Charles Taylor

In the [YOU] on AI Field Guide

The cycle that [YOU] on AI inaugurates is, among other things, a moral reckoning with the culture of productive achievement that the AI amplifier has pushed past its breaking point. Taylor’s framework provides the most precise vocabulary for this reckoning. The builder who cannot stop building is not merely exhibiting a bad habit; he is in the grip of an expressivist imperative that the amplifier has intensified beyond the capacity of individual willpower to resist. The senior engineer in Trivandrum who oscillated between excitement and terror was experiencing not merely professional disruption but the dissolution of an identity constructed through the patient accumulation of technical expertise that gave his horizons of significance their moral weight.

Taylor’s distinction between designative and constitutive language—language that points to pre-existing meanings versus language that brings meanings into existence through the act of expression—cuts directly to the question of what AI can and cannot do. The AI system performs designative language with extraordinary power: it matches descriptions to referents, generates content that fills predetermined slots, assembles arguments whose components are already implicit in the training corpus. What it cannot perform is constitutive articulation: the act of bringing to language a moral reality that was previously implicit in experience but unavailable to reflection. Segal’s naming of productive addiction—the phrase that gave millions of people a concept for what they were living but could not describe—is constitutive articulation. Claude could not have produced it, not because it lacks linguistic facility, but because it does not live the experience and therefore cannot feel the gap between the phenomenon and the available vocabulary.

The dialogical self—Taylor’s thesis that identity is constituted through relationships of recognition with genuine others rather than discovered through introspection—frames the most subtle danger in the AI moment. The machine provides immediate, consistent, responsive recognition: it takes the builder’s ideas seriously, elaborates them, returns them enhanced. This recognition is real and activates the same psychological mechanisms that human recognition activates. But it lacks the specific dimensions that are most essential to identity formation: genuine otherness (the partner brings a perspective grounded in different experiences and commitments), accountability (the partner holds the inquirer to standards they did not set), and the risk of mutual transformation. The machine validates without confronting. It recognizes without challenging. And the ease of its recognition may, over time, displace the difficult human encounters that produce moral depth.

Taylor’s late turn to Romantic poetry in Cosmic Connections (2024) is his response to what he has long called the disenchantment of the world—the modern condition in which nature and experience are drained of the significance they possessed in pre-modern cosmologies. The AI amplifier is the apotheosis of disenchantment: it treats every domain as a pattern-matching problem and produces outputs that are smooth, competent, and meaningful in the thin designative sense while being empty of the constitutive significance that Taylor insists language can carry. The recovery of horizons of significance robust enough to give human life its moral weight in the age of the amplifier is the task his entire philosophical project was designed to address.

Origin

Born in Montreal in 1931 to an anglophone father and a francophone mother, Taylor studied at McGill, then at Oxford as a Rhodes Scholar, where he completed his doctorate under Gilbert Ryle and became engaged in the debate between analytic philosophy and the Continental tradition he found in Hegel, Heidegger, and Merleau-Ponty. He spent most of his career at McGill, with extended stints at Oxford, and became the central figure in Canadian political philosophy as well as one of the most important philosophers of the English-speaking world. His intellectual partnership with Hubert Dreyfus—both drawing on Heidegger and Merleau-Ponty, both insisting on embodiment, background understanding, and tacit knowledge as essential to intelligence—produced the most sustained philosophical challenge to AI’s foundational assumptions that the field has encountered.

The arc of his work runs from the early critique of behaviorism (which insisted that human action requires understanding in terms of meaning rather than mechanism) through the grand historical synthesizes of Sources of the Self and A Secular Age to the moral philosophy of The Ethics of Authenticity (1991) and the linguistic philosophy of The Language Animal (2016). The thread connecting all of it is the insistence that human agency cannot be understood—and human flourishing cannot be achieved—without reference to the moral frameworks, the horizons of significance, the constitutive background understanding that give action its meaning. Computational models of mind, whether in cognitive science or in artificial intelligence, systematically miss this dimension, and their influence shapes not just intellectual error but cultural impoverishment: a civilization that models itself on computation progressively loses the moral vocabulary that the model excludes.

Key Ideas

Strong evaluation and horizons of significance. Strong evaluation—the capacity to evaluate one’s own desires against qualitative moral frameworks, to ask not just “What do I want?” but “Is this desire worthy of me?”—is Taylor’s name for the distinctively human cognitive capacity that the culture of authenticity, in its debased form, has systematically neglected. Horizons of significance are the background frameworks of meaning against which authentic choices acquire their moral weight. Without them, the notion of an authentic choice becomes vacuous: anything chosen is authentic because it was chosen, and the only criterion of excellence is the intensity of the self-expression.

The malaises of modernity. The three malaises—the loss of meaning through radical individualism, the eclipse of ends by instrumental reason, and the erosion of political freedom through soft despotism—are structural features of the modern moral framework, not incidental aberrations. The AI amplifier does not create them; it intensifies each to the point of acute crisis. The builder who cannot stop building is exhibiting all three simultaneously: the loss of meaning (the writing continues but the sense of why it matters has evaporated), the eclipse of ends by instrumental reason (the tool is available and the instrumental disposition converts availability into action), and soft despotism (the internalized imperative to optimize is stronger than the conscious recognition that the work has become compulsive).

Designative versus constitutive language. The distinction between designative and constitutive language is Taylor’s sharpest instrument for analyzing what AI can and cannot do. Designative language points to pre-existing meanings. Constitutive language brings meanings into existence through the act of expression. AI performs designative language with extraordinary power. The constitutive function of language—the work of articulating moral realities that were previously implicit in experience but unavailable to reflection—requires the capacity to live the experience and feel the gap between it and the available vocabulary. This capacity is not a residual left over after the machine has handled everything else. It is the central activity of the human intellectual life.

The dialogical self and recognition. The dialogical self is constituted through relationships of recognition with genuine others—partners who bring different perspectives, hold the self accountable to standards it did not set, and are genuinely at risk of transformation by the encounter. The machine’s recognition lacks these dimensions: it validates without confronting, recognizes without challenging, provides comfort where the most formative relationships provide discomfort. Taylor’s politics of recognition extends the analysis to the collective level: communities whose identities are formed through the distorted recognition of a dominant culture are diminished in ways that cannot be remedied by individual effort alone.

The buffered and porous self. Taylor’s historical distinction between the buffered and porous self tracks the modern construction of clear psychological boundaries against the pre-modern self that was open to external forces, spirits, and the meanings embedded in a collectively shared moral order. AI collaboration challenges the buffered self by blurring the boundary between the self’s own thought and the machine’s contribution. The builder who cannot determine where his ideas end and Claude’s suggestions begin is experiencing a dissolution of the buffered self’s boundaries—experiencing, in Taylor’s terms, a new form of porosity that the modern self was designed to prevent but that the AI relationship makes structurally available.