Musk joins the AIGC battle! Rush to buy 10,000 GPUs to poach Deepmind

This article is reprinted with the authorization of AI New Media Qubit (public account ID: QbitAI). Please contact the source for reprinting.

Musk entered the AIGC war and rushed to buy about 10,000 GPUs!

What is the concept?

Tesla’s most powerful supercomputer, Dojo, doesn’t take advantage of so much.

The data released by Dojo after the upgrade in 2022 is 7,360 pieces of A100. Although it is not clear what model it bought this time, the quantity alone is terrifying enough.

According to Business Insider, these 10,000 graphics cards were purchased for Twitter.

The whistleblower said that the Twitter version of GPT is already in preparation, and Twitter itself has massive data and has certain advantages.

Wait a moment? Didn’t Musk sign an open letter a while ago calling for a six-month suspension of AI models more powerful than GPT-4?

Did you really respond to what netizens said, "Just stop and wait for me to catch up"?

Regarding Musk’s big move this time, some netizens believe that stocking up on GPUs is a good investment in itself.

Focus on large models, poaching two DeepMind engineers

According to people familiar with the matter, Musk’s artificial intelligence project is still in its early stages, and buying so many The additional computing power was enough to demonstrate his determination.

As mentioned at the beginning, these 10,000 GPUs are currently mainly involved in developing a large language model-

Of all the companies under Musk, I am afraid Twitter is the most suitable to do this. , because it can provide a large amount of training data.

As one netizen opined, Twitter, founded in 2006, has accumulated a database for almost 20 years. This is completely an excellent weapon for Twitter to enter the LLM field.

In addition to spending a lot of money to buy GPUs to build large models, Musk is also vigorously grabbing people.

Currently, two DeepMind engineers have been successfully recruited:

One is Igor Babuschkin, who is engaged in AI system research and is a senior scientific research engineer at DeepMind. He has worked at DeepMind for a total of about 5 years. , and also jumped to OpenAI for a year and a half.

The other is Manuel Kroiss, a software engineer who has worked at DeepMind for six years and was a solution engineer at Google before that. The position Musk assigned him after recruiting him was senior director of software engineering at Twitter.

According to BusinessInsider, in fact

Since February

, Musk has begun to frequently "get close" to people in the AI ​​field, including Igor Babuschkin. As for the role of this large model, we don’t know yet, but there are two guesses:

One is to improve search. Because Lao Ma complained about this feature, he even hired George Hotz to repair it for three months.

However, it seems that he and Musk have a "discord". Not only did he reject Tesla's offer invitation, but he also stayed on Twitter for only one month and left after agreeing to do so for three months. .

In addition to search,

another use may be advertising marketing. Advertising is the backbone of social media. Unfortunately, Musk introduced many new policies after acquiring Twitter. Many advertisers directly cut expenditures or simply refused to cooperate. As a result, Twitter's revenue in December last year alone dropped by about 40% compared with the same period in the past year.

And if there are large models for generative AI creation, whether it is designing materials or providing ideas, it may improve Twitter's efficiency and benefits in advertising and marketing.

emmm... Speaking of Twitter’s current

(always)

not very optimistic financial situation, this time people familiar with the matter also said that Musk may have to build this time to buy a GPU. Tens of millions of dollars. After all, it is possible to buy NVIDIA, because it occupies 95% of the market share, and a product for a large AI model costs about US$10,000.

So, have you found the reason why Twitter has been short of money?

(Manual dog head)

Failed to take over OpenAI 5 years ago, Musk’s counterattack is coming

Since the birth of ChatGPT, Musk has frequently expressed his views on OpenAI.

His name is still listed on the open letter page that suspended AI development for 6 months.

#And often expressed dissatisfaction with the current situation of the company he co-founded.

How did the non-profit organization I established with a donation of US$100 million turn into a for-profit company with a market value of US$30 billion? If this is legal, why don't others do it?

As the world continues to pay attention to ChatGPT and OpenAI, more insider information about Musk and OpenAI’s past has been unearthed.

According to Semafor, Musk previously tried to take over the CEO position in 2018, but ultimately failed.

At that time, Musk believed that OpenAI was far behind Google technically and needed to make major changes.

After a fierce internal power struggle, the board rejected Musk's meaning. In response, Musk withdrew from OpenAI completely and poached Andrej Kaparthy to be Tesla's AI director.

Sam Altman, the current CEO of OpenAI, also said in an interview, "I have always regarded Musk as a hero, although he has been quite a jerk on Twitter recently (being a jerk) ."

In addition to making offensive remarks, Musk has also taken practical actions against OpenAI.

In the past, OpenAI could use Twitter data for AI training, but on the fifth day after ChatGPT was released, Musk ordered the suspension of this permission.

Three months later, Musk announced that he would set up his own AI laboratory and chose BasedAI as the team name.

Now, Musk mysteriously posts his latest views on AI, perhaps revealing the development direction of the Twitter version of GPT:

A curious AI might be the best.

Cats may object to (Curiosity Killed the Cat) , but Curious Monkey George (from popular enlightenment books) will strongly support it.

The above is the detailed content of Musk joins the AIGC battle! Rush to buy 10,000 GPUs to poach Deepmind. For more information, please follow other related articles on the PHP Chinese website!