The technical director of the Musk company is accusing an anonymous employee of the change.
The internal censorship of Elon Musk’s Grok language model raised an uproar in the X service.
Elon Musk’s language model Grok temporarily censored sources criticizing Donald Trump or Musk. Censorship is shown in the Grok’s public system prompt, where the language model is prohibited from referring to the sources mentioned by Trump or Musk to be distributed to the sources of the Miss Information.
Among other things The Verge.
The technical director of XAI Igor Babuschkin In the X service, the company has removed the language model after feedback from users. Babuschkin claims that Musk himself had not pushed for change.
– When people pointed out about a problematic prompt, we immediately changed it. Elon was never involved, Babuschkin says tweet.
According to Babuschkin, the body change was added by an individual employee. Babuschkin believesthat the error was, after all, the cause of Openai.
“The worker who made the change was a former Openai employee who has not yet fully adopted XAI’s culture,” says Babuschkin.
Some users criticize Babuschkin’s way of dealing with the problem. For example Joanne Yang He replied that public prosecution of an individual employee does not create a psychologically safe work environment.
“If I were you, I wouldn’t throw anyone under the bus so publicly,” Jang says.
