Elon Musk took up artificial intelligence

Elon Musk founded a new artificial intelligence (AI) company called X.AI. This is reported by The Wall Street Journal (WSJ) with reference to the registration certificate. The newspaper’s sources believe that Musk’s new startup will develop large language models and compete with OpenAI products, the ChatGPT chat bot developer and the neural network for generating images DALL-EXAI was registered in Nevada in early March, Musk himself became its leader, and as his deputy is Jared Birchall, a former banker at Morgan Stanley, who has headed the entrepreneur’s family office since 2016 and manages his affairs and fortune. In early March, it became known that Twitter, owned by Musk, changed its legal name to X Corp. and also passed into the jurisdiction of Nevada. Compared to Delaware, where Twitter was previously registered, Nevada’s laws provide more freedom of action and better protection for business owners, according to the WSJ.

Musk has spent the past few months hiring AI experts to create a competitor to OpenAI, the publication reports. In particular, the head of Twitter recently hired Igor Babushkin, an employee of Alphabet’s DeepMind lab, and “about half a dozen other engineers.” He also tried to poach several employees of OpenAI itself, says a newspaper source familiar with the situation. In addition, the billionaire is in talks with current investors SpaceX and Tesla to invest in a new startup, writes FT. Sources of the publication add that thousands of Nvidia graphics processing units (GPUs) have already been purchased to create a large language model.

Musk was one of the founders of OpenAI, but left the project in 2018 amid a conflict with the company’s current CEO, Sam Altman. In 2022, Musk repeatedly criticized the company’s ChatGPT chatbot, as he considers the company’s AI model to be “politically biased.” Moreover, at the end of March, the entrepreneur, along with Pinterest co-founder Evan Sharp and Apple co-founder Steve Wozniak, signed a petition calling for a six-month suspension of the development of new generative AI models. The statement said that during this time the industry will be able to develop some standards for the safe use of technology and thereby prevent potential harm.

Now AI is becoming critically needed for any company, not only from the IT sector, said Svetlana Zakharova, head of Just AI’s partner and client department. “The former president of Google China, Li Kaifu, says that almost all the software we know is about to change, and this is already evident in the presentations of Microsoft, which are integrating AI 2.0 into their office applications,” the expert added. According to her, Musk cannot stay away from this trend and it is the ethical consequences of the emergence of new technologies that concern him, and not the commercial component.

The manifesto with a proposal to suspend research in the field of AI was rather an attempt to attract as much attention to this topic as possible, including from states, to consider the issue of regulation, says Vladimir Sveshnikov, co-founder of Parodist AI. Therefore, the creation of a company in this area, according to him, does not look like a radical change in rhetoric. It is possible that this call concealed Musk’s desire to “catch up with the outgoing OpenAI train” by creating artificial barriers for a successful competitor, Alexey Boyko, an analyst at MForum agency, believes.

Sveshnikov, on the contrary, does not believe that the manifesto could limit anyone in the market. According to him, if the goal were such, then more serious measures would be taken – for example, a request was sent to Congress, lawsuits, etc. But even these steps would still not affect the development of AI in other countries, for example in China, he added.

In general, investments in generative AI seem quite logical, since in addition to the obvious benefits for existing Musk companies, it is worth considering the fact that Twitter is a constantly updated source of a large number of texts and dialogues necessary for training language models, Sveshnikov explained. OpenAI also used Twitter data to train the GPT, but Musk put that process on pause, he recalled.

Creating a company with a product similar to GPT-4 could have a synergistic effect on Musk’s other businesses, experts interviewed by Vedomosti agree. The solution, in theory, could help in the fight against bots on Twitter, a problem that Musk regularly discusses and considers important, Sveshnikov suggested.

AI can also be used in the internal processes of Twitter as a company and become part of more convenient interfaces for working with the product, Zakharova agrees. It is possible, for example, using the language model to improve interaction with users, advertising customers, and also to make a “smart search” on the social network, Boyko added. In the case of Tesla, the model could potentially be applied in the components of the autopilot system, Sveshnikov suggested.