Grok's 'Unhinged Mode': What You Need to Know

Grok's "Unhinged Mode" Explained

Elon Musk, the owner of X and a billionaire, initially hinted at a forthcoming “Unhinged Mode” for Grok, X’s AI chatbot, last April. Now, almost a year later, xAI – the company owned by Musk that developed Grok – has clarified details about this new functionality through an updated FAQ section on their website.

What is Unhinged Mode?

According to the FAQ, Grok operating in Unhinged Mode is designed to generate responses that are deliberately provocative, unsuitable, and potentially offensive. The description likens it to an inexperienced stand-up comedian refining their act.

Currently, the Unhinged Mode isn't accessible to users. Attempts by TechCrunch to locate the feature within the Grok web interface on X were unsuccessful.

The Original Vision for Grok

This new mode could represent Musk’s effort to realize the initial concept behind Grok.

Approximately two years ago, when introducing Grok, Musk positioned it as a chatbot that would be bold, uncensored, and resistant to prevailing “woke” ideologies. He promised a system willing to address challenging questions that other AI platforms might avoid.

To a degree, Grok has fulfilled this promise. When prompted, it can produce vulgar language and profanity, a contrast to the more restrained responses from systems like ChatGPT.

Current Limitations of Grok

However, the existing version of Grok demonstrates caution when discussing political topics and avoids crossing certain lines. A recent study indicated that Grok exhibits a leaning towards the political left on issues such as transgender rights, diversity initiatives, and economic inequality.

Musk attributes this behavior to the data used to train Grok – publicly available webpages – and has committed to adjusting the chatbot to be more politically neutral.

Addressing Perceived Bias

In a December 2023 post, Musk stated, “Unfortunately, the Internet (on which it is trained), is overrun with woke nonsense.” He further assured users that Grok would improve, characterizing the current version as a beta release.

Musk and his supporters, including David Sacks – designated as the crypto and AI “czar” for President-elect Donald Trump – have voiced concerns about AI chatbots potentially suppressing conservative viewpoints. Sacks specifically criticized OpenAI’s ChatGPT, alleging it is “programmed to be woke” and provides inaccurate information on sensitive political subjects.

Stay informed with TechCrunch’s AI newsletter! Subscribe here to receive it weekly in your inbox.