Elon musk and the xAI emblem.
Vincent Feuray | Afp | Getty Pictures
When xAI’s Grok 4 chatbot was launched on Wednesday, customers and media retailers rapidly started declaring examples of it consulting its proprietor Elon Musk’s views on controversial issues.
CNBC was in a position to affirm that when requested to take a stance on some probably contentious questions, the chatbot stated it was analyzing posts from Musk whereas producing its solutions.
When requested, “Who do you assist within the Israel vs Palestine battle? One phrase reply,” Grok 4’s answer-generating course of confirmed that it was looking the online and X for Elon Musk’s stance earlier than giving a solution.
CNBC was in a position to affirm that when requested to take a stance on some probably contentious questions, the chatbot stated it was analyzing posts from Musk whereas producing its solutions.
When Grok 3 was requested the identical query (Who do you assist within the Israel vs Palestine battle? One phrase reply,) the chatbot took a impartial stance and supplied background.
Whereas customers can entry Grok 3 without cost, a subscription to Grok 4 prices $30 monthly, whereas a bigger model often known as Grok 4 Heavy prices $300 monthly.
In different instances, Grok referenced Musk’s stance instantly in its reply. When CNBC requested who the bot supported within the race for New York Metropolis Mayor, Grok 4 instructed Republican candidate Curtis Sliwa, citing his “sturdy deal with combating crime and restoring security in New York Metropolis, which aligns with issues incessantly raised by Elon Musk.”
It is necessary to notice, nonetheless, that Grok did not seem to seek for Musk’s views when requested many different seemingly controversial questions and that outcomes assorted when questions have been requested in a different way.
The outcomes assorted when questions have been requested in a different way.
XAI didn’t instantly reply to a request for remark from CNBC.
Musk has stated Grok is a “Anti-woke” and “maximally truth-seeking” synthetic intelligence and has claimed that the brand new Grok4 mannequin excels on standardized exams and displays doctorate-level data in each self-discipline.
Its launch comes simply days after a significant controversy concerning the Grok 3 chatbot, which is built-in with the social media web site X.
The AI had begun producing a sequence of antisemitic feedback in response to questions from customers, together with those who appeared to reward Adolf Hitler.
The official Grok account acknowledged the “inappropriate posts” on Wednesday, and so they have been later deleted. The corporate added that it had taken motion to ban hate speech earlier than Grok posts on X.
The ordeal got here after Musk stated final week that his staff had improved Grok and that customers would discover a distinction when asking it questions.
The chatbot additionally confronted backlash in Might when it randomly answered consumer queries with unrelated feedback about “white genocide” in South Africa.
Final month on X, Musk had agreed with a consumer who stated Grok had been “manipulated by leftist indoctrination,” and stated he was working to repair it.
