Unlock Editor’s Digest for free
Roula Khalaf, editor of the FT, picks her favorite stories in this weekly newsletter.
Elon Musk’s artificial intelligence start-up xAI has released its first AI model, as the tech billionaire aims to take on OpenAI, Google and Meta with a cheeky chatbot tightly integrated into X, formerly Twitter.
Grok, the new AI system, has “real-time access” to information from X, the social media platform that Musk bought a year ago for $44 billion, he said in a post on Saturday evening, giving it a “massive advantage over other models” that relied largely on older archives of internet data.
The chatbot “loves sarcasm” and responds with “a little humor,” Musk added, hoping Grok can stand out in an increasingly crowded market by adding more personality.
“It will also answer tough questions rejected by most other AI systems,” xAI said as it announced a “very early” test version of Grok.
So-called generative AI companies – whose technology can create human-like text, code and images in seconds – have raised billions of dollars this year as investors flocked to an industry that proponents say could be as transformative as the internet. However, others fear the inflating of a new technology bubble as commercialization of the technology is still at an early stage.
xAI’s ability to release a powerful model with reportedly just two months of training shows how new entrants are starting to erode the huge lead of OpenAI, which released its groundbreaking chatbot ChatGPT nearly a year ago.
Musk, who said last week that “AI will eventually do everything” and make human labor obsolete, founded xAI earlier this year. Its team of engineers who formerly worked at Google DeepMind and Microsoft is struggling to catch up with more established rivals like OpenAI, which Musk co-founded in 2015 but left three years later.
Although Grok was developed in just a few months, Musk claims that Grok’s capabilities are comparable to the latest models from Meta, which released its LLaMA 2 model in July, and Inflection, the AI start-up led by the former DeepMind co-founder Mustafa Suleyman, benchmark tests give competition.
xAI said Grok’s ability to respond to math questions or demonstrate arguments is similar to OpenAI’s GPT-3.5, the model that powered the first version of ChatGPT when it launched last November. The startup added that Grok passed a Hungarian final exam in mathematics with a grade of C, which corresponds to Anthropic’s Claude model.
However, xAI also admitted to being behind OpenAI, which released its latest GPT-4 model in March, which has demonstrated “human-level performance” on professional benchmarks such as the US Bar exam and is already being embedded into apps by partner companies.
“[Grok] is only surpassed by models trained with a significantly larger amount of training data and computational resources such as GPT-4,” xAI said. “This demonstrates the rapid progress we at xAI are making in training LLMs with exceptional efficiency.”
Oracle, the software company chaired by Musk ally Larry Ellison, said in September that xAI trains its technology using its cloud computing platform.
User of