Jensen Huang predicts: Physic AI, the next disruptive wave of artificial intelligence

At the theme event of the Advanced Manufacturing Chain of the 3rd Chain Expo, Jensen Huang, founder and CEO of NVIDIA, said that the next wave of artificial intelligence is physical AI. Artificial intelligence is a new way to make software. It predicts results by using algorithms instead of human coding based on basic principles. We use algorithms to learn how to predict results from examples, information, and data. This method of using computers to learn how to conduct projects has been proven to be very scalable.

Jensen Huang said that generative AI is the second wave of artificial intelligence. It allows us to convert from one modality to another, from English to Chinese, from English to pictures, from pictures to English, and from Chinese to videos. Now artificial intelligence can understand information and generate information. The state we are in now is incredible because artificial intelligence can understand and solve problems it has never seen before, just like humans, and can break down problems step by step to solve problems that have never been really solved before. The next wave of artificial intelligence is physical AI-Physic AI, and AI can enter physical machines such as robots.

How to understand Physic AI?

Looking back at the development of AI, from the early discriminative AI, which focused on the identification and classification of data, helping machines to achieve basic image and voice recognition; to the rapidly rising generative AI, represented by DALL-E and ChatGPT, which relied on given prompts to generate new texts, images and even videos, completely changed the way of content creation and interaction, and transformed AI from a simple understander to a creator.

Today, we are in the process of the development of Agentic AI. This type of AI can make decisions, plan and perform tasks autonomously in complex environments, as if it has independent "thinking" ability, and has begun to show its edge in the fields of intelligent customer service and smart home.

According to Jensen Huang's vision, Physic AI will be the next chapter in the development of AI. Physic AI, in simple terms, is to enable AI to have the ability to understand and abide by various physical laws, and to achieve dynamic interaction and autonomous operation between the virtual and real worlds through the collaboration of embodied physical systems (such as robots, self-driving cars, etc.) and disembodied intelligent systems (such as cloud-based intelligent algorithms).

For example, robots in factories can accurately understand and perform complex assembly tasks. When encountering emergencies (such as missing parts and equipment failures), they can adjust the operation process autonomously based on physical principles and real-time environmental information to solve problems efficiently. Autonomous driving cars can not only identify road signs and obstacles under complex road conditions, but also accurately plan driving paths based on the principles of mechanics and dynamics to achieve safe and smooth driving. These scenarios are the vivid embodiment of Physic AI in the real world.

Jensen Huang's understanding

The reason why Jensen Huang is so confident about the future of Physic AI is not groundless, but is supported by deep technical accumulation and market insights.

From a technical perspective, many key technologies in the current AI field are developing rapidly, providing fertile soil for the implementation of Physic AI. On the one hand, the continuous evolution of large language models has made its ability to understand and generate complex semantics more powerful. These models can accurately convert abstract instructions contained in human language into operation processes that conform to the rules of the physical world, building a bridge for efficient communication between Physic AI and humans. For example, in a smart factory, workers only need to give instructions to robots through voice, such as "assemble this batch of parts into products according to a specific mechanical structure", and robots equipped with Physic AI can quickly plan and execute assembly tasks by relying on their understanding of language and their own "mastery" of physical principles.

On the other hand, the rapid progress of sensor technology has provided Physic AI with "eagle eyes" to perceive the real world. High-resolution cameras, high-precision lidar, sensitive tactile sensors, etc. can collect massive amounts of environmental information in real time, from the shape, color to material, texture, spatial position to motion trajectory of objects, everything. These rich data sources are continuously input to AI, enabling it to build an accurate model of the physical world and make scientific decisions based on it. Taking self-driving cars as an example, multiple sensors work together to enable vehicles to perceive the surrounding road conditions, the dynamics of other vehicles and pedestrians in real time, and then combine with Physic AI algorithms to achieve safe and efficient driving.

From the perspective of market demand and application prospects, Physic AI also shows great potential. In the manufacturing industry, with the continuous rise in labor costs and the increasing requirements for production efficiency and product quality, enterprises have an increasingly urgent need for intelligent and automated production. Physic AI-driven robots can undertake high-precision and highly repetitive production tasks, greatly improve production efficiency, and reduce the defective rate. At the same time, they can also operate in dangerous environments (such as high temperature, high pressure, toxic and harmful) to ensure the safety of workers. According to relevant research institutions, in the next five years, the global manufacturing industry's investment in Physic AI-related technologies and equipment will grow at a rate of more than 20% per year.

In the medical field, surgical robots can achieve more precise operations than human doctors with the help of Physic AI technology. Taking heart bypass surgery as an example, surgical robots equipped with Physic AI can accurately control surgical instruments and complete blood vessel connections based on the real-time three-dimensional model of the patient's heart, greatly improving the success rate of surgery and reducing postoperative complications. In addition, Physic AI also has a broad application space in rehabilitation therapy, telemedicine and other aspects, and can provide patients with personalized and high-quality medical services.

What will we usher in the era of Physic AI?

If Physic AI really becomes the next wave of AI development as Jensen Huang predicted, then our lives and society will usher in huge changes. In daily life, smart homes will become more intelligent and considerate. Household robots equipped with Physic AI can not only complete daily housework such as sweeping, mopping, and washing dishes, but also quickly call emergency numbers and take preliminary emergency measures based on human physiological characteristics and physical environment information when the elderly suddenly fall ill; when children encounter difficulties in learning, they will become smart tutors to help children overcome difficulties through understanding of knowledge and physical experiment simulation.

In terms of urban transportation, autonomous driving technology will usher in a qualitative leap. Vehicles supported by Physic AI can achieve truly intelligent collaborative driving. Real-time information exchange between vehicles and between vehicles and road infrastructure, dynamically adjust driving speed and route according to factors such as traffic flow, road conditions, and weather, completely solve traffic congestion problems, and significantly reduce the incidence of traffic accidents. By then, people's travel will become more convenient, efficient, and safe.

From the macro perspective of social development, Physic AI will promote a new round of industrial revolution and give birth to countless new business models and employment opportunities. In the logistics industry, with the help of Physic AI, intelligent warehousing and distribution systems can realize the automated storage, sorting and transportation of goods, improve logistics efficiency and reduce costs. At the same time, in order to support the development of Physic AI, from chip research and development, algorithm optimization to sensor manufacturing and system integration, each link will require a large number of professional talents to inject new vitality into the employment market.

Of course, Physic AI is also accompanied by many challenges and risks. From a technical perspective, how to ensure that Physic AI always maintains the accuracy and reliability of decision-making in a complex and dynamic physical environment is a difficult problem that needs to be overcome. The real world is full of uncertainty, and a small environmental change (such as a sudden change in light, a slight difference in the material of an object) may affect the perception and decision-making of AI. In addition, the security of the Physic AI system is also a top priority. Once it is attacked by hackers, the consequences are disastrous. For example, if a malicious attacker tampers with the Physic AI algorithm of an autonomous vehicle, it may cause the vehicle to lose control and cause a serious traffic accident.

From a social ethical perspective, the widespread use of Physic AI may trigger a series of ethical controversies. As robots gradually replace humans in jobs, how to solve the resulting unemployment problem and how to redefine the roles of humans and machines in social production are issues that need to be thought about in depth. At the same time, when robots can make decisions independently, the definition of responsibility becomes unclear. For example, when an accident occurs in an autonomous vehicle, who should be held responsible, the car manufacturer, the programmer, or the vehicle user?

Jensen Huang's assertion that Physic AI will become the next wave of artificial intelligence paints an exciting blueprint for future technology. The development of Physic AI will be like a new technological revolution, profoundly changing our lives, work and society. Of course, the road to development will be full of uncertainties, but the revolution brought about by technological progress will definitely completely change our lives.