recent Aye The models are surprisingly humanlike in their ability to generate text, audio, and video when prompted. However, until now these algorithms have been largely limited to the digital world rather than the physical, three-dimensional world in which we live. In fact, whenever we try to apply these models to the real world, even the most sophisticated struggle to perform adequately. For example, just think how challenging it has been to develop safe and reliable self-driving cars. Despite being artificially intelligent, these models not only have no understanding of physics, but they also often have hallucinations, which leads them to make inexplicable mistakes.
However, this is the year AI will finally arrive Take the leap from the digital world to the real world we live inExtending AI beyond its digital limits requires reworking the way machines think, combining the digital intelligence of AI with the mechanical power of robotics. This is what I call “physical intelligence,” a new form of intelligent machine that can sense dynamic environments, cope with unpredictability, and make decisions in real time. Unlike the models used by standard AI, physical intelligence is rooted in physics; Understand fundamental principles of the real world, such as cause-and-effect.
Such characteristics allow physical intelligence models to interact with and adapt to different environments. In my research group at MIT, we are developing models of physical intelligence that we call fluid networks. For example, in one experiment, we trained two drones – one driven by a standard AI model and the other by a fluid network – to detect objects in the forest during the summer using data captured by human pilots. While both drones performed equally well when tasked with doing exactly what they were trained to do, when they were asked to detect objects in different conditions – during winter or in an urban setting – only liquid The network drone successfully completed its task. This experiment showed us that, unlike traditional AI systems, which stop evolving after their initial training phase, Liquid Networks continue to learn and adapt from experience, just like humans.
Physical intelligence is also able to interpret and physically execute complex commands received from text or images, bridging the gap between digital instructions and real-world execution. For example, in my lab, we have developed a physically intelligent system that, in less than a minute, can train small robots based on signals such as “robot that can walk” or “robot that can catch up.” Iteratively design and then 3D-print. objects”.
Other laboratories are also making important breakthroughs. For example, robotics startup Covariant, founded by UC-Berkeley researcher Peter Abele, is developing a chatbot – similar to ChatGTP – that can control robotic arms when given a signal. They have already secured over $222 million to develop and deploy sorting robots in warehouses globally. A team from Carnegie Mellon University has also done this recently Exhibited A robot with just a single camera and precise actuation can execute dynamic and complex parkour movements using a single neural network trained through reinforcement learning – including jumping over jumps twice its height and over intervals twice its length.
If 2023 was the year of text-to-image and 2024 was text-to-video, then 2025 will mark the era of physical intelligence, with a new generation of devices — not just robots, but anything from power grids to smart homes. . -Can interpret what we are telling them and perform tasks in the real world.