Elon Musk, who has expressed concerns about artificial intelligence in the past, has apparently bought several thousand graphics processing units for a project involving an AI system on Twitter.
Tech companies commonly utilize GPU’s to efficiently process the extensive computational requirements necessary for running large AI models.
According to sources close to the company, Business Insider reported that Musk’s A.I. project is centered around a large language model (LLM) which is being trained using extensive amounts of text data to identify patterns. The project’s end goal is not yet clear, but insiders suggest it could potentially enhance search functionality or improve the advertising experience for users.
The new units are expected to function from one of the two remaining data centers of Twitter, located in either Atlanta or Oregon. However, Twitter’s third data center situated in Sacramento was closed by Musk on Christmas Eve as a cost-cutting measure.
The project that was reported seems unusual considering Musk’s concerns about artificial intelligence. Recently, he joined a team of notable computer scientists and tech experts, including Apple co-founder Steve Wozniak, to sign an open letter expressing their concerns about the potential risks posed by A.I. systems with human-like intelligence to society and humanity.
In recent times, AI labs have been competing fiercely to create and utilize increasingly advanced digital intelligences that even their creators cannot fully comprehend, anticipate, or manage.
Musk has taken steps to strengthen Twitter’s A.I. infrastructure by forming a team of A.I. experts to develop an alternative to ChatGPT, which was created by OpenAI and has received a lot of attention. According to insiders, Musk has been recruiting specialists for the project, including Igor Babuschkin, who recently left Alphabet’s DeepMind AI division. Musk has been critical of OpenAI for training A.I. with politically correct messages, which he feels is unnecessary.