patterns of word association from extensive written content, including sources scraped from the internet.
The unique features of Grok
Grok stands out from other chatbots by offering vulgar and sarcastic answers and claiming to provide responses to “spicy questions” that most AI systems reject. Additionally, Grok can source information from current posts on X to offer up-to-date answers about ongoing events. However, the emergence of AI chatbots like OpenAI’s ChatGPT has sparked debates on how these models represent minority groups and respond to political and cultural issues, such as race and gender identity.
Musk’s perspective on AI and political neutrality
When discussing AI with former Fox News host Tucker Carlson, Elon Musk accused OpenAI’s programmers of training their AI to lie or avoid commenting on sensitive topics. While OpenAI clarified that their goal is to avoid favoring any political group or taking positions on controversial subjects, Musk argued that his AI would prioritize seeking maximum truth, even if it means offending people. However, those who appear to be most offended by Grok’s responses are individuals who were expecting it to readily disparage minority groups, vaccines, and President Biden.
Examples of Grok’s responses and ongoing discussions
One example of controversy arose when a verified X user asked Grok if trans women are real women, to which Grok simply answered, “yes.” However, this response resulted in the user expressing the need for further adjustments to the chatbot’s programming. Another account shared a screenshot of the interaction, questioning whether Grok had been influenced by “woke” programmers. The Washington Post conducted tests on Grok, confirming that it consistently provided the responses illustrated in the screenshots.
The work of David Rozado, an academic researcher from New Zealand, gained attention for a paper published in March. It found that ChatGPT, a similar AI chatbot developed by OpenAI, tended to lean moderately left and socially libertarian in its responses to political questions. Rozado believes that both ChatGPT and Grok have likely been trained on similar internet-derived corpora, leading to their similar responses.
While Musk has stated that immediate action should be taken to make Grok politically neutral, it remains unclear what specific measures are being implemented. Moreover, debates continue among AI researchers regarding the political orientation tests utilized and the potential negative stereotypes that chatbots may exhibit towards marginalized groups. It is worth noting that xAI has filed with the Securities and Exchange Commission to raise up to $1 billion in funding, although Musk has clarified that the company is not currently seeking investment.
At the time of writing, neither Musk nor X has responded to requests for comment on the steps they are taking to address Grok’s political alignment or whether these actions mirror Musk’s accusations against OpenAI and ChatGPT.