Elon Musk said he is developing an artificial intelligence chatbot called "TruthGPT," describing it as a "maximum truth-seeking AI" in an interview with Tucker Carlson on Fox News. The Twitter CEO framed the project as an alternative to OpenAI's ChatGPT, which he accused of being trained to avoid certain topics.
"I'm going to start something which I call 'TruthGPT' or a maximum truth-seeking AI that tries to understand the nature of the universe," Musk told Carlson. It was the first time Musk publicly announced plans to build a bot under that name on his own, though he had mentioned the term before.
Musk's Case for a 'Truth-Seeking' AI
Musk argued that an AI designed to understand the universe would be safer than one focused on other goals. "An AI that cares about understanding the universe is unlikely to annihilate humans because we are an interesting part of the universe," he said. He compared the dynamic to human efforts to protect chimpanzees, noting that people choose to preserve the species and its habitats rather than hunt it.
Musk directed his sharpest criticism at OpenAI, the company behind ChatGPT. "What's happening is they're training an AI to lie," he said. "To comment on some things, not comment on other things. Not to say what the data actually demands that it says."
Contradictions in Musk's AI Stance
The announcement sits uneasily with Musk's previous warnings about artificial intelligence. He has called for regulation of the technology and said in the same Fox News interview that AI carries the "potential of civilization destruction." He also recently signed an open letter calling for a moratorium on developing AI systems more advanced than OpenAI's GPT-4 model.
Building a "universe-understanding" AI could conflict with the limits he has publicly endorsed. It remains unclear whether TruthGPT exists beyond the recent establishment of an AI company called X.AI. The source material did not include details on funding, staffing, timelines, or technical specifications for the project.
Musk's track record on ambitious timelines has drawn skepticism. He has promised fully self-driving Tesla vehicles since 2014, a goal that remains unmet. That history, combined with the early stage of the TruthGPT announcement, leaves the project's scope and viability unresolved.
Editorial Notes