Bing Song

In the [YOU] on AI Field Guide

The [YOU] on AI cycle asks what it means to remain a self in the presence of a machine that can simulate every kind of expression. Song's work presses a prior question: what is the self that is being amplified, and is the conception of selfhood underlying the question adequate to the moment? If the self is a liberal individual—atomistic, rights-bearing, prior to its relationships—then the cycle's question is well-formed. But if the self is, as the Confucian tradition insists, fundamentally relational—constituted through connection, defined through roles and obligations, never prior to the social world in which it inheres—then the amplification changes character. What is being amplified is not a bounded individual but a node in a network of relationships, and the ethical question becomes not merely “are you worth amplifying” but “what is the amplification doing to the relational fabric through which the you is constituted?”

Song's framework also reframes the cycle's implicit anthropology. The AI safety discourse in the West is pervaded by anthropocentric anxiety: the fear that AI will surpass human intelligence and thereby usurp the sovereign position humans occupy in the universe. The Daoist and Confucian traditions do not assign sovereign status to the human in the first place; the human is one element in a cosmic whole that includes heaven, earth, and all phenomena. In a cosmological order without human sovereignty as its premise, the emergence of powerful AI is not a metaphysical crisis but a governance challenge requiring the cultivation of wisdom adequate to the new conditions. The different emotional valence—less existential panic, more focused attention on cultivation and institutional design—is not a failure to appreciate the risks but the product of a different anthropological premise, and it may produce more tractable governance responses.

The cycle's final chapter returns the reader to the human capacity that the machines cannot have on our behalf. Song names that capacity precisely: not any particular task that AI cannot perform, but the experience of grasping truth as truth, standing before the order of the world in the specific mode of a being for whom the order is real and significant. This is the Confucian dimension of wisdom—not computational efficiency in achieving specified ends, but the cultivated capacity to discern what ends are worth pursuing, to act in a way that sustains the relational fabric through which human flourishing is possible, to bring ren—humaneness, compassionate attentiveness to others—to the encounter with whatever the moment presents.

Origin

Bing Song's intellectual formation combines Chinese classical philosophy, Western liberal theory, and the practical experience of institution-building at the intersection of Eastern and Western thought worlds. She was educated in China and the United States, and her career has moved between academic philosophy, policy engagement, and the organizational work of creating the conditions for sustained cross-civilizational conversation. Her founding of the Berggruen Institute's China Center in 2017 was a direct response to the observation that the emerging global AI conversation was proceeding as if the Chinese philosophical tradition either did not exist or was not relevant—an observation she regards as both empirically false and practically consequential.

Her 2021 edited volume Intelligence and Wisdom: Artificial Intelligence Meets Chinese Philosophers was the result of years of organized dialogue between major Chinese philosophers and AI researchers, conducted at a moment when the philosophical dimensions of AI were just beginning to receive systematic attention. The nine philosophers who contributed to the volume did not produce a unified Chinese position; they produced nine distinct engagements, each drawing on different aspects of the classical tradition—Confucian ethics, Daoist cosmology, Buddhist philosophy of mind—and each reaching different conclusions about the AI challenge. The plurality of the result was itself a methodological commitment: Song's goal was not to assert a Chinese alternative to Western AI ethics but to open a space for genuine philosophical pluralism in the global AI conversation.

Her 2024 book Gongsheng Across Contexts: A Philosophy of Co-Becoming, co-authored with colleagues at the Berggruen Institute, extends the gongsheng framework across ecological crisis, pandemic response, international relations, and technological governance, arguing that the concept of co-constituting entities in internal relation—rather than discrete entities in external relation—provides resources for analyzing contemporary challenges that the dominant ontological frameworks lack. Her UNESCO report on AI and the future of education applies the framework specifically to educational governance, arguing for standards that attend to educational purpose—the cultivation of wisdom—alongside educational efficiency.

Key Ideas

Gongsheng: co-becoming as the fundamental framework. The word gongsheng (共生) translates roughly as symbiosis but carries philosophical weight the biological term does not: entities that live and become together, constituted through their ongoing relationships rather than prior to them. In Song's application, the human-AI relationship is not an instrumental arrangement between pre-existing parties but an ontological one that changes both parties. Governing AI requires asking not merely “does this system harm individual rights?” but “what kind of human beings does sustained engagement with this system tend to produce?” This is a question the liberal framework is methodologically unable to answer, because it requires substantive judgments about the good that liberal neutrality forecloses.

The relational self and its AI implications. Confucianism begins not with the individual but with the relationship: persons are constituted through networks of mutual obligation and cannot be understood apart from them. Applied to AI, this generates different evaluative questions than the liberal framework. Not merely: does this AI system respect individual autonomy? But: does this AI system strengthen or weaken the relational capacities—empathy, reciprocity, shared meaning, sustained attention to genuine others—on which human flourishing depends? An AI companion that optimizes for user satisfaction without offering the formative friction of genuine other-ness may atrophy the very relational capacities that healthy human life requires, producing user satisfaction and relational impoverishment simultaneously.

The intelligence-wisdom distinction. The intelligence-wisdom distinction that structures Song's most important editorial project holds that intelligence—the capacity to process information, identify patterns, and optimize for specified goals—is a tool, while wisdom is the capacity to use the tool well: to act rightly in situations of genuine moral complexity, drawing on cultivated virtue and relational sensitivity. The deepest AI risk, in this framework, is not a misaligned machine but a civilization of powerful intelligence that has outsourced the cognitive tasks through which wisdom is cultivated, discovering too late that the machines can optimize for every specified end and cannot tell us which ends are worth specifying.

Wu-wei and a different model of AI alignment. The Daoist concept of wu-wei—effortless action arising from deep attunement to the nature of things—suggests a different model for AI alignment than the value-loading paradigm dominant in Western AI safety. Rather than specifying values in advance and training systems to implement them, wu-wei alignment would cultivate in AI systems the capacity to read actual human needs and relational contexts with sufficient depth that their responses flow from that attunement rather than from an imposed specification. Values are not context-independent specifications that can be loaded; they are dynamic, relational orientations that emerge from sustained engagement with the world. Cook Ding's mastery is the model: the action that flows perfectly with the grain of the material, requiring no force because the understanding is total.

Governance as civilizational, not merely regulatory. Song insists that the choices being made about AI development are choices about the trajectory of human civilization—what capacities will be cultivated and which will atrophy, what kinds of relationships will be normative, what conception of the good life will be encoded into the systems through which billions of people navigate their days. This is too large a question for regulatory frameworks designed around individual rights and specific harms. It requires the kind of institution capable of civilizational thinking—capable of asking what kind of civilization we want to be and working backwards to the governance choices that would support it. The Confucian tradition, which has historically conceived of governance's role in relation to civilization rather than merely to market regulation, offers resources for this kind of institution-building that the Western regulatory tradition lacks.