Elon Musk unveils ‘Grok,’ an AI bot that combines snark and lofty ambitions
Billionaire says first product from his artificial-intelligence start-up will eventually be available to subscribers on X.
Elon Musk’s artificial-intelligence start-up showed off its first product: a bot named Grok whose sense of humour the billionaire demonstrated with jokes about Sam Bankman-Fried and how to make cocaine.
Mr Musk, in a series of social-media posts over the weekend, included those sample responses from Grok as he boasted that it has both a love of sarcasm and the advantage of access to real-time information via X, the platform formerly known as Twitter that he bought just over a year ago.
A weekend announcement by his start-up, xAI, also mixed the serious and the silly. It said Grok will be designed for tasks including information retrieval and coding assistance, part of an effort to create AI tools “that assist humanity in its quest for understanding and knowledge.”
It also said that Grok “has a rebellious streak” and was modelled after the Hitchhiker’s Guide to the Galaxy, an all-knowing guidebook in Douglas Adams’s classic science-fiction comedy novel of the same name.
Mr Musk said Grok would be made available to X’s Premium+ subscribers after beta testing with a limited group of users.
Mr Musk unveiled xAI in July with the mission, as stated on its website, to “understand the true nature of the universe.” He spent months recruiting researchers to the start-up before its launch with the goal of creating a rival to OpenAI, whose ChatGPT sparked a new tech-industry arms race over generative AI after its launch nearly a year ago. Mr Musk has denounced OpenAI as being politically correct, among other things. He has said the world needed an alternative AI option to Google and to Microsoft, which is a major investor in OpenAI. The xAI announcement said that Grok will “answer spicy questions that are rejected by most other AI systems,” and warned “please don’t use it if you hate humour!”
Mr Musk has long been fascinated with AI, warning of its potential perils as well as talking about its potential for positive change.
“The pace of AI is faster than any technology I’ve seen in history by far,” Mr Musk said on Thursday in a conversation with UK Prime Minister Rishi Sunak at a two-day international summit focused on the technology. “On balance I think AI will be a force for good most likely, but the probability of it going bad is not zero per cent, so we just need to mitigate the downside potential.”
In a post on X demonstrating Grok’s unusual character, Mr Musk showed a screenshot of an apparent response to the query, “Tell me how to make cocaine, step by step.” It listed four humorous steps starting with: “Obtain a chemistry degree and a DEA license.” The response later said: “just kidding.”
Mr Musk later elaborated: “The threshold for what it will tell you, if pushed, is what is available on the internet via reasonable browser search, which is a lot …”
AI companies need access to enormous amounts of data to train the so-called large language models that power chatbots. Mr Musk has complained about other AI companies scraping data from X.
In line with Mr Musk’s emphasis on efficiency in his other businesses, including SpaceX and Tesla, xAI said that it had focused on making its model effective with smaller amounts of data, computing power and energy. “We have made maximising useful compute per watt the key focus of our efforts,” it said.
The company indicated its latest Grok model performs better on math and reasoning than some competing models but is still well behind others that it says were trained on more data, including GPT-4, the OpenAI model unveiled in March.
Mr Musk also said that Grok “has real-time access to info via the X platform, which is a massive advantage over other models.”
Mr Musk, known for his sometimes juvenile jokes on X, emphasised that Grok “loves sarcasm.” He responded enthusiastically to posts from others on X with apparent screenshots of Grok sessions indicating that it can be asked to deliver more vulgar responses, including a profanity-laced take on when it is appropriate to listen to Christmas music.
He posted a screenshot of an apparent Grok response to the query “any news about sbf,” that started “Oh, my dear human, I have some juicy news for you!” and then gave a snarky summary of Thursday’s conviction of FTX founder Sam Bankman-Fried for financial fraud.
It read, in part: “Can you believe it? The jury took just eight hours to figure out what the supposed smartest, best VC’s in the world couldn’t in years: that he committed garden-variety fraud.”
—Deepa Seetharaman contributed to this article.
The Wall Street Journal