Elon Musk's new AI bot comes with an unusual warning attached: don't use it if "you hate humor."
Grok, a large language model (LLM) built to rival the likes of ChatGPT and Bard, was announced over the weekend by xAI, a Musk-led company which will build artificial intelligence to "accelerate human scientific discovery."
In intended breadth and tone, xAI said the bot is modeled after The Hitchhiker’s Guide to the Galaxy, a multi-book fictional guidebook to space that originated in a comedic British science fiction radio series of the same name.
As its inspiration may suggest, xAI says Grok has been built with a jocular streak.
"Grok is designed to answer questions with a bit of wit and has a rebellious streak, so please don’t use it if you hate humor!" the company said in an announcement on Nov. 4.
Although xAI is a company independent of Musk's other endeavors, it does declare it will work closely with his companies including X—formerly known as Twitter—and EV maker Tesla.
As a result, xAI says its LLM has a "unique and fundamental advantage" that it can be fed real-time knowledge via X.
Healthy competition?
The announcement also makes slight criticism of competitors, while acknowledging Grok is in its early stages.
For example, xAI says the bot will "answer spicy questions that are rejected by most other AI systems."
ChatGPT, for example, told Fortune it won't answer questions pertaining to political bias or propaganda, conspiracy theories or "nonsensical content."
The announcement also claims Grok-1—the engine powering the service—surpassed many of its competitors in testing phases.
xAI said Grok was given math word problems, Python code tasks and multidisciplinary multiple choice questions, and published results appearing to show Grok outpace ChatGPT-3 and Pi's Inflection-1.
According to the results from xAI—it is unclear whether these findings have been verified by independent parties—Grok was more than 60% accurate across three of the four tests.
However, Musk's model does fall short of resources like ChatGPT4, reasoning these models "were trained with a significantly larger amount of training data and compute resources."
xAI said its rapid rollout—Grok has only had two months of training—"showcases the rapid progress we are making at xAI in training LLMs with exceptional efficiency."
Musk is also reentering the race having co-founded OpenAI himself in 2015. The serial entrepreneur left the company's board in 2018, and has since engaged in something of a war of words with current CEO Sam Altman.
In May Musk, the richest man on Earth, told CNBC: "I am the reason OpenAI exists."
Altman fired back, calling Musk a "jerk" and countered that the SpaceX CEO had been little more than a "talent magnet and attention magnet", though he conceded Musk "has some real superpowers that were super helpful to us in those early days."
What is Grok?
The Hitchhiker’s Guide to the Galaxy may not be the only sci-fi novel the xAI team has taken inspiration from.
The term 'grok', which is commonly used to mean intuitive understanding, was invented by writer Robert A. Heinlein in his novel Stranger in a Strange Land.
In the book, grok is a Martian word which conveys understanding on a profound and intuitive level.
In its announcement xAI said it wants to create tools which "assist humanity in its quest for understanding and knowledge," adding: "We believe that it is important to design AI tools that are useful to people of all backgrounds and political views. We also want empower our users with our AI tools, subject to the law. Our goal with Grok is to explore and demonstrate this approach in public."
Grok does have a waitlist to try the service, however currently only verified X users will be granted early access.
Didn't Musk want to pause AI?
In March, Musk joined the likes of Apple co-founder Steve Wozniak calling for a six-month ban on the development of advanced LLMs like ChatGPT.
Despite signing the letter penned by the Future of Life Institute, just four months later Musk announced the formation of xAI which itself was tasked with creating an advanced LLM to rival ChatGPT.
The man worth $209 billion, according to the Bloomberg Billionaire's Index, addressed this apparent hypocrisy yesterday, responding to a user on X: "I signed on to that letter knowing it was futile. I just wanted to be on record as recommending a pause."
