Elon Musk announced that his AI company xAI will retrain its AI model, Grok, on a new knowledge base purified of "garbage" and "uncorrected data" by first having it rewrite history. The upcoming Grok 3.5 model is said to feature "advanced reasoning" and aims to correct errors, add missing information, and eliminate bias. Musk envisions Grok as an "anti-woke" AI free from political correctness, a stance consistent with his approach to moderating Twitter's content since acquiring it.
This plan has drawn criticism from AI experts, including Gary Marcus, who likened it to a dystopian rewriting of history tuned to Musk's personal beliefs. University of Milan professor Bernardino Sassoli de’ Bianchi also called the plan dangerous due to the risk of controlling narratives rather than promoting innovation. Musk invited X users to submit "divisive facts" that are "politically incorrect but factually true" to train Grok, which led to a mixture of conspiracy theories and debunked misinformation appearing in the discourse.
The initiative attracted notable industry attention, including from Binance CEO Changpeng Zhao (CZ), who expressed concerns about potential biases introduced when an AI attempts to rewrite human knowledge. CZ noted the complexities in historical and factual data could lead to subjective interpretations. Despite the controversy, no immediate financial or market impacts tied to Grok AI have been observed, and the project's scale on a social platform like X is unprecedented.