To get smarter AI, DARPA wants to get inside our brains
The Pentagon’s outside-the-box research group thinks pushing AI forward requires better understanding of how our own brains work.
Background: DARPA has developed human-computer interfaces that let paralyzed patients learn to move robotic limbs. But there’s a problem: the brain never stops learning and experimenting with new ways to carry out tasks, and the software that translates brain signals into commands for robotic limbs can’t keep up. In an op-ed in the Wall Street Journal, a director at DARPA says AI could help.
How to do it: DARPA wants to use reinforcement learning, a process whereby machines learn by trial and error, to improve their software. Reinforcement learning has proved great at playing video games, but everyday movements like grabbing a cup are much more complicated. To create AI that can accomplish DARPA’s goals, we will need more insight into how the brain accomplishes these tasks and others so effortlessly.
But: At the moment this kind of tech is a dream, not a reality. Tech companies are spending billions on AI, but for the most part interfacing with the brain isn’t on their agenda. And we still don’t understand much about how the brain achieves its incredible ability to keep learning and adapting. So yeah, there’s a long way to go before AI-powered robotic limbs are commonplace.
Deep Dive
Artificial intelligence
Large language models can do jaw-dropping things. But nobody knows exactly why.
And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.
OpenAI teases an amazing new generative video model called Sora
The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.
Google’s Gemini is now in everything. Here’s how you can try it out.
Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.
Google DeepMind’s new generative model makes Super Mario–like games from scratch
Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.
Stay connected
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.