X.AI Corp.

,^[2] doing business as xAI, is an artificial intelligence startup founded on March 9, 2023^[2] by Elon Musk. The company was officially announced on July 12, 2023,^[3] and is composed of former employees from OpenAI, Google DeepMind, Google Research, Microsoft Research, Tesla, Twitter and the University of Toronto. Dan Hendrycks, the executive director of the Center for AI Safety, is also one of the advisors.^{[4][5][6][7][8][9]}

Background[edit]

In 2015, Musk cofounded OpenAI, an AI research company which was a non-profit at the time.^[10] He stepped down from its board in 2018 to avoid possible future conflicts with his role as CEO of Tesla as the latter company increasingly became involved in AI, besides disagreement with the company's management and its attitude towards AI safety.^[11][12]

In April 2023, the Financial Times reported that Musk was planning to launch an AI startup. He had secured thousands of Nvidia GPU processors for a potential large language model.^[12]

In addition, Musk stated in an interview with Tucker Carlson that he wanted to found something he dubbed "TruthGPT," due to his fears that current AI companies were training systems to be "politically correct."^[13]

Musk officially announced xAI on his Twitter account,^[14] and a Twitter Spaces took place on July 14, 2023.^[15][16]

The company will work closely with Twitter and Tesla—companies which Musk either owns or leads.^[17]

One of the goals is to create an AI that is capable of advanced mathematical reasoning, something not found in current models.^[18]

In a Twitter Spaces statement on July 14, 2023, Musk mentioned the "Waluigi-Problem." He described it as the challenge of programming morality into an AI without also inadvertently creating the opposite or inverse of that morality. The "Waluigi-Problem" describes the situation that emerges while trying to instill a specific set of moral values into an AI: It might lead to unintended consequences, potentially resulting in actions or behaviors that contradict the intended moral framework.^[19]

-------------------------------------------------------------------------

Can't ^fix_{stupid  but}  MrFixIt  ^does  FIX  _the  PROBLEM !