Highlights
- Elon Musk unveils Grok 4, calling it “the smartest AI in the world”
- Grok 4 reportedly trained 100 times more than Grok 2
- Musk says it performs at PhD-level across nearly all subjects
- The launch follows controversy around Grok 3’s offensive responses
Musk launches Grok 4 AI model with bold claims of intelligence
Elon Musk has launched Grok 4, the latest artificial intelligence model developed by his xAI company, claiming it surpasses PhD-level intelligence across all academic fields.
Speaking at the launch, the Tesla and SpaceX CEO said Grok 4 had been trained using 100 times more data than Grok 2, which was replaced by Grok 3 in February this year. Musk described the new model as “the smartest AI in the world” and said it could achieve near-perfect results in graduate-level exams in almost every subject.
“Grok 4 is smarter than nearly all graduate students in all subjects simultaneously,” Musk said. He referred to the current pace of AI development as an “intelligence big bang” and suggested the model may soon contribute to significant technological breakthroughs.
Potential and concerns over AI ethics
While praising Grok 4’s capabilities, Musk acknowledged that the model “may lack common sense” despite its advanced reasoning skills. He stressed that the most important quality for any AI system is to be “truth-seeking” and stated that it is possible to embed values such as honesty and honour into AI, much like instilling values in a child.
Musk also predicted that the first “watchable half-hour” of television produced entirely by AI could be released by the end of 2025.
The launch of Grok 4 comes amid criticism of the previous version, Grok 3, after it generated antisemitic content, including messages that praised Adolf Hitler. In response, the official Grok account on X (formerly Twitter) said: “We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts.”
Musk distances Grok from 'woke' competitors
Musk has positioned Grok as an alternative to other chatbots such as ChatGPT and Google’s Gemini, which he has previously criticised for being “woke”. In June, he invited users on X to help train Grok using “divisive facts” — a term he used for politically incorrect but factually accurate statements.
The update also follows news that X CEO Linda Yaccarino is stepping down, nearly two years after being appointed by Musk to lead the platform he acquired for $44 billion (£32.4 billion) in 2022.
