Musk's XAI needs to retrain Grok, experts criticize: deliberately distorting history

Elon Musk ( expressed his dissatisfaction on Twitter ) X( on June 22, regarding his AI model Grok producing "garbage" and "incorrect content." Therefore, he plans to allow Grok to rewrite the human knowledge base by itself and then use this data to retrain Grok. The news immediately sparked criticism from the outside.

Grok 3.5 Rewrite Human Knowledge Base

Elon Musk first revealed that his AI company xAI will make significant adjustments to Grok. He pointed out:

"We will use Grok 3.5, which possesses advanced reasoning capabilities, to rewrite the entire human knowledge base, filling in the missing information and deleting incorrect content. We will then use this data to retrain the model."

He believes that all AI models currently on the market rely too heavily on uncalibrated training data and are filled with too much junk content.

Experts have continuously criticized Musk, claiming he is "deliberately manipulating history."

Elon Musk's post immediately sparked strong backlash from experts. Gary Marcus, a professor emeritus at New York University, replied beneath the post:

"This is basically like a plot from the novel '1984'. You can't make Grok align with your personal views, so you might as well rewrite the entire history to fit what you think."

Professor Bernardino from the University of Milan in Italy also vented on LinkedIn:

"Using ideology to modify AI training data is wrong on all levels!"

) Note: "1984" is a strong critique of totalitarianism, thought control, historical revisionism, and speech repression. The terrifying surveillance society depicted in the book is still widely referenced to criticize government abuse of power or information manipulation. (

It's frustrating that too many AI models are overly politically correct; Grok is the only truth.

Elon Musk has always believed that many mainstream AI models, such as ChatGPT, are overly politically correct and deliberately avoid certain sensitive topics that are not in line with the mainstream. Therefore, he has repeatedly emphasized the need to create AI tools that are anti-political correctness )Anti-Woke(, and Grok was born for this purpose.

Since Musk bought Twitter )X( in 2022 and relaxed the platform's content moderation and misinformation rules, there has been a surge of conspiracy theories, extreme rhetoric, and fake news, with him personally retweeting many extreme statements.

Once promoted "Community Notes" to combat fake news, but the effect was limited.

In response to the issue of rampant misinformation, Musk has launched "Community Notes" ), hoping to allow users to provide additional explanations or clarifications under controversial posts.

However, the outside world generally believes that such a mechanism cannot truly and effectively suppress misinformation, but rather allows the platform to freely spread more unverified statements.

(Elon Musk's xAI engineer reportedly asked to delete a post criticizing Grok 3 for being inferior to OpenAI, leading to a controversy over freedom of speech; the engineer subsequently resigned.)

This article discusses how Musk's xAI aims to retrain Grok, with experts criticizing it for deliberately distorting history. It first appeared in Chain News ABMedia.