Embodied intelligence may enhance public cultural services. Image generated by AI

After more than two decades of systematic development, China’s public cultural services have made remarkable progress in both coverage and service delivery efficiency. Digital platforms, such as smart library systems and public cultural cloud services, have enabled cultural resources to transcend physical and temporal constraints and reach individual users directly. However, while screen-mediated and increasingly individualized forms of cultural engagement have enhanced access to cultural resources, they have also drawn users away from shared cultural spaces, gradually weakening the relational dimension of public cultural services. Today, as artificial intelligence (AI) advances rapidly and access to cultural resources becomes increasingly convenient, it is increasingly important to reconsider what the “public” in public cultural services actually means and to explore how the shift toward embodied AI might offer new opportunities for reconstructing publicness.

Theoretical connotations and practical erosion of publicness

The Public Cultural Service Guarantee Law of the People’s Republic of China establishes four fundamental requirements: public welfare, basic provision, equality, and convenience. Public welfare defines the nature of the service; basic provision sets the scope of what is supplied; convenience concerns the mode of delivery; and equality provides a criterion for equitable distribution of outcomes. Core indicators such as coverage, per capita expenditure, and service radius all center on equality, making it, in practice, the operational core of publicness. This institutional logic can be summarized as distributive publicness: Through public finance, public institutions, and institutionalized provision, cultural resources are made open and accessible to all citizens on increasingly equal terms.

The explanatory power of distributive publicness, however, has clear limits. Measures of equalization take comparable individual accessibility as their starting point and assume that the value of services can be broken down and attributed to individual recipients. Yet during a librarian-led reading session, an exhibition tour, or a local opera rehearsal, participants do not merely receive knowledge as separate individuals. They collectively direct their attention toward the same text, exhibit, or performance, perceive the reactions of others, and adjust their own understanding through questions, silence, laughter, singing along, or collaboration. Such an experience belongs to no single individual, nor can it be reconstructed simply by aggregating individual satisfaction scores. This relational quality—the aspect of publicness that the equalization framework cannot capture—may be termed “constitutive publicness.”

The distinction between distributive and constitutive publicness lies in the different questions each framework asks. Distributive publicness takes a pre-existing public as its recipient and asks whether cultural resources can reach that public equitably. Constitutive publicness, by contrast, asks whether a public is itself formed through cultural practice. The notion of “atmosphere” gets to the heart of constitutive publicness: It exists between people, emerges through their awareness of one another, and cannot be measured simply by aggregating individual reading or usage data.

This suggests that public cultural services should not merely deliver cultural resources; they should also create spaces for co-presence. Yet the conditions for such co-presence are undergoing structural erosion. The root of this erosion lies in the historical decoupling of “access” from “co-presence.” Before the widespread application of digital technologies, physical presence naturally brought resource access and shared presence together, allowing constitutive publicness to emerge in the process. Digital and intelligent transformation now allows resources to be accessed through individual devices, making physical presence optional. Although new types of public cultural spaces have restored the physical settings for shared cultural activity, they have yet to incorporate relational mechanisms through which those present can become aware of one another. Constitutive publicness has therefore shifted from being a by-product of service delivery to an independent objective that must be deliberately designed.

Embodied intelligence as a technological mediator of co-presence

Co-presence requires ongoing coordination and therefore calls for mediators that can be embedded within service processes. Such mediators should operate within physical spaces and sustain the conditions for co-presence through low-intensity forms of intervention that participants can easily disengage from.

Embodied cognition theory holds that the body is not merely a vehicle for transmitting information, but a site where meaning emerges. By coupling perception and action with the physical environment and making its outputs visible and perceptible to those present, embodied intelligence can move beyond screen-bound symbolic computation and into real-time interaction among people. Its capacity to sustain the conditions of co-presence can be understood as progressing through four levels. At the perceptual level, it moves beyond recognizing individual behavior to representing group states. At the mediating level, it guides shared attention through physical action. At the collaborative level, multimodal interfaces foster intercorporeality. At the orchestration level, it maintains spatial conditions that allow people to linger, approach, and observe together.

As embodied intelligence is increasingly applied in museums and other cultural institutions, this technological logic is moving from theoretical possibility into institutional practice. Importantly, the value of embodied intelligence lies in its capacity to accommodate differences dynamically, which also reflects an essential characteristic of publicness: Multiple actors become perceptible to one another while retaining unpredictable differences. The application of embodied intelligence must therefore be strictly bounded. Once representations of co-presence themselves become optimization targets, preprogrammed systems risk eliminating precisely this unpredictability, thereby simulating publicness rather than realizing it.

From technological possibility to institutional capacity

Whether embodied intelligence can contribute to the reconstruction of publicness ultimately depends on whether it can be integrated into the institutional objectives of public cultural services rather than remaining merely a technological offering. This transformation can be understood as a positive cycle between distribution and constitution: Accessible cultural resources provide shared focal points that embodied mediation turns into shared experiences. These accumulated experiences are then converted into new cultural resources that feed back into service provision. Institutionalization requires coordinated efforts across four dimensions: spatial deployment, interaction design, content production, and evaluation mechanisms.

Spatial deployment should prioritize county-level cultural centers, township cultural stations, and community libraries, while intelligent agents should be placed in open areas rather than isolated compartments. Interaction design should support simultaneous participation by multiple users, with responses addressed to everyone present. When local knowledge is involved, intelligent agents should actively solicit participants’ experiences and act as initiators of discussion rather than authoritative providers of answers. Incorporating multimodal channels—including sign language, dialects, and haptic interaction—into conventional interfaces can further broaden participation in co-presence.

For content production, if intelligent agents merely provide standardized knowledge, the expansion of distributive publicness may come at the expense of constitutive publicness. Intelligent agents should therefore be embedded over the long term in settings such as reading groups, local opera rehearsals, and intangible cultural heritage workshops, where they can document oral narratives, dialects, craft procedures, and other materials generated through the process itself. When community memories are transformed into searchable and displayable cultural resources, members of the public shift from being recipients of services to becoming co-interpreters of meaning. In this way, the material foundations provided through distribution can generate new meaning at the constitutive level.

Evaluation mechanisms should likewise incorporate qualitative dimensions alongside existing quantitative indicators. The distribution of community-level service sites, the frequency of co-presence events, and the return of locally generated linguistic and cultural materials to the resource system should all be incorporated into monitoring frameworks.

Evaluation, however, must build in an auditing logic: Indicators of co-presence should serve to make such phenomena visible rather than to drive optimization. Once these indicators become performance targets, institutions risk shifting from maintaining the conditions for publicness to manufacturing its appearance. Institutional safeguards are likewise needed to address the risks of turning technology into spectacle, compromising privacy, and exacerbating spatial inequalities through high costs.

Technologies for group perception and low-intensity intervention remain immature, and constitutive publicness still lacks reliable methods of measurement. These remain practical constraints on the use of embodied intelligence to reconstruct publicness. Nevertheless, when embodied intelligence ceases to function merely as a spatial extension of individualized services and instead becomes an institutionalized mechanism for sustaining the conditions of co-presence, public cultural services may be able to fulfill a more fundamental promise: enabling people not only to access cultural content as individuals, but also to become visible and perceptible to one another through shared cultural practices.

Wu Dan is vice president of and a professor at Central China Normal University (CCNU); He Xiaoyang is an associate research fellow from the School of Information Management at CCNU.