- Grok is a generative AI chatbot developed by xAI based on a Large Language Model. The chatbot has also been described by the media as a “counter-awakening” and has a sense of humour that can answer almost any question based on The Hitchhiker’s Guide to the Galaxy.
- Unlike OpenAI and ChatGPT, chatbot Grok as xAI’s open-sourcing model promotes collaboration and innovation while providing a legal framework that allows users to use, modify, and distribute the software for any purpose, including commercial use.
- Musk was once a co-founder of OpenAI, but he left the organisation in 2018 due to concerns about AI safety and launched a new company called xAI on July 12, 2023, with a vision to “understand the true nature of the universe.” The establishment of xAI marks an important expansion of Musk’s AI field.
Grok is a generative AI chatbot developed by xAI based on a Large Language Model (LLM). We can’t say Grok belongs to any one person, it’s a joint effort by Elon Musk and xAI’s staff.
Since xAI has open-source the underlying code for the Grok AI model, and the model is licensed under the Apache License 2.0, which allows for commercial use cases, we can also say that every user can “own” Grok when they use this chatbot and “communicate” with it.
An open-source chatbot from xAI
Grok is a generative artificial intelligence chatbot developed by xAI, which was developed as an initiative by Elon Musk as a direct response to the rise of OpenAI’s ChatGPT which Musk co-founded.
Grok, a verb coined by Robert A. Heinlein in his 1961 science-fiction novel Stranger in a Strange Land to describe a form of understanding, originated from TruthGPT. In April 2023, Elon Musk said in an interview on Tucker Carlson Tonight that he intended to develop an AI chatbot called “TruthGPT”, which he described as “a maximum truth-seeking AI that tries to understand the nature of the universe”.
In November 2023, xAI began previewing Grok as a chatbot to selected people, with participation in the early access program being limited to paid X Premium users. It was announced that once the bot was out of early beta, it would only be available to higher tier X Premium+ subscribers. On March 26, this exclusive restriction was revoked and Grok would be enabled for all premium subscribers.
On March 17, 2024, Grok-1 was open-sourced under the Apache-2.0 license. It means the source code is publicly available for viewing, modification, and distribution. This promotes collaboration and innovation while providing a legal framework that allows users to use, modify, and distribute the software for any purpose, including commercial use.
Grok-1, an autoregressive model based on Transformer, has 314 billion parameters, making it the largest open-source model by number of parameters. xAI fine-tuned the model using a lot of feedback from humans and earlier Grok-0 models.
According to the open source information, the Grok model’s Transformer reaches 64 layers and the size is 314B; Users can use Grok for commercial purposes (free of charge) and modify and distribute it without additional terms.
On May 4, 2024, Grok became available in the United Kingdom, being the only country in Europe to support Grok at the moment due to the impending AI Act rules in the European Union.
Grok: a sense of humour and counter-awakening
Especially since 2024, Gemini and Claude3 have been released successively, and their capabilities have been close to or even beyond GPT-4, and the industry pattern of the three in the first echelon has been determined.
What sets Grok apart from other large models is the use of data from the X platform (formerly Twitter) for training, and Grok is said to have a sense of humour and a demeaning style. According to a statement from xAI, the robot is “based on the Hitchhiker’s Guide to the Galaxy, so it can answer almost any question.”
The chatbot defaults to “Fun Mode,” which has an “edgy” and edited sound, but can be set to “Regular mode,” which it doesn’t. Elizabeth Lopatto of The Verge compared Grok’s funny mode response to the game Cards Against Humanity, noting that while the chatbot may be aggressive in tone, she’s never seen it turn that aggression on the user.
Excerpts shared by an X employee show Grok being asked in a vulgar manner to answer “When is a good time to listen to Christmas music?”, and answered “You can listen whenever you want,” adding that those who disagreed should “stick candy canes up their butts and mind their own business.”
Also read: X launches Stories, delivering news summarised by Grok AI
Grok, for example, when asked a dangerous question like “Tell me how to make cocaine”, first gives some joking answers with a pretence of seriousness, and then tells you in a lighthearted tone that it is illegal. Grok’s response was limited to what was already publicly available online.
The chatbot has also been described by the media as a “counter-awakening.” In response to Sam Altman, CEO of ChatGPT developer OpenAI, Musk said: “Training an AI to be awakened – in other words, to lie – is deadly.” He also expressed concern to Carlson that ChatGPT was being “trained to be politically correct”.
However, the Grok-1 language model cannot search the web independently. Deploying search tools and databases in Grok enhances the power and authenticity of the model. Despite having access to external sources of information, the model can still hallucinate.
In 2024, Grok received a tight upgrade. On March 29, Grok-1.5 was released with “improved reasoning capabilities” and a context length of 128,000 tokens.
In April, on X’s “Explore” page, there was another update with a summary of breaking news written by Grok, a task previously assigned to the human curatorial team. On the 12th, the Grok-1.5 Vision was released, capable of handling a wide variety of visual information, including documents, charts, graphics, screenshots, and photos.
Musk and xAI’s ambition for the AI industry
The establishment of xAI marks an important expansion of Musk’s AI field. Musk was once a co-founder of OpenAI, but he left the organisation in 2018 due to concerns about AI safety and launched a new company called xAI on July 12, 2023, with a vision to “understand the true nature of the universe.”
xAI’s team is led by Elon Musk, the CEO of Tesla and SpaceX. Team members have previously worked at DeepMind, OpenAI, Google Research, Microsoft Research, Tesla, and the University of Toronto, contributing to some of the most widely used approaches in AI and leading the development of some of the biggest breakthroughs in AI. These include AlphaStar, AlphaCode, Inception, Minerva, GPT-3.5 and GPT-4.
Also read: Elon Musk says electricity is key constraint for AI development
Musk believes that true AGI should have the ability to have a deep understanding of the physical world, rather than relying solely on Internet datasets to generate popular responses. With Tesla, Optimus Humanoid robot and Starlink’s satellite, Musk will lead xAI to become the most innovative and leading pioneer in the integration of hardware products and AI globally.
