https://www.livescience.com/technology/computing/humans-cannot-really-understand-them-weird-ai-designed-chip-is-unlike-any-other-made-by-humans-and-performs-much-better Skip to main content (*) ( ) Open menu Close menu Live Science [ ] Search Search Live Science [ ] Sign in * * View Profile * Sign out Subscribe RSS * * Space * Health * Planet Earth * Animals * Archaeology * Physics & Math * Human Behavior * Technology * Chemistry * More + Science news + Life's Little Mysteries + Science quizzes + About us + Newsletter + Follow us + Story archive How It Works Magazine How It Works Magazine Why subscribe? * The ultimate action-packed science and technology magazine bursting with exciting information about the universe * Subscribe today and save an extra 5% with checkout code 'LOVE5' * Engaging articles, amazing illustrations & exclusive interviews * Issues delivered straight to your door or device From$26.49 View Trending * Invisible DNA * Science news this week * Asteroid 2024 YR4 'just the tip of the iceberg' * AI-designed computer chips * Polar vortex 1. Computing AI-designed chips are so weird that 'humans cannot really understand them' -- but they perform better than anything we've created News By Tim Danton published 20 February 2025 AI models have, within hours, created more efficient wireless chips through deep learning, but it is unclear how their 'randomly shaped' designs were produced. * * * * * * * Comments ( 0 ) () When you purchase through links on our site, we may earn an affiliate commission. Here's how it works. Two people in a lab using a microscope to view the chip on a monitor. The AI treated the chip design as one complete system rather than a collection of parts. (Image credit: Princeton University) Engineering researchers have demonstrated that artificial intelligence (AI) can design complex wireless chips in hours, a feat that would have taken humans weeks to complete. Not only did the chip designs prove more efficient, the AI took a radically different approach -- one that a human circuit designer would have been highly unlikely to devise. The researchers outlined their findings in a study published Dec. 30 2024 in the journal Nature Communications. The research focused on millimeter-wave (mm-Wave) wireless chips, which present some of the biggest challenges facing manufacturers due to their complexity and need for miniaturization. These chips are used in 5G modems, now commonly found in phones. Manufacturers currently rely on a mix of human expertise, bespoke circuit designs and established templates. Each new design then goes through a slow process of optimization, based on trial and error because it is often so complex that a human cannot fully understand what is happening inside the chip. This leads to a cautious, iterative approach based on what has worked before. Related: 6G speeds hit 100 Gbps in new test -- 500 times faster than average 5G cellphones In this case, however, researchers at Princeton Engineering and the Indian Institute of Technology posited that deep-learning-based AI models could use an inverse design method -- one that specifies the desired output and leaves the algorithm to determine the inputs and parameters. The AI also considers each chip as a single artifact, rather than a collection of existing elements that need to be combined. This means that established chip design templates, the ones that no one understands but probably hide inefficiencies, are cast aside. Sign up for the Live Science daily newsletter now Get the world's most fascinating discoveries delivered straight to your inbox. [ ][ ]Contact me with news and offers from other Future brands[ ]Receive email from us on behalf of our trusted partners or sponsors[Sign me up] By submitting your information you agree to the Terms & Conditions and Privacy Policy and are aged 16 or over. The future of chip design? In this experiment, the resulting structures "look randomly shaped," said lead author Kaushik Sengupta, a professor of electrical and computer engineering at Princeton. "Humans cannot really understand them." And when Sengupta'steam manufactured the chips, they found the AI creations hit performance levels beyond those of existing designs. An enlarged image of the chip's circuitry shows unusual patterns. (Image credit: Princeton University) Although the findings suggest that the design of such complex chips could be handed over to AI, Sengputa was keen to point out that pitfalls remain "that still require human designers to correct." In particular, many of the designs produced by the algorithm did not work- equivalent to the "hallucinations" produced by current generative AI tools. RELATED STORIES --AI could crack unsolvable problems -- and humans won't be able to understand the results --Mathematicians devised novel problems to challenge advanced AIs' reasoning skills -- and they failed almost every test --AI 'hallucinations' can lead to catastrophic mistakes, but a new approach makes automated decisions more reliable "The point is not to replace human designers with tools," said Sengputa. "The point is to enhance productivity with new tools." The speed with which iterative designs can be developed opens up new possibilities, too. Some chip designs can be geared towards energy efficiency, others to outright performance or to extending the frequency range. Wireless chips are of growing importance, with an ever-growing demand for miniaturization, so this research is a valuable step forward. But Sengupta said that if his team's method can be extended to other parts of a circuit's design, it could change the way we design electronics in the future. "This is just the tip of the iceberg in terms of what the future holds for the field." Tim Danton Tim Danton Social Links Navigation Tim Danton is a journalist and editor who has been covering technology and innovation since 1999. He is currently the editor-in-chief of PC Pro, one of the U.K.'s leading technology magazines, and is the author of a computing history book called The Computers That Made Britain. He is currently working on a follow-up book that covers the very earliest computers, including The ENIAC. His work has also appeared in The Guardian, Which? and The Sunday Times. He lives in Buckinghamshire, U.K. You must confirm your public display name before commenting Please logout and then login again, you will then be prompted to enter your display name. Logout More about computing Microsoft's Majorana 1 quantum computing chip Breakthrough quantum chip that harnesses new state of matter could set us on the path to quantum supremacy 3d rendered illustration of atoms with blue and red lines. Quantum simulation breakthrough will lead to 'discoveries impossible in today's fastest supercomputers,' Google scientists claim Latest An image of red swirling dust and sparkling stars in outer space Space photo of the week: James Webb telescope reveals mysterious 'light echo' in the broken heart of Cassiopeia See more latest Most Popular [missing-im] Humanoid 'Protoclone' robot twitches into action while hanging from ceiling in viral video [missing-im] 'Just the tip of the iceberg': Why risky asteroids like 2024 YR4 will pester Earth for decades to come [missing-im] Deadly motor-neuron disease treated in the womb in world 1st [missing-im] How to watch 'The Americas': TV and streaming as Tom Hanks and Hans Zimmer join forces for new docu-series [missing-im] Invisible DNA lurks everywhere in the environment -- and we're on the verge of decoding its secrets [missing-im] What is double pneumonia? Pope Francis's diagnosis explained. [missing-im] Savonoski Crater: The mysterious, perfectly round hole in Alaska that scientists can't explain [missing-im] Asteroid YR4 impact odds plummet as NASA changes threat level of 'city-killer' [missing-im] 2 'exclusively indoor' cats in Michigan caught bird flu, potentially from their owners [missing-im] 14-year-old known as 'the human calculator' breaks 6 math world records in 1 day [missing-im] Decapitated woman found in Irish bog is 2,000-year-old ritual sacrifice LATEST ARTICLES 1. An image of red swirling dust and sparkling stars in outer space 1 Space photo of the week: James Webb telescope reveals mysterious 'light echo' in the broken heart of Cassiopeia 2. 2 Why do cats have 'toe beans'? 3. 3 James Webb Space Telescope reveals how a cosmic 'Phoenix' cools off to birth stars 4. 4 Crocodiles are overheating due to climate change -- and it's changing their behavior 5. 5 Why America is losing its 50-year 'war on cancer,' according to scientist Nafis Hasan Live Science is part of Future US Inc, an international media group and leading digital publisher. Visit our corporate site. * About Us * Contact Future's experts * Terms and conditions * Privacy policy * Cookies policy * Accessibility Statement * Advertise with us * Web notifications * Careers * Editorial standards * How to pitch a story to us (c) Future US, Inc. Full 7th Floor, 130 West 42nd Street, New York, NY 10036. [] Close Please login or signup to comment Please wait... Login Sign Up