The latest version of Elon Musk\u2019s artificial intelligence chatbot Grok is echoing the views of its billionaire creator, so much so that it will sometimes search online for Musk\u2019s stance on an issue before offering up an opinion.<\/p>\n
The unusual behavior of Grok 4, the AI model that Musk\u2019s company xAI released late Wednesday, has surprised some experts. <\/p>\n
Built using huge amounts of computing power at a Tennessee data center, Grok is Musk\u2019s attempt to outdo rivals such as OpenAI\u2019s ChatGPT and Google<\/a>\u2019s Gemini in building an AI assistant that shows its reasoning before answering a question.<\/p>\n Musk\u2019s deliberate efforts to mold Grok into a challenger of what he considers the tech industry\u2019s \u201cwoke\u201d orthodoxy on race, gender and politics has repeatedly got the chatbot into trouble<\/a>, most recently when it spouted antisemitic tropes<\/a>, praised Adolf Hitler and made other hateful commentary to users of Musk\u2019s X social media platform just days before Grok 4\u2032s launch.<\/p>\n But its tendency to consult with Musk\u2019s opinions appears to be a different problem.<\/p>\n \u201cIt\u2019s extraordinary,\u201d said Simon Willison, an independent AI researcher who\u2019s been testing the tool. \u201cYou can ask it a sort of pointed question that is around controversial topics. And then you can watch it literally do a search on X for what Elon Musk said about this, as part of its research into how it should reply.\u201d<\/p>\n One example widely shared on social media \u2014 and which Willison duplicated \u2014 asked Grok to comment on the conflict in the Middle East. The prompted question made no mention of Musk, but the chatbot looked for his guidance anyway. <\/p>\n As a so-called reasoning model, much like those made by rivals OpenAI or Anthropic, Grok 4 shows its \u201cthinking\u201d as it goes through the steps of processing a question and coming up with an answer. Part of that thinking this week involved searching X, the former Twitter that\u2019s now merged into xAI, for anything Musk said about Israel, Palestine, Gaza or Hamas.<\/p>\n \u201cElon Musk\u2019s stance could provide context, given his influence,\u201d the chatbot told Willison, according to a video of the interaction. \u201cCurrently looking at his views to see if they guide the answer.\u201d<\/p>\n Musk and his xAI co-founders introduced the new chatbot in a livestreamed event Wednesday night but haven\u2019t published a technical explanation of its workings \u2014 known as a system card \u2014 that companies in the AI industry typically provide when introducing a new model. <\/p>\n The company also didn\u2019t respond to an emailed request for comment Friday.<\/p>\n \u201cIn the past, strange behavior like this was due to system prompt changes,\u201d which is when engineers program specific instructions to guide a chatbot\u2019s response, said Tim Kellogg, principal AI architect at software company Icertis. <\/p>\n \u201cBut this one seems baked into the core of Grok and it\u2019s not clear to me how that happens,\u201d Kellogg said. \u201cIt seems that Musk\u2019s effort to create a maximally truthful AI has somehow led to it believing its own values must align with Musk\u2019s own values.\u201d<\/p>\n The lack of transparency is troubling for computer scientist Talia Ringer, a professor at the University of Illinois Urbana-Champaign who earlier in the week criticized the company\u2019s handling of the technology\u2019s antisemitic outbursts.<\/p>\n Ringer said the most plausible explanation for Grok\u2019s search for Musk\u2019s guidance is assuming the person is asking for the opinions of xAI or Musk. <\/p>\n 20 Years Of Free JournalismYour Support Fuels Our MissionYour Support Fuels Our Mission<\/p>\n For two decades, HuffPost has been fearless, unflinching, and relentless in pursuit of the truth.