Scientists Built Smarter Chips That Save Power and Money

Imagine if your phone’s brain could do simple math when you’re just browsing social media, then switch to super-precise mode for editing photos. That’s the idea behind a new chip design called CircuitsDNA. Instead of using fixed rules for how chips calculate numbers, this system uses a kind of digital evolution—like breeding the best-performing chip designs—to create chips that can change their accuracy on the fly.

The big advantage? Energy savings. In tests, these chips used up to 56% less power than traditional chips while handling AI tasks like image recognition. That could mean your phone or laptop battery lasts much longer between charges. For companies, it could also cut costs by reducing the amount of energy needed to run data centers.

There’s a trade-off, though. These chips sometimes make tiny mistakes in calculations to save energy—but the errors are so small they don’t affect most tasks. For example, in tests with AI image recognition, the accuracy loss was less than 2%. The system is also faster at designing these chips than older methods, cutting design time by up to 33%.

So why does this matter to you? If this technology becomes widely used, it could lead to cheaper, longer-lasting gadgets and more efficient AI tools. But don’t expect to see it in stores tomorrow—this is still in the research phase.