Problem

Traditional human-machine interaction models are predominantly human-centric, treating artificial intelligence (AI) as subordinate tools. However, with the rapid evolution of advanced AI, particularly large language models, machines are demonstrating increasing autonomy and intelligence, challenging this conventional paradigm. This shift necessitates a reevaluation of the relationship between humans and AI, moving towards a more symbiotic coexistence where AI entities are recognized as perceptual centers of equal significance. The challenge lies in developing a framework that can accommodate this dual-centered interaction across both physical and virtual environments.

Method

The paper introduces the concept of "symmetrical reality" as a novel descriptive framework for harmonious human-machine coexistence. This framework posits a symmetrical structure where humans primarily inhabit the physical world and AI entities reside in the virtual world. Crucially, both humans and AI are conceptualized as capable of simultaneously perceiving and interacting with both the physical and virtual realms. This establishes a dual-center architecture for human-machine interaction, moving away from the traditional human-centric model. For AI entities to interact with the physical world, embodiments like robotic arms are required. Furthermore, the framework emphasizes the necessity for autonomous machines to possess intrinsic value systems to guide their behaviors, fostering genuinely autonomous and harmonious interactions.

Results

Symmetrical reality provides a foundational solution for long-term harmonious human-machine coexistence, embodying a dual-center architecture that acknowledges AI's significant role due to its autonomy and independence. This paradigm highlights symmetrical perceptual and interactional capabilities between humans and AI in a blended physical-virtual world. The paper differentiates symmetrical reality from related concepts like "digital twins" and "extended reality" by emphasizing AI's autonomous perception and interaction, and its reliance on high-level machine autonomy. It proposes a four-level progression: starting from basic physical human-machine interaction, moving through digital twins (adding virtual world mapping), then combining extended reality and AI (cross-reality interaction), and culminating in symmetrical reality (dual-center architecture with symmetrical cognition).

Implications

The symmetrical reality framework has profound implications for reshaping the relationship between humans and AI, extending the cognition system from a single human-centric model to a dual-centered one. This paradigm is expected to generate numerous innovative ideas within the human-machine interaction and virtual reality communities. It aims to establish an infrastructure supporting harmonious human-machine coexistence by addressing the challenges posed by advanced AI. The concept suggests that highly autonomous machines, equipped with value systems, could behave similarly to humans in certain scenarios, potentially blurring the boundaries between machines and living organisms. As an emerging field, symmetrical reality requires further research and inclusive development to provide forward-looking solutions for future human-machine societies.