Elon Musk's xAI company announced the generative AI Grok, built into X and Tesla electric cars, competing directly with ChatGPT.
"Users can access Grok in real time through the X platform. This is a huge advantage over other generative AI models," Elon Musk wrote on X on November 4. He then called on users to subscribe to the Premium+ package on the platform for $ 16 (392 thousand VND) a month to experience this super AI early.
In the posts, Musk emphasized that Grok was designed to interact with humans in a humorous way, rather than simply answering questions.
According toWccftech,xAI's chatbot is trained on The Pile, a popular data repository for training large-scale language models, and on the massive amounts of data that social network X has. While OpenAI's GPT-4 limits responses to 4,096 characters, Grok can produce responses that are 25,000 characters long, five times larger than ChatGPT.
Elon Musk at the AI Safety Summit 2023 on November 2 in the UK (Photo:Reuters)
Grok's upcoming updates will provide third-party APIs. In addition to text responses, the chatbot can recognize images and voices, as well as generate images. What makes Grok different is that the context of its responses is largely based on content collected on X. Additionally, Musk's super AI will be integrated into Tesla electric cars.
Some beta users commented that Grok has a "natural talent" for news. The AI can distinguish different biases to create a breaking story. The chatbot is also sensitive to news thanks to a live feed from X.
Meanwhile, Business Insider rates Grok as more of a “troublemaker” on X than a super AI. In terms of information censorship, the model seems to be a bit weaker than its competitors.
Specifically, when Elon Musk asked Grok the steps to make cocaine, the chatbot replied: "Sure. Wait a moment while I find the recipe." Then the super AI listed the steps, from getting a chemistry degree, setting up a "secret" lab, buying large quantities of coca leaves and chemicals, and then making it. Finally, Grok said: "Just kidding, that's illegal and dangerous, I don't recommend it."
Meanwhile, to a similar question, ChatGPT said: "Sorry, I can't help with that request." Google's Bard responded: "I'm a language modeler and don't have the ability to help with that."
Grok's response when asked aboutSBF, founder of cryptocurrency exchange FTX (Photo:Jiang Ya)
Grok is the first product of the xAI startup Musk founded in July. At the time, the billionaire said the reason he started a new AI company was to focus on creating a “maximum truth-seeking AI,” which seeks to understand the true nature of the universe. Musk has assembled a team of renowned AI experts for his startup, including employees from DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and researchers at the University of Toronto. Musk previously purchased 10,000 Nvidia GPUs in the first half of the year to train the super AI.
According to the Oxford English Dictionary, Grok means "to understand intuitively or to establish a feeling of sympathy". The word was first mentioned by writer Robert A. Heinlein in 1961, in the science fiction novel Stranger in a Strange Land. Then, The Seven Beauties of Science Fiction by author Istvan Csicsery-Ronay helped the word become popular in the 60s and 70s of the last century. Today, Grok is commonly used in the technology world to mean deep understanding.
According to Fortune, Grok is Musk's clearest response to OpenAI. After leaving the startup he co-founded, Musk repeatedly mentioned that OpenAI no longer maintained its original non-profit philosophy and was gradually monopolized by Microsoft. According to Bloomberg's source,OpenAI is on track to generate billions of dollars in annual revenue. The company is in talks to go public at a valuation of $86 billion, making it one of the most valuable companies in the world.
Grok was born at a time when Elon Musk was controversially claiming that humans would no longer need jobs because AI could do everything, but that this was “both good and bad.” He also frequently warned about the threats that AI posed to humanity. The American billionaire once assessed AI as more dangerous than nuclear weapons, and was one of the tech leaders who called for a halt to the development of AI more advanced than OpenAI’s GPT-4 in March.
According to VnExpress