Researchers in China have developed an artificial intelligence tool capable of designing a working computer in under five hours. The team of 19 computer scientists from five different institutions made the AI breakthrough after setting out to prove that machines can create computer chips in a similar way to humans. The feat was performed 1,000 times faster than a human team could have achieved it, the researchers claimed, marking a major step towards building self-evolving machines. “Design activity... distinguishes humanity from other animals and traditional machines, and endowing machines with design abilities at the human level or beyond has been a long-term pursuit,” the scientists wrote in a paper detailing their research. “We present a new AI approach to automatically design a central processing unit (CPU), the brain of a computer, and one of the world’s most intricate devices humanity has ever designed.” The project involved the layout of an industrial-scale RISC-V CPU, capable of running the Linux operating system and achieving an accuracy of 99.99 per cent in validation tests. The AI bypassed the manual programming and verification process of the typical design cycle, which the researchers said “consumes more than 60-80 per cent of the design time and resources” of human teams. The AI was also able to autonomously make discoveries involving computer design, uncovering something called the von Neumann architecture, first invented in 1945. The overall performance of the CPU is relatively modest compared to modern computers, with the researchers saying it can perform at a similar level to a 1991 Intel 80486SX CPU. Developing the AI approach, however, has the potential to “reform the semiconductor industry by significantly reducing the design cycle”, the researchers said. The research is detailed in a study, titled ‘Pushing the limits of machine design: Automated CPU design with AI’. Leading AI chip maker Nvidia has previously used artificial intelligence to optimise its computer chip designs, publishing a new approach to AI-powered chip design in March that could significantly improve the cost and performance of CPUs. Read More 10 ways AI will change the world – from curing cancer to wiping out humanity Harvard’s new computer science teacher is a chatbot Google’s DeepMind unveils AI robot that can teach itself unsupervised Robots can now learn new skills like picking up knives by watching YouTube videos

Fortnite vaulted the popular Red-Eye Assault Rifle and brought back the MK-Alpha Assault Rifle in Chapter 4 Season 3.

It's true: You can add mentions in an Instagram Story even after it's posted. And

LOUISVILLE, Ky.--(BUSINESS WIRE)--May 22, 2023--

Here's the best F1 23 Monza setup for a race victory in Italy, including tyre pressure, brakes, and aerodynamic settings.

OSLO, Norway--(BUSINESS WIRE)--Sep 21, 2023--

A new lawsuit filed against Nvidia alleged that a file accidentally left on display by one of its employees during a video conference contains data stolen from the car tech company Valeo. The lawsuit alleged that employee Mohammad Moniruzzaman downloaded the “entirety of Valeo’s advanced parking and driving assistance systems source code” in early 2021 without authorization. It says he also downloaded “scores of Valeo Word documents, PowerPoint presentations, PDF files, and Excel spreadsheets explaining various aspects of the technology” before leaving to join Nvidia in August of that year. Valeo is accusing Nvidia of having benefited from the stolen trade secrets. “Mr Moniruzzaman ended his employment at Valeo and took the stolen source code and technical documentation with him to Nvidia, receiving a promotion to a senior position working on the software development for the very same project,” the lawsuit alleged. The alleged data theft reportedly came to light the following year when staff from both Nvidia and Valeo were working over a Microsoft Teams video conference call on a joint parking assistance project. Nvidia was developing the software, while Valeo was providing ultrasonic sensor hardware in the collaborative project for an unnamed automotive parts manufacturer. “On March 8, 2022, one of these videoconference meetings was scheduled. Mr Moniruzzaman, now employed by Nvidia, attended the video conference call… and shared his computer screen during the call,” the lawsuit noted. “When he minimized the PowerPoint presentation he had been sharing, however, he revealed one of Valeo’s verbatim source code files open on his computer,” it said. Valeo participants on the conference call recognised the source code and took a screenshot of it, the company said. “So brazen was Mr Moniruzzaman’s theft, the file path on his screen still read ‘ValeoDocs,’” the lawsuit noted. Mr Moniruzzaman had admitted to stealing Valeo’s software and using it while employed at Nvidia when he was questioned by German police, according to Valeo’s lawsuit. Nvidia has responded that it has no interest in using the stolen code, according to The Verge. But Valeo alleged that its competitor has benefited from the data, which it says would save “millions of dollars in development costs”. Citing these reasons, Valeo is seeking recovery of damages and an injunction to stop Nvidia and its staff from using its trade secrets. The Independent has reached out to Mr Moniruzzaman through Nvidia for a comment. Read More Trump’s Truth Social sues 20 media outlets over financial loss reports AI breakthrough could help us build solar panels out of ‘miracle material’ ‘We let you down’: Peloton apologises for Thanksgiving ride AI breakthrough could help us build solar panels out of ‘miracle material’ ‘We let you down’: Peloton apologises for Thanksgiving ride HSBC app not working on one of the biggest shopping days of the year

Amazon just dropped refurbished device deals ahead of Prime Day. Here are the best ones:

SAVE 49%: ExpressVPN is the best VPN for gaming. A one-year subscription to ExpressVPN is

Listen to The Big Take podcast on iHeart, Apple Podcasts, Spotify, Terminal. Millions of Americans depend on public

BOSTON--(BUSINESS WIRE)--May 9, 2023--

LEHI, Utah--(BUSINESS WIRE)--May 17, 2023--