Elon Musk said on June 22 that his company xAI will retrain its chatbot Grok AI using a new data set.
According to a post on X, Musk wants Grok AI to “rewrite the entire corpus of human knowledge.” He claimed that current models are trained on “uncorrected data” filled with “garbage.”
He stated that the new Grok 3.5 model will feature “advanced reasoning.” Musk added that the model will first rewrite human knowledge, then train itself on that revised content.
He did not clarify what information will be deleted or added, or how the rewriting process will work.
Musk criticized most existing foundation models. He said they include too much incorrect or biased content.
His solution is to use Grok AI to replace that content with what he calls “factually true” information.
Grok AI Will Use Politically Incorrect Data, Says Musk
Musk said that Grok AI should be trained on content that is “politically incorrect, but nonetheless factually true.”
He invited users on X to share “divisive facts” for training the model. Many of the replies included conspiracy theories and false claims.
Some examples included Holocaust denial, anti-vaccine misinformation, racist pseudoscience, and climate change denial.
Musk did not say how Grok AI would separate facts from misinformation in those submissions.
In 2022, Musk took over X and reduced content moderation. That led to a rise in false and misleading posts.
In response, he launched Community Notes, where users can add context to viral content. However, experts say this approach still allows a large amount of false data to circulate.
The plan to retrain Grok AI using politically incorrect content remains unverified by external reviewers. No oversight body or independent fact-checking process has been announced.
Critics Warn of Dangers in Grok AI Retraining Strategy
AI expert Gary Marcus reacted to Musk’s Grok AI announcement on X, calling it dystopian.
“Straight out of 1984,”
he wrote. He argued that Musk’s plan to rewrite history using Grok AI reflects an attempt to shape facts to fit personal views.
Bernardino Sassoli de’ Bianchi, professor of logic and science philosophy at the University of Milan, raised similar concerns.
In a LinkedIn post, he called the idea “dangerous” and said rewriting data to match ideology “is wrong on every conceivable level.”
Both Marcus and Sassoli de’ Bianchi emphasized the risks of allowing a private company to reshape human knowledge using AI. They noted the absence of details about how Grok AI will be audited or kept in check.
Grok AI, first launched by Musk’s company xAI, has been promoted as an “anti-woke” alternative to models like ChatGPT.
The company has not released technical documentation for Grok AI retraining. The upcoming Grok 3.5 version is expected soon, but no release date has been shared.
Disclosure:This article does not contain investment advice or recommendations. Every investment and trading move involves risk, and readers should conduct their own research when making a decision.
Kriptoworld.com accepts no liability for any errors in the articles or for any financial loss resulting from incorrect information.
