Elon Musk wants to call his own AI chatbot TruthGPT, he says in an interview. His language model must understand the universe and therefore not pose a threat to humanity, the businessman argues.
Elon Musk at Fox News
The name seems to indicate that it is a General Pre-trained Transformer model, such as OpenAI’s GPT. The prefix Truth is reminiscent of the name Truth.social, which former US President Donald Trump gave to his own social medium. Musk announced the name in an interview with the American TV channel Fox Newsa TV channel that has supported Trump a lot.
Musk’s idea is that OpenAI’s GPT series and Google’s Bard are ideologically driven and would therefore be “woke,” a sentiment he has expressed before. Musk wants to counter this with a chatbot with fewer rules for what is not allowed. TruthGPT must have a fundamental understanding of the universe. That makes it safe, Musk claims. “And I think that might be the best way to ensure safety, because an AI that wants to understand the universe is unlikely to destroy humans, because we’re an interesting part of the universe.”
It’s no surprise that Musk is working on its own GPT application. Recently a rumor came out that he had bought ten thousand GPUs for that purpose. Musk also speaks a lot about the risks of AI. He was one of the signatories of a letter calling for AI development to be paused in order to create rules. According to Musk, AI is a technology that can threaten humanity. That is why it would be good to provide frameworks through regulations to ensure that AI is used for the good. Without those rules, according to Musk, AI will quickly become more intelligent than humans and take over power, after which we don’t know what will happen.
Musk’s view is not new. Many researchers working with AI have also been working on the safety of the technology for years. The current AI race among tech companies is forcing companies to develop and release products faster, raising concerns about the risks.