Musk's reservations about AI already seem to be a public thing.
Previously, he has called artificial intelligence a dangerous technology more than once in public, and at the same time, various "yin and yang" OpenAI have drawn a line with it. Just a few weeks ago, he also issued an appeal along with the big names in the AI research community and industry
Suspend the training of AI large models for half a year and strengthen the supervision of AI technology
's open letter.
But under the current surging wave of artificial intelligence, is Musk really not interested in artificial intelligence and unwavering in his position as he himself said? Recent indications suggest that this does not seem to be.
The latest news shows that Musk recently purchased 10,000 GPUs for Twitter to advance a new AI project, and is most likely developing his own large-language model. In addition, Musk has recently poached various corners in the field of artificial intelligence, and there is a posture to do a big job.
Calling for a moratorium selfishly? Twitter's internal AI project is secretly advancing
According to Business Insider, Musk recently spent about tens of millions of dollars on about 10,000 GPUs (graphics processing units) for use in one of the company's two data centers.
Given that Musk just said not long ago that Twitter is "short of money" and needs to spend about $1.5 billion a year to pay off debt, the analysis believes that Twitter's purchase of these GPUs is mainly for the development of generative artificial intelligence, after all, if not for AI training, Twitter really has no reason to spend so much money on data center-class GPUs. You know, in 2020, Microsoft built a supercomputer for OpenAI, and that computer used exactly 10,000 GPUs.
Image from Engadget
Objectively, Twitter has every foundation to develop its own big language model, and Twitter is arguably one of the companies with the largest amount of data in recent decades, and it can train on its own massive data.
It is unclear how Musk will consider how to use AI technology in Twitter, and some speculate that it may be used to improve Twitter's search function or help expand advertising business, but Silicon Star believes that based on Musk's recent statements and integration actions on Twitter, the layout of this heavy investment on AI may not be so simple.
Recently, it was discovered that a legal document related to the case filed in the California court on the 4th of this month showed that Twitter had been merged into Company X and no longer existed. Yesterday afternoon, Musk sent a tweet on Twitter containing only the letter "X" that also seemed to officially announce the de facto "end" of Twitter.
The documents show that X is a newly formed shell company that is privately held, while its parent company is X Holdings. Musk serves as president of Company X and Group.
Image from public legal documents
Anyone who knows Musk probably knows that he has always had an obsession with building "Empire X". PayPal was the online bank he founded in 1999 called X.com, and Musk said last year that he would relaunch the X.com plan on Twitter, which was bought to accelerate his "universal app" plan.
Musk's "universal app" is actually inspired by WeChat. Musk has publicly recognized and praised WeChat's model more than once, believing that it integrates chat, payment, financial management, inquiry, and information consultation into one application, believing that this super application has great potential, but has not seen the emergence of a similar model in markets outside China.
He made clear his plans to overhaul Twitter and make it the center of users' lives and finances. In an internal letter last month, he acknowledged that Twitter's valuation was almost cut from when he bought it, but he had a very clear plan for Twitter, which he believed could be worth $250 billion in the future.
Image from Twitter
Therefore, if you use the positioning of "American version of WeChat" to look at Twitter, then what artificial intelligence can do for Twitter in the future may have gone far beyond the scope of search and advertising, and may also be used in various scenarios such as dialogue and communication, financial payments, etc.
Build your own AI team and recruit people from DeepMind
In addition to buying hardware equipment for Twitter's heavy investment, Musk has recently worked hard in the battle for artificial intelligence talents.
As early as February this year, Musk was revealed to have been looking for talents in the field of artificial intelligence, planning to recruit an artificial intelligence research and development team to form a new research laboratory, with the direct goal of competing with OpenAI's ChatGPT.
Musk has recruited at least two senior engineers from Google's AI research company DeepMind, and has joined Twitter in March. Igor Babuschkin, a senior researcher in DeepMind's AI division, has previously worked at OpenAI, is a member of the ChatGPT project team, and will currently be responsible for the development of Twitter's ChatGPT-like products. Another Manuel Kroiss is a 6-year veteran of DeepMind and currently serves as Twitter's new senior director of software engineering.
Musk's recent move to quietly carry out a series of AI layouts has also once again made him controversial. Many netizens believe that this also confirms the reason why Musk previously took the lead in calling for the suspension of AI research, because he wanted to buy himself time.
Last week, LinkedIn founder Hoffman also directly put this view of Ming Xiu plank road and secret crossing Chen Cang on the table. He said the letter was an attempt by some parties to slow down industry leaders like OpenAI so they could catch up, and as a former business partner, he saw Musk's high-profile participation in the letter more of a strategic move.
Do you dislike artificial intelligence, or do you dislike Open AI?
In fact, when you look back at Musk's love-hate entanglement with artificial intelligence over the years, you may come to the conclusion that Musk does not really dislike artificial intelligence, and his opinion may be more derived from OpenAI.
As Musk joins the AI battle and the "war of words" with OpenAI heats up, more stories behind the break between the two have recently been picked up.
As we all know, Musk was previously a co-founder of OpenAI, but the story behind the sudden split between Musk and OpenAI in 2018 was once a mystery. At the time, the official statement was that there was a conflict of interest between Tesla's artificial intelligence system and OpenAI's business, and Musk had to leave OpenAI for Tesla, but would still support the development of OpenAI.
But the fact is that the most important reason for Musk's separation from OpenAI was different from Ultraman's development philosophy, Musk thought that OpenAI lagged too far behind Google and wanted to fully control the operation and management of OpenAI, but Musk's proposal was rejected by the board. After some internal struggle, Musk left OpenAI. In addition to the $100 million already donated, his previous commitment to $1 billion in support was interrupted.
Musk's divestment made OpenAI, which needed a lot of money for training at the time, very difficult, so that it had to create a for-profit company and take $1 billion from Microsoft to "survive". Since then, with Microsoft's all-round support from funds, hardware and software, OpenAI has embarked on the fast track of development, but it has also completely parted ways with Musk.
After the launch of ChatGPT beta last November, Musk was not excited, but directly shut down OpenAI's access to Twitter data, and since then began a series of bombardments against OpenAI.
Image from Twitter
For example, the naming and shelling of OpenAI has moved away from the original intention of the non-profit organization, saying that it "donated $100 million to establish a non-profit organization, how can it become a for-profit company with a market value of 30 billion?" If this is legal, why don't others do it? ”
Image from Twitter
For example, it said that OpenAI was established to compete with large institutions such as Google, but now it has fallen into the hands of Microsoft.
Another example is to continuously question the security of OpenAI technology, saying that the emergence of ChatGPT is a very dangerous signal, and so on.
And, of course, the famous recent open letter calling for a moratorium on training for large models. And it is reported that Musk is actually the original initiator of the Future of Life Institute, the agency that issued the open letter.
To be sure, Musk is not rejecting AI technology. In addition to his current layout on Twitter, Tesla's humanoid robots and automatic driving are actually closely related to artificial intelligence. However, Musk may have his own definition of safe artificial intelligence and his vision of a future artificial intelligence society, and obviously, he does not approve of OpenAI.
So what is Musk's ideal artificial intelligence, and can he really do better than Open AI and truly achieve the standardized development of artificial intelligence? We may be able to wait for future Twitter, or Company X, to give us answers.
Note: The cover image comes from the Google Public Copyright Gallery, and the copyright belongs to the original author. If you do not agree to the use, please contact us as soon as possible and we will delete it immediately.
![Musk is going to be hunted down? His 50,000 knives of low-cost housing have been exposed[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)