Is Grok getting the so-called “Unhinged” AI feature soon?

xAI’s website shows an updated FAQ that provides insight into Grok’s so-called Unhinged Mode. The update states that Unhinged Mode is meant to create responses “intended to be objectionable, inappropriate, and offensive.”

The FAQ section compares this feature to an amateur stand-up comedian “still learning the craft.” However, the feature doesn’t appear to be live yet.

In April last year, Elon Musk, the billionaire behind X, also tweeted a teaser for the feature of the AI chatbot developed by his company xAI.

Coming soon, Grok will offer “Unhinged Mode” news 🤣🤣 https://t.co/Utwd5xbQU6

— Elon Musk (@elonmusk) April 5, 2024

Musk’s route toward “Unhinged Mode” could be to fulfill Grok’s “anti-woke” vision.

Two years ago, when Musk introduced Grok, he said it was bold, raw, and unafraid to call out politically correct AI systems. Unlike most chatbots, Grok was offered as edgy and willing to broach controversial subjects. True to that pitch, the chatbot offers vulgar and colorful responses when prompted, in distinction to more restrained systems like ChatGPT.

Despite advocating for being unfiltered, the chatbot shows bias in certain political subjects and will avoid crossing lines altogether. Even a study showed that Grok usually sides with the left regarding issues like transgender rights, diversity initiatives, and social inequality.

Elon Musk pointed fingers towards its training data, which relies on publicly available content from the web. But he’s also promised to pivot the chatbot from a more politically neutral stance.

Musk wrote in a December 2023 post that the Internet, from which Grok gets its training data, is “overrun with woke nonsense.” He promised Grok would improve as this is just the beta version.

Source