Elon Musk's xAI artificial intelligence lab unveiled its first AI chatbot over the weekend, dubbed Grok, which is designed to compete with OpenAI's ChatGPT and Google's Bard, but with more irony and live posts and content from X and access

. xAI Lab was launched in July after Mask became upset that other large-scale language models were being censored and restrictions placed on speech. He demonstrated Grok on Saturday and released a screenshot of himself sarcastically answering an inquiry about cocaine.

Initially, it will be offered only to a small number of invited testers in the U.S., but according to Mask, it will eventually be offered as a benefit to X Premium+ subscribers.

In a post promoting the chatbot, xAI claims that it is modeled after Douglas Adam's fictional galactic hitchhiking guide and is "intended to answer almost anything and, much harder, even suggest what questions to ask."

Grok is a term used to describe an intuitive understanding of a topic or idea, to "get it" without further elaboration. Robert A. Heinlein first used the term in his 1961 science fiction classic, "A Stranger in a Strange Land."

Unlike ChatGPT and Anthropic's Claude and Bard, Mask says Grok is designed to "answer questions with a little wit and rebelliousness," as one would expect from a billionaire. This includes encouraging designers to answer questions that would be ignored by other tools.

To promote Grok, Mask showed a screenshot of a question about how to make cocaine; where something like ChatGPT might have issued a warning for this question, Grok gave a humorous answer suggesting the need to get a chemistry degree and license.

Grok accesses live data feeds by pulling in comments, posts, links, and media from the X platform itself; it is not clear how much of the X content was used to train the model, but it likely forms a large part of the data set .

The model was trained over the past two months using Oracle hardware; the model behind ChatGPT requires months to years of training, tweaking, and retraining before release.

According to xAI, GPT-4 will be regularly enhanced with new features. The lab suggests that it can already outperform other models in its class, such as GPT-3.5, which is included in the free version of ChatGPT.

Having live access to major data sources like X can be both a plus and a minus: feeding the AI new data, querying it regularly, and writing in a more human, natural, conversational style than Wikipedia or other traditional publishing platforms Often. There is more vitriol, offensive posts, and racism in response to X posts detailing the latest scientific breakthroughs.

This may have been fine in the early days of biogenic AI, but as President Joe Biden's White House, the EU, and others seek to introduce restrictions and guardrails on AI, models that react quickly to racism or are not careful in their output will find themselves shut down or xAI may face fines.