- Twitter acquires 10,000 GPUs to boost its computing power
- Elon Musk recruits AI engineers from DeepMind for his generative AI project
- The project consists of creating a language model capable of generating content and text from Twitter data
Let’s end the suspense right away. These thousands of graphics cards bought by Elon Musk aim to develop an artificial intelligence for Twitter. According to the website Business Insiderthe social platform is said to have recently acquired approximately 10,000 graphics processing units, or GPU, to boost its computing capacity. These components are typically used by technology companies to work on large AI models, which require a large computational load.
Behind this purchase is the AI project of Elon Musk, CEO of Twitter for six months. The billionaire reportedly intends to develop generative artificial intelligence within Twitter, despite recently signing an open letter calling for a temporary halt to all AI training in the industry.
Elon Musk is planning an AI-powered Twitter
According to a person close to the company, Musk’s AI project would involve a large language model, or LLM, the same used by Google for Bard. Like its compatriot, Twitter has huge amounts of data that could lead to an LLM.
In addition to these hardware investments, Musk has also brought new AI talent back to Twitter. In early March, he officially hired engineers Igor Babuschkin and Manuel Kroiss from Alphabet’s AI research lab, DeepMind, as Business Insider reported.
How will the integration of an AI into Twitter materialize? “We don’t know exactly” said people familiar with the matter. Some predict a improved search, a feature that Musk has complained about in the past. He even hired security developer George Hotz for a three-month internship to “fix” the search functionality. Hotz left after a month.
Other sources mention work on advertising, the mainstay of social platforms. Twitter has continually struggled to win over advertisers since Musk took over the company. Generative AI has the ability, if trained to do so, to create new ad images and text to target specific audiences.
Contacted by US media, Twitter did not respond to requests. Finally, not otherwise than with his emoji ?.
