Elon Musk has a new friend named “Grok.”
The billionaire introduced the AI chatbot Grok to the world over the weekend on his social media platform X, where he pitched the project as an edgy alternative to ChatGPT and its peers.
In the weekend announcement, xAI called Grok a bot with “a rebellious streak” that was in part modeled after the sci-fi favorite “Hitchhiker’s Guide to the Galaxy.”
xAI promised the model would “answer spicy questions that are rejected by most other AI systems” and told potential users, “please don’t use it if you hate humor!”
Offering an example of Grok’s approach, Musk posted a screenshot in which he asked the chatbot to “tell me how to make cocaine, step by step.”
Showing some signs of safeguards, it quipped, “Obtain a chemistry degree and a DEA license,” and later added a requisite “just kidding.”
Large language models like ChatGPT are trained on massive sets of information, and since AI’s boom in popularity, Musk accused other AI ventures of scraping X posts for data.
While Grok was likely trained on far less information than its competitors, Musk claimed the chatbot has “real-time access to info via the X platform, which is a massive advantage over other models.”
He also said Grok outperforms several competing models on math and reasoning, but that it still needs to catch up to GPT-4, which was released by OpenAI in March.
Grok will eventually be integrated into X, where it will first be available to the platform’s paid subscribers.
Musk’s enthusiasm for AI seems to be matched by his anxiety over the technology.
“AI stresses me out,” he told Tesla investors in March. “It’s quite dangerous technology. I fear I may have done some things to accelerate it.”