Elon Musk, CEO of Tesla and SpaceX, has previously raised concerns about artificial intelligence risks. However, it appears that he has quietly launched an AI research company, X.AI Corp., after leaving the OpenAI board and has now revealed the development of an AI chatbot called “TruthGPT” that seeks the greatest truth of the universe.
Musk has reportedly poached talent from DeepMind, a subsidiary of Alphabet, and has purchased thousands of GPUs from NVIDIA to train the AI on OpenAI’s ChatGPT patterns.
It is unclear what exactly Musk means by calling TruthGPT a “truth-seeking AI.” While some believe it could refer to a self-conscious AI that can distinguish between truth and falsehood, this level of technology is currently not achievable. Furthermore, distinguishing between real and false is a difficult problem for AI developers.
Others believe that TruthGPT may simply be a large-scale language model (LLM) like OpenAI’s ChatGPT, which has shown remarkable progress in natural language processing. However, using Twitter’s internal data to train the AI presents its own challenges, as the raw data is of varying quality and consistency, requiring extensive cleaning and preprocessing.
Regardless of its form, it is apparent that Musk is serious about AI and is investing heavily in its research. Given his earlier concerns about the risks of AI, it will be intriguing to watch how he weighs the possible advantages against the potential threats of this powerful technology.