Grok is about to launch 'Runaway Mode': Responses are even more pointed and offensive

Company As early as April last year, Musk mentioned this mysterious mode, saying that Grok would provide some "disgusting, inappropriate and offensive" responses in this mode, likening it to "an amateur who is still learning." comedian".

Currently, "Runaway Mode" has not yet been officially launched. TechCrunch tried to find this feature in Grok's web interface but failed. This new model may be part of Musk's attempt to realize Grok's original vision. When Musk first launched Grok nearly two years ago, he emphasized that it would be a fringe, unfettered and anti-"woke culture" chatbot that could answer controversial questions that other AI systems were unwilling to touch.

In a previous feature, when Grok was asked to use vulgar language, it was very helpful, fluently outputting some curse words that were inappropriate to hear in ChatGPT. However, at this stage, Grok still seems to have reservations on political topics, especially when it comes to issues such as gender identity rights, equality projects and diversity. Some studies have found that Grok's tendency is more towards the political left.

In response, Musk blamed Grok's performance on its training data and promised to make it "more politically neutral." "Unfortunately, the internet is filled with woke culture nonsense. Grok will get better, this is just a beta," he said in a December 2023 social media post.

Additionally, Musk and many of his supporters, including Trump's encryption and AI adviser David Sachs, have accused AI chatbots of censoring conservative views. Sachs singled out OpenAI's ChatGPT, saying it was "programmed to be woke" and not truthful enough on politically sensitive topics.

Overall, Grok’s new feature “Runaway Mode” has not yet been launched, but its arrival may bring more controversial and offensive responses from users, further promoting Musk’s vision for this AI system.