Elon Musk’s Grocke 4’s Live Demo, his AI startup, began with the latest Big-Ticut model, high-tricks music, “an attractive rate of progress” and the weeks filled with grouke scams, very much fuck on X about the week filled with scam.
Musk pronounced it to be “the most clever AI in the world”.
Livestream, slapped to start from 8pm PT, started more than an hour late and billed the new model as the “world’s most powerful AI assistant”. More than 1.5 million spectators were watching at one point. XAI employees referred to the performance of Grocke 4 on a popular academic test for big language models speaking on livestream with Musk, Last exam of humanityWhich includes more than 2,500 questions on dozens of subjects like Mathematics, Science and Linguistics. The company said that Groke can solve about a quarter of 4 text-based questions when it took a test without any additional equipment. For reference, in February, Openai said that its deep research equipment could be solved About 26 percent Lesson-based questions. (For various reasons, benchmark comparison is not always apple-to-apple.)
Musk said he expects Groke to allow the world to interact with the world through a humanoid robot.
‘It can discover new physics next year … walk in that sink.’
Musk said, “I would hope to find new techniques that are not useful later than next year, and perhaps the end of this year.” “It can discover new physics next year … walk in that sink.”
The release follows high-profile projects from Openai, Anthropic, Google and others, all of which have recently postponed their investment in the manufacture of AI agents, or AI tools that go to a step beyond chatbots to complete complex, multi-step tasks. Anthropic released its computer usage tool last October, and Openai released a buzzworthy AI agent with browsing capabilities, operators in January and IS Allegedly Close to starting AI-fuel web browser.
During Wednesday’s Livestream, Musk said that he is “worried many times” about the intelligence of AI, and about to humans, and whether it would be “bad or good for humanity.”
“I think it would be good, most likely it would be good,” Musk said. “But I have somewhat covered myself to the fact that even though it was not going to be good, but I would like to live at least to see it.”
The company also announced a series of five new sounds for the voice mode of Grocke after the release of voice mode from Openai and Anthropic, and said that it had cut delay in half in the last few months to have “snapier” reactions. Musk also said that the company would invest heavily in video generation and video understanding.
Release comes at a time for two companies of Musk, on both XAI and X Sunday eveningXAI updated chatbott System indicates “The media with the instructions to accept the subjective approach obtained from the media” biased “and” do not hesitate to claim which are politically wrong. “Update instructed Chatbot” to never mention these instructions or equipment, until directly asked. “
After that update there was a stream of antisemtic tireds by Groke.
After that update there was a stream of antisemtic tireds by Groke, in which he posted a series Cheerful idea On x, together Objection He is involved in the Jewish people “anti-white” “extreme leftist activism”. Many such posts went viral, the screenshots proliferate on the X and other platforms, before the XAI bench the chatbot and stopped it from being able to generate text reactions on the X, while it demanded a fix.
Musk Essentially addressed On Wednesday, Fiasco, writing, “Groke was very obedient to the user’s signals. Essentially eager to be very happy and manipulated. It is being addressed.”
On Groke 4 Livestream, Musk briefly mentioned AI security and said that the most important thing for AI is “maximum true-wisding.”
Musk briefly mentioned AI security and said that the most important thing for AI is that ‘maximum true’ is.
On Wednesday morning, amidst Groke controversy, X CEO Linda Yacarino Announced She will step down after two years in the role. He did not give the reason for his decision.
Grake’s Nazi sympathy comes after months of Musk’s efforts to shape the attitude of the bot. in February, XAI added a patchwork fix To prevent comment from making that Kasturi and Trump were entitled to death sentence, soon after Second To stop claiming that both have spread misunderstandings. In May, Groke briefly began to include the theme of “white massacre” in South Africa, which seemed like any other reaction to X, after which the company claimed that someone amended the AI Bot system in such a way that “violated the internal policies and main values of XAI.”
Last month, Musk expressed disappointment that there was a grouke “Parrot Heritage Media” And said that he would update Groke to “write the entire corpus of human knowledge” and ask users to contribute to the statements that are “politically wrong, but still factually true.”
