Twitter CEO Elon Musk recently purchased approximately 10,000 GPUs for a generative AI project within Twitter, despite calling for a six-month pause in the development of powerful AI models. The specific idea of the task, notwithstanding, is as yet a secret.
Graphics processing units, or GPUs, are chips made specifically for computer graphics. However, due to their massive parallelism, they can also be used to process generative AI. Since training (creating) a new AI model typically requires a lot of computing power, including a lot of GPUs, Musk’s acquisition of Twitter may represent a significant investment in the creation of a deep-learning AI model.
For this project, Musk is said to have invested tens of millions of dollars in GPUs. Business Insider’s source says that one of Twitter’s two data centers in Atlanta will most likely house the GPU units.
The development of novel advertising strategies or enhancements to Twitter search are two examples of potential generative AI applications. Business Insider quotes a person familiar with the project as saying that the purchase of the GPUs demonstrates that Musk is “committed” to AI development within the company. This suggests that the purchase of the hardware could also be interpreted as public posturing.
At the end of March, Musk signed an open letter that was widely discussed. The letter was written by the Future of Life Institute, which Musk partially funds. It called for a six-month halt to the creation of AI models that are more powerful than GPT-4, the released language model that is currently OpenAI’s most powerful. The researchers cited the potential dangers that runaway AI poses to human civilization.
Even though the idea of a pause didn’t work out, some people said that Musk wanted a “pause” so that his companies could catch up with OpenAI. Even though it hasn’t been said whether Twitter’s AI model will be “more powerful” than GPT-4, these latest actions may be interpreted as undermining Musk’s warnings to be cautious.