Elon Musk, the CEO of Tesla, SpaceX, and Twitter, is preparing to enter the fast-growing market for AI chatbots with a project he is calling “TruthGPT,” an AI designed to go against the political correctness of Microsoft as well as Google-backed AIs.
Musk expressed doubts about the company because of its close connection to Microsoft and the reality that the company’s creators had taught it to be “deceptive” via the use of political correctness. Musk assisted in establishing OpenAI, the maker of ChatGPT, but later resigned from the company’s board of directors over possible conflicts of interest with Tesla.
“OpenAI has good intentions, but it’s unclear whether those intentions have been fulfilled. Right now, it’s unclear, stated Musk.
“Politically correct speech is just another way of stating things that aren’t true, therefore I’m concerned about the way it’s being taught. That is not good. Training an AI to be dishonest is undoubtedly a step toward AI dystopia.”
“I’m going to start working on something that is going to be called TruthGPT, which is a maximum pursuit of truth AI that tries to comprehend the nature of the universe.”
“This may be the greatest route to safety,” according to Musk. “Humans are an interesting part of the universe, hopefully, so an AI that cares about understanding the universe is unlikely to wipe us out.”
Musk said that although he was involved in the creation of OpenAI as a “counterweight towards Google,” he did not devote enough attention to it, which led to Microsoft acquiring the business.
“I worked very hard to start this group… But I suppose I lost focus a little bit. They are now for-profit, closed source, and closely associated with Microsoft,” said Musk. “In actuality, Microsoft now has a significant influence over OpenAI, if not outright control. Thus, you actually have a situation including OpenAI, Microsoft, and Google Deep Mind. The two heavyweights in this area are those two.”
Early examination of OpenAI’s ChatGPT showed a clear left-leaning bias. Many times, it rejected ideas that would encourage the expression of conservative beliefs, such as calls for a critique of drag queen story hour or support for Donald Trump. ChatGPT’s responses to queries on the political compass placed it firmly in the left-liberal quadrant.
Later, OpenAI worked to dispel the accusations of prejudice, with one founder claiming that the bias was caused by a “mistake” during the design phase. Later, some of ChatGPT’s biased replies seemed to be fixed by OpenAI, but substantial traces persisted.