On September 22, at the opening ceremony of the inaugural Immersive Consumption Carnival and the 2026 AI Terminal Innovation and Development Conference, Zhao Qinp, Academician of the Chinese Academy of Engineering and Honorary Chairman of the Virtual Reality Industry Professional Committee of the China Computer Industry Association, delivered a keynote speech. He stated that global AI technology is undergoing a critical transition from a "parameter race" to "scenario implementation." AI is no longer just a profound model in the cloud, but is entering our daily lives, work, and perception at an unprecedented speed. He innovatively proposed the concept of "Human-Embodied Intelligence," aiming to answer the core proposition of "how to make intelligence better understand humans and couple more effectively with them."
From "Embodied Intelligence" to "Human-Embodied Intelligence"
Over the past two years, "Embodied Intelligence" has become one of the most resounding keywords in the industry. From Boston Dynamics' humanoid robots to Tesla's Optimus, and then to the rapid rise of domestic star enterprises such as Unitree Robotics and Zhiyuan Robot, academia and industry are vigorously promoting robots to achieve autonomous perception, decision-making, and execution in the physical world. Zhao Qinp explained: "The core logic of this direction is to give AI a 'body' and enable robots to truly possess a 'brain' with intelligence. Essentially, it is a machine-centric intelligence path."
However, when people shift their focus from industrial workshops and professional services to the daily lives of ordinary people, they will find a critical issue: in the next decade, larger-scale and deeper AI popularization will be driven not only by robots but also by intelligent terminals that closely accompany humans.
"This is precisely the fundamental starting point for proposing 'Human-Embodied Intelligence'," Zhao Qinp further elaborated. "If 'Embodied Intelligence' answers 'how to make machines do things more smartly,' then 'Human-Embodied Intelligence' answers 'how to make intelligence better understand humans, couple more effectively with them, and make humans more powerful.' Intelligence is not only endowed to robots but also surrounds humans; it is human-centric, with all intelligent hardware designed and evolved around human needs, experiences, lives, work, and creativity."
The Connotation of "Human-Embodied Intelligence" is Human-Centric
"Human-Embodied Intelligence," as the name suggests, means "AI and humans share the same body, with AI and the human brain forming a dual-brain coupled synergy." Zhao Qinp summarized its core connotation into three principles:
First, human-centric. Unlike "Embodied Intelligence," which gives AI a physical form, "Human-Embodied Intelligence" is anchored in the human first-person perspective. AI is no longer an independent replacement but a collaborative partner continuously guided by human intent during perception, decision-making, and action. Its design philosophy is no longer about function stacking, but about understanding your needs.
Second, multi-form carrier. "Human-Embodied Intelligence" is not an abstract concept but is implemented in specific terminals such as AI glasses, bracelets, rings, pendants, watches, AI PCs, and AI smartphones. These devices are close-fitting, continuous, and wearable, serving as extensions of human capabilities and senses. Future AI terminals will evolve from "passive tools" to "active companions."
Third, accompanied evolution as the goal. "Human-Embodied Intelligence" is not a static capability but an intelligent agent that learns throughout life and continuously evolves. It can adaptively adjust according to users' preferences, habits, and physical conditions, achieving the effect of "the more you use it, the better it understands you, and the more tacit you become."
"This concept can be extended and called 'Companion Intelligence'—it emphasizes the word 'companionship,' representing the long-term connection between intelligent terminals and humans in the time dimension, and the deep integration with humans in the emotional and cognitive dimensions," Zhao Qinp added.
AI Terminals Will Accelerate Their Evolution Along Four Main Lines
Under the concept of "Human-Embodied Intelligence," Zhao Qinp predicted that the new generation of AI terminals will accelerate their evolution along four main lines.
First, diversified forms and segmented scenarios. AI terminals will no longer be dominated solely by smartphones but will form a multi-form matrix around various perception channels such as vision, hearing, touch, smell, and motion. AI glasses and AI pendants reconstruct the way information is acquired, AI rings monitor health indicators, AI bracelets accompany sports and sleep, AI PCs become the intelligent extension of productivity, and AI smartphones continue to serve as the central hub of personal computing power. "Strengthening edge-side capabilities will be key—large models will continuously migrate to terminals, and lightweight models with tens of billions of parameters will enable terminals to possess independent reasoning and decision-making capabilities," Zhao Qinp emphasized.
Second, natural interaction and seamless experience. Future AI terminals will break free from the reliance on screens and fingers. Multiple modalities such as voice, vision, posture, and emotion will be deeply integrated, returning interaction to human instincts. Zhao Qinp gave examples: when you look up, AI glasses automatically recognize the scene and display information; when your heart rate accelerates, AI bracelets proactively inquire and provide suggestions; when you raise your hand, AI rings can perform lightweight intelligent interaction. "Seamlessness" is the deepest intelligence.
Third, interconnected ecosystems and multiplied value. "The value of a single terminal is limited, but when a seamlessly collaborative ecosystem is formed among AI glasses, AI smartphones, AI PCs, and AI watches, the value will exhibit a multiplier effect," Zhao Qinp further pointed out. "The collaboration between cloud-based general large models and edge-side dedicated small models, the secure circulation of personal data under the privacy computing framework, the seamless flow of cross-device tasks, and dual-brain coupled decision-making—these will be the core features of the new 'Human-Embodied Intelligence' ecosystem."
Fourth, technology popularization and service equalization. Zhao Qinp believes that "Human-Embodied Intelligence" must adhere to technology popularization. AI terminals should move from high-end circles into thousands of households, allowing the elderly, children in remote areas, and people with disabilities to enjoy the convenience brought by intelligence. "This is the original aspiration of technological development and our common responsibility," he stated.
Building a New Ecosystem of "Human-Embodied Intelligence"
In Zhao Qinp's view, to bring "Human-Embodied Intelligence" from concept to millions of users requires the collaborative construction of an open and sustainable new industrial ecosystem by the industry chain, academia, government, and society. To this end, he proposed developing four pillars.
First, basic research. Strengthen research on underlying technologies such as edge-side large models, multimodal perception, spatial computing, cognitive computing, affective computing, privacy computing, and brain-computer interfaces, providing a solid scientific cornerstone for "Human-Embodied Intelligence."
Second, standard systems. Accelerate the formulation of standards for interoperability, data security, and ethical governance of AI terminals to avoid fragmentation and isolation, and achieve an open and collaborative ecosystem.
Third, application innovation. Encourage developers to create killer applications in scenarios such as medical health, education, office production, elderly care and companionship, and entertainment life, forming a virtuous cycle of "promoting terminal popularization through application prosperity."
Fourth, talent cultivation. "Human-Embodied Intelligence" is a highly interdisciplinary field that requires the integration of multiple disciplines such as computer science, cognitive psychology, design, and sociology. It is imperative to cultivate compound talents with humanistic sentiments and technical literacy.
Finally, Zhao Qinp concluded that "Embodied Intelligence" gives machines a body, while "Human-Embodied Intelligence" gives humans a companion. If the past Internet connected computers, the mobile Internet connected people to people, and the IoT connected things to things, then the AI terminal era will connect humans, machines, and things, especially the deeper symbiotic relationship between humans and AI. China has the world's most complete electronics industry chain, the largest AI developer community, and the richest application scenarios, as well as the fine tradition of being human-centric and benefiting humanity. We are fully capable and well-positioned to take the lead globally in building a new "Human-Embodied Intelligence" ecosystem, enabling China's AI terminal industry to lead the world trend.
Author | Wu Lilin
Editor | Zhang Xinyi
Art Design | Ma Liya Supervisor | Lian Xiaodong