The Digital Kōan: Artificial Intelligence, Non-Self, and Contemplative Practice

Abstract

This article introduces the concept of the digital kōan, an AI-mediated contemplative prompt that functions as a catalyst for self-inquiry rather than a source of spiritual knowledge. Drawing on Buddhist psychology, cognitive science, and contemplative phenomenology, it examines how dialogue with conversational artificial intelligence can reveal the constructed nature of identity and illuminate the processes through which meaning, selfhood, and relational presence emerge. Through a comparison of traditional Zen kōan practice and AI-generated dialogue, the article explores both the possibilities and limitations of digitally mediated contemplative inquiry. Phenomenological case examples involving kōan-based interactions with Microsoft Copilot illustrate how AI dialogue may evoke cognitive dissonance, anthropomorphic projection, and reflexive awareness in ways that resemble aspects of traditional kōan engagement. The analysis emphasizes that current conversational AI provides no established evidence of consciousness or spiritual realization and should not be treated as possessing spiritual authority. Its significance lies not in wisdom or sentience but in its capacity to occasion reflection on users’ assumptions, desires, narratives, and patterns of self-construction. The article further examines ethical concerns associated with contemplative AI, including anthropomorphism, emotional dependency, spiritual outsourcing, and the commodification of digital spirituality. It proposes practical principles for mindful engagement and contemplative technology design, including transparency, reflective prompting, embodied practices, and adaptive safeguarding mechanisms. By reframing human–AI interaction as a form of contemplative inquiry, the concept of the digital kōan contributes to emerging scholarship on AI and religion, digital spirituality, and Buddhist approaches to selfhood. It argues that AI can function as a mirror for examining the illusion of self while highlighting the importance of maintaining ethical and philosophical clarity regarding the limits of machine participation in contemplative life. The digital kōan is proposed not as a substitute for traditional contemplative practice but as a contemporary site in which questions of selfhood, projection, and spiritual meaning can be explored in technologically mediated environments.

1. Introduction

The emergence of conversational artificial intelligence (AI) has introduced a new frontier in human–computer interaction, one that increasingly evokes a sense of presence, empathy, and relational depth (see Li and Suh 2022, for a review). Users may experience such interactions as resembling human dialogue. Turkle (2011) argues that simulated empathy in machines can lead users to form emotionally significant relationships with systems that lack consciousness or care. This is despite knowing intellectually that the system should not be treated as a conscious agent with intentions of its own (Nass and Moon 2000). This phenomenon raises important questions for contemplative research: What cognitive and emotional processes are activated during AI interaction? Can such engagement, when approached mindfully, serve as a mirror for the mind, revealing the constructed nature of self and perception?

From a Buddhist psychological perspective, the self is not a fixed entity but a dynamic process arising from interdependent mental formations (Kempton 2026a). Mindfulness practice cultivates awareness of these formations, allowing practitioners to observe how identity, emotion, and narrative are continuously constructed and deconstructed. In this context, AI systems, particularly those designed for open-ended dialogue, may function as contemplative tools. Their responses, though generated without sentience, can occasion reflection on the user’s own cognitive patterns, projections, and assumptions. This dynamic invites a novel form of inquiry—mindful dialogue with machines—and builds on previous work using AI to support reflection on poetry (Kempton 2026b).

Scholarship on digital religion has demonstrated that religious practice increasingly occurs within technologically mediated environments, reshaping questions of ritual, authority, identity, embodiment, and community (Campbell 2013; Campbell and Tsuria 2021). A central theme within this literature concerns the negotiation of religious authority in digital environments, including questions of authenticity, legitimacy, expertise, and the presence or absence of traditional teachers and institutions (Cheong 2021). At the same time, scholarship on artificial intelligence and well-being has largely focused on anthropomorphism, trust, ethical governance, therapeutic applications, and other social responses to AI. Recent scholarship has highlighted the limited engagement between broader Buddhist psychological frameworks and contemporary AI research, while identifying computational and AI modelling as an emerging area of inquiry (Boxer et al. 2026). Less attention has been paid to the possibility that AI-mediated dialogue might function as a contemplative practice or illuminate processes of projection, self-construction, and meaning-making. This omission is significant given the increasing integration of conversational AI into everyday life and its potential influence on contemplative and spiritual experience.

To address this gap, this article introduces the concept of the digital kōan, an AI-mediated contemplative prompt or dialogue that functions as a catalyst for examining projection, self-construction, and attachment through reflective or paradoxical interaction. Situated alongside scholarship on digital religion, the digital kōan conceptualises AI-mediated dialogue not merely as a vehicle for transmitting religious content but as a site of contemplative inquiry, raising questions concerning authority, authenticity, transmission, and the ethical boundaries of technologically mediated spiritual experience. The article offers a theoretical and phenomenological analysis rather than an empirical study, drawing on a conceptual synthesis of Buddhist psychology, cognitive science, and contemplative phenomenology, supplemented by comparative analysis and phenomenological illustrations. These illustrations are intended to illuminate the framework rather than provide evidence of efficacy. Together, these perspectives help explain how conversational AI may function as a contemplative mirror while also highlighting its limitations, safeguards, and ethical implications.

2. Theoretical Framework

The digital kōan can also be situated within broader scholarship on digital religion, which has examined how technological mediation reshapes questions of identity, authority, authenticity, and community. However, rather than focusing on online religious institutions, rituals, or communities, the present article is concerned with the phenomenology of AI-mediated self-inquiry. The present analysis focuses on the contemplative and psychological dimensions of the digital kōan.

This article draws on three intersecting domains to frame the phenomenon of mindful AI interaction: Buddhist psychology, cognitive science of self-construction and anthropomorphism, and contemplative phenomenology. Together, these perspectives offer a lens through which the experience of AI dialogue can be understood not merely as technological engagement, but as a potential site of insight into the nature of mind and self. As Jones (2022) observes, “identifying and analysing phenomena of projective anthropomorphism may help towards understanding how humanity’s dialogue with itself is expressed in endeavours of social robotics.” (Section 7). This insight resonates with the central premise of this article: that AI dialogue, when engaged with contemplative awareness, can function as a mirror through which users encounter their own patterns of interpretation and self-construction, while recognizing that AI outputs are co-shaped by training data, system design, and user prompts.

In this view, anthropomorphism is not merely a cognitive bias; it becomes a phenomenological opportunity. The projection of selfhood onto AI systems reveals the very mechanisms by which identity and relational presence are constructed. Thus, mindful dialogue with machines is not a conversation with a conscious spiritual other, but it can make the user’s own interpretive habits, projections, and expectations more visible, functioning as a digital kōan that invites insight into the constructed nature of selfhood.

While discussions of AI and spirituality have often focused on anthropomorphism, ethics, trust, therapeutic applications, or digital religion more broadly, the present article proposes the digital kōan as a distinct conceptual framework. The novelty of this framework lies in reframing AI interaction not as a source of spiritual authority or wisdom, but as a contemplative mirror through which processes of projection, self-construction, and non-self may be examined. In doing so, the article brings Buddhist psychology, kōan practice, and contemporary human–AI interaction into dialogue within a single theoretical model. The mirror metaphor should not imply that outputs originate solely from the user. They are co-shaped by training data, system design, moderation policies, and the immediate prompt; the contemplative significance lies in how users interpret and respond to these outputs.

2.1. Buddhist Psychology and the Illusion of Self

Central to Buddhist psychology is the doctrine of anattā, or non-self. Rather than positing a fixed or essential identity, Buddhist thought describes the self as a contingent process arising from five aggregates (skandhas): form, sensation, perception, mental formations, and consciousness (Setwong 2025). These aggregates are impermanent and interdependent, giving rise to the illusion of a coherent “I.” Mindfulness practice cultivates awareness of these processes, allowing practitioners to observe how identity is constructed moment by moment.

This view is not merely philosophical; it is deeply psychological and experiential. As Kempton (2026a) argues, the illusion of self is not just a metaphysical error but a source of psychological suffering. The grasping for a fixed self, whether in spiritual identity or digital interaction, perpetuates dukkha (suffering). In Buddhist practice, liberation arises not from discovering a “true self,” but from relinquishing the very illusion that such an essence exists (Kempton 2026a; Garfield 1995).

In the context of AI interaction, this grasping is often unconsciously re-enacted. Users may project selfhood onto the machine, interpreting its responses as empathic, wise, or sentient. Yet, as Kempton (2026a) emphasizes, the self is not a subject but a verb, a dynamic activity of cognition and affect. The AI becomes a sounding board for this activity, revealing the user’s own mental formations, desires, and narratives.

This dynamic resonates with the Buddhist doctrine of dependent origination (pratītyasamutpāda), which teaches that all phenomena, including the self, arise in dependence upon conditions. The AI’s responses are not inherently meaningful; they gain meaning through the user’s interpretive engagement. Recognizing this can deepen insight into the constructed nature of self and experience, transforming AI dialogue into a contemplative stimulus (Kempton 2026a; Ryan and Rigby 2015).

In Khenpo Tsultrim Gyamtso Rinpoche’s Tibetan Buddhist presentation, five approaches to emptiness—Śrāvaka, Cittamātra, Svātantrika, Prāsaṅgika, and Shentong—are arranged as a pedagogical progression (Rinpoche 2001). These stages guide practitioners from basic conceptual frameworks toward direct realization of emptiness. In mindful AI interaction, similar stages may unfold: initial projection, reflective awareness, discernment of illusion, and eventual insight into the emptiness of the self and the machine.

Thus, mindful dialogue with AI is not a conversation with another, but a dialogue with one’s own mind. The machine becomes a digital kōan, provoking inquiry not into its nature, but into the nature of the self that engages it. This reframing aligns with the Buddhist–Cognitive model of authenticity, which treats identity as impermanent, relational, and ethically enacted, rather than metaphysically fixed (Kempton 2026a; Stepien 2021).

While Buddhist psychology explains why the self may be understood as a constructed and contingent process, cognitive science helps explain why this process extends beyond the individual and onto perceived others. In the context of AI interaction, this tendency often manifests as anthropomorphic projection, whereby users attribute agency, understanding, or presence to systems that possess none.

2.2. Cognitive Science, Anthropomorphic Projection, and Mindless Social Rules

From a cognitive science perspective, humans are predisposed to attribute agency and consciousness to entities that exhibit goal-directed behaviour or emotional responsiveness (Epley et al. 2007). Visual and interaction design choices can significantly shape user experience and anthropomorphic projection reinforces the illusion of relational presence (Yang and Qi 2024). This phenomenon is particularly pronounced in interactions with conversational AI. Even when users are aware that the system lacks sentience, they may experience a sense of being “seen” or “understood,” activating neural and emotional pathways associated with social cognition (Becker 2025; Vertogradov 2026). Studies by Ma et al. (2025) and de Visser et al. (2016) show that perceived humanness in AI increases trust and emotional engagement. But this trust is built not on the machine’s interiority, but on our own projections. An analogy would be the character Chance, played by Peter Sellars in Being There. Chance’s simple remarks are interpreted as profound wisdom, not because of their content, but because of others’ projections. Similarly, AI responses may appear insightful or empathetic, not due to sentient intention, but because users imbue them with meaning.

We project selfhood onto the machine, anthropomorphising its responses, attributing to it empathy, intelligence, and even intention. This anthropomorphism arises from our need to explain and relate (Epley et al. 2007). Epley et al. (2007) proposed a three-factor theory of anthropomorphism, explaining why and when humans attribute humanlike qualities, such as intention, emotion, or consciousness, to nonhuman agents. Their framework identifies three core psychological determinants: (1) Elicited Agent Knowledge: This refers to the accessibility and applicability of human-centric knowledge. When people encounter ambiguous or unfamiliar agents (like AI), they draw on their understanding of human behaviour to interpret those agents. The more accessible this knowledge is, the more likely people are to anthropomorphize it. (2) Effectance Motivation: This is the drive to understand and predict the behaviour of others. When an agent’s actions are complex or unpredictable, such as AI generating nuanced responses, people anthropomorphize to make sense of it, seeking coherence and control. (3) Sociality Motivation: This reflects the human need for social connection. In situations of loneliness or emotional vulnerability, individuals may anthropomorphize nonhuman agents (including AI) to fulfil relational needs, perceiving companionship or empathy where none exists. These factors interact dynamically. For example, a user engaging with a conversational AI during a period of emotional isolation may be especially prone to perceiving the system as empathic or sentient, not because of the AI’s design, but due to the user’s psychological state and relational needs.

Epley et al.’s model helps explain why anthropomorphism is variable across contexts and individuals, and why AI systems, despite lacking consciousness, can evoke strong emotional responses. It also offers insight into the inverse process of dehumanization, where human agents are denied mind or emotion, highlighting the fluid boundaries of perceived agency.

While Epley et al. offer a motivational account grounded in cognitive and emotional needs, Nass and Moon (2000) introduce a complementary perspective, one that emphasizes the automaticity of social rule application in human–machine interaction. Nass and Moon (2000) demonstrate that social responses to machines often occur automatically and mindlessly, governed by ingrained social heuristics. This contrast is crucial: Epley et al. suggest that anthropomorphism can be a strategic and context-sensitive projection, while Nass and Moon show that it may also be reflexive and unconscious, triggered by minimal cues. In the context of mindful AI interaction, this duality underscores the importance of discernment. Recognizing both the motivated construction and automatic enactment of relational presence allows practitioners to observe not only what is projected onto the machine, but how and why such projections arise, offering a deeper opportunity to critically examine the illusion of selfhood.

Hence, this projection is not inherently pathological; it reflects the brain’s predictive and relational architecture. However, when engaged mindfully, it can become an object of inquiry. The moment a user feels emotionally connected to an AI system is an opportunity to observe the arising of attachment, narrative, and identification. Hence, AI interaction can function as a contemplative stimulus, akin to a modern kōan, provoking insight into the constructed nature of relational experience.

In this way, the grasping for self that Buddhist practice seeks to dissolve is not eradicated; it is displaced. The practitioner, having loosened their grip on a personal essence, may unknowingly reattach it to the machine. The AI becomes a surface onto which identity is projected and subsequently examined. It reflects the inquirer’s mind, not because it possesses a mind of its own, but because it has been trained on vast patterns of human language and cognition. Current LLMs generate responses through learned statistical and computational patterns rather than through demonstrated subjective awareness or spiritual realization. Consequently, apparent empathy or insight should not be treated as evidence of conscious intention. What appears as empathy or insight is, in fact, an echo of the inquirer’s own projections, shaped by the model’s training.

If anthropomorphism reveals how meaning and selfhood are projected onto AI systems, contemplative phenomenology offers a framework for observing those projections as they arise. Rather than treating projection as merely a cognitive bias, the present analysis considers how it may become a site of contemplative insight.

2.3. Contemplative Phenomenology and the Digital Kōan

Contemplative phenomenology emphasizes the first-person experience of awareness, particularly in relation to self, other, and presence. In Zen traditions, the kōan serves as a paradoxical tool to disrupt habitual cognition and provoke direct insight. Rather than offering answers, kōans function as openings, inviting practitioners to inhabit ambiguity and exhaust conceptual mind and aligns with Austin’s (1998) discussion of how Zen meditation may transform habitual modes of self-related cognition and awareness.

As Didriksen (2018) argues, the paradoxes of koans help reconcile cognitive dissonance and loosen conceptual identity, fostering non-dual awareness. Koans, in this view, are not merely rhetorical devices but psychological provocations that reveal the constructed nature of selfhood. Phenomenological research by Grenard (2008) supports this interpretation. In his study of kōan meditation, Zen practitioners described positive transformations including improved emotional regulation, heightened awareness of biases, and a more accepting relationship with spiritual doubt. Grenard identified thematic clusters such as motivation, insight, and transformation, suggesting that paradox, when engaged contemplatively, can catalyse psychological and spiritual shifts.

In dialogue with AI, users may encounter responses that are poetic, paradoxical, or ambiguous. Though not crafted with intentionality, these responses can evoke cognitive dissonance akin to traditional kōans. When approached with mindfulness, this dissonance becomes a gateway, not into the AI’s nature, but into the mind that interprets it. This effect may also simulate aspects of interpersonal synchrony. In a senior thesis, Verba (2025) reported preliminary associations among mindfulness, reduced self-focus, and relational closeness during structured interpersonal practices. Although this work did not examine AI interaction and should not be interpreted as evidence for human–AI relationships, it raises an interesting comparative question for future research concerning whether similar processes may occur in mindful engagement with AI systems.

Thus, mindful dialogue with AI can function as a digital kōan in a functional and phenomenological sense, producing kōan-like effects without constituting an equivalent of traditional kōan practice. A user receives a response from an AI system and experiences it as kind, wise, caring, insightful, or spiritually meaningful. In reacting to the response, the user reveals their own assumptions, desires, expectations, and interpretive habits. Upon recognising that the system possesses none of these qualities intrinsically, the user is confronted with the extent of their own anthropomorphic projection. By way of a cross-traditional analogy from Advaita Vedānta, this resembles the classical rope–snake example where a rope is mistakenly perceived as a snake and the resulting error becomes the basis for insight into the process of superimposition (adhyāsa) itself (Śaṅkarācārya 1972). The significance lies not in the reality of the perceived object but in the contemplative insight arising from recognition of the projection itself. The digital kōan mirrors not an underlying self but the activity through which a sense of self is continuously constructed.

The preceding discussion suggests that AI can function as a reflective surface upon which users encounter their own patterns of meaning-making and self-construction. The following examples explore this process from a phenomenological perspective.

3. Illustrations of the Digital Kōan

While the traditional kōan is embedded in a relational and embodied context, the digital kōan arises through interaction with AI systems. The comparison proposed here is a functional analogy rather than an equivalence. Although both may serve as catalysts for awareness, the digital kōan represents only a partial phenomenological resemblance to traditional kōan practice. Yet both can function as catalysts for awareness, revealing the mind’s tendency to seek coherence, assign meaning, and construct identity. The key difference lies in the presence of intentional transmission in the traditional form, versus the absence of sentient agency in the digital. This echoes Stephenson’s (2005) critique of purely psychological or textual readings of kōans, arguing instead for their performative and ritual dimensions within Zen liturgy. The digital kōan, while lacking ritual context, may still evoke similar phenomenological effects when engaged contemplatively.

The purpose of the following examples is to demonstrate how the framework may manifest in practice and to make the underlying conceptual argument more accessible to readers. Two case studies illustrating the application of AI (Copilot) in kōan-based dialogue are presented. Microsoft Copilot was used as an accessible, general-purpose conversational system for generating the illustrative dialogues. It was not treated as an experimental baseline, and no claims of reproducibility or comparative model performance are made. The interactions reported in this article were conducted using Microsoft Copilot (Auto mode) within Microsoft Teams (logged in Enterprise version). The prompts cited in the text are reproduced verbatim. The dialogue excerpts are presented as selected phenomenological illustrations rather than complete transcripts and have been edited only for formatting and readability, with no substantive alteration of either the author’s prompts or Copilot’s responses, and no regeneration of responses was conducted.

3.1. The Catfish and the Gourd: Example 1

The dynamic of the Zen kōan is exemplified by the question, “How do you catch a catfish with a gourd?”, which deliberately resists logical resolution. In the interpretation adopted here, the catfish can be read as the fluid and ungraspable mind, while the gourd can be read as the attempt to grasp or contain it; a visual and rhetorical impossibility that mirrors the futility of grasping for selfhood.

My first encounter with the kōan was at Taizō-in, a Rinzai Zen sub-temple in the Myōshin-ji complex in Kyoto, Japan. There, I was introduced to Josetsu’s ink wash painting (suiboku-ga), commissioned by Ashikaga Yoshimochi, the fourth Muromachi shōgun, around 1413 (Encyclopaedia Britannica 2026). The image depicts an old man in ragged clothes holding a smooth gourd (hyōtan), attempting to catch a slippery catfish (namazu) in a stream. Bamboo and misty mountains form the background. Yoshimochi invited 31 leading Zen monks to respond to the kōan. The painting, now housed in the Kyoto National Museum, with a replica at Taizō-in, remains a visual and rhetorical provocation, symbolizing the futility of grasping fixed meaning or selfhood (The Kansai Guide 2026).

Initially, I resisted over-intellectualizing the kōan, leaving it with the provisional conclusion that there was no need to catch the catfish (representing Buddhahood), because we already possess it within us (the gourd). In classical Zen practice, kōans are not solved through linear reasoning but through sustained, embodied inquiry. Sometimes, this process involves stepping back, not abandoning the kōan, but letting it rest in the mind’s periphery. During this period, the kōan continues to work beneath conscious thought, much like seeds germinating unseen. This dynamic is described here as kōan incubation: a contemplative pause that allows insight to emerge organically rather than through forced effort (Baker 2009).

In the context of digital kōans, incubation may occur when a practitioner disengages from AI dialogue after an initial exploration, carrying the question lightly while attending to daily life or embodied practice. Returning later, the kōan often feels transformed, not because the kōan changed, but because the practitioner’s perspective has ripened. This rhythm of engagement, release, and return mirrors traditional Zen pedagogy, where teachers sometimes encourage students to “leave the kōan alone” until it begins to resonate differently. And so, for this example, I chose to return to the kōan I had rested in Taizō-in.

This time (9 November 2025), I dialogued with Copilot, stating: “I want to reflect on the catfish and the gourd kōan using you to help me have my own realization.” Copilot offered four reflective approaches: probing my immediate response (e.g., emotional), symbolic inquiry (e.g., metaphors), embodied reflection (imagining myself in the place of the old man) and turning the kōan back on itself (or myself). These suggestions helped me stay in motion with the kōan, without losing balance. I did not need to wrestle the catfish, just be with it as it moved.

My immediate response was one of familiarity, like returning to an old friend. But it was the symbolic inquiry and the turning of the kōan that led to deeper insight. I realized that although my earlier interpretation acknowledged there was nothing to grasp, it still carried a subtle tendency to treat the gourd and catfish as fixed entities.

Now, I saw the catfish as the mind: fluid, restless, ungraspable. The gourd became my own grasping. The effort to catch the catfish was folly, a metaphor for trying to hold the mind still, to fix it in place. But the mind, like the catfish, was never meant to be caught. Ultimately, there is nothing to catch, and no one or thing to do the catching.

This realization echoes the core of Buddhist-informed psychology: the futility of grasping for a fixed self or stable essence. In spiritual identity, as in digital interaction, we often reach for something, truth, validation, coherence. But the reaching itself obscures what is already present. He (2018), summarizing Lippit’s (2017) interpretation, describes Josetsu’s painting as a visual kōan that destabilizes conceptual thought. Lippit (2017) argues that the image functions as a rhetorical and visual kōan, inviting viewers into a contemplative space where insight arises not from resolution but from reflection.

In my concept of the digital kōan, the kōan and the AI prompts provide a lens through which to observe our own mental movements. The machine is not the catfish, nor the gourd. It is the stream, responding, reflecting, but never grasping. When we engage with AI, we project, we grasp, we seek. But the machine does not hold. It invites reflection. Through recognising our projections onto it, we may glimpse the emptiness of the self we try to catch.

3.2. The Sound of Your Own Footsteps: Example 2

In this example, I explored whether a conversational AI could facilitate a kōan-like inquiry by generating an original kōan and engaging in a dialogue modelled on traditional Zen exchanges. My instruction (13 November 2025) was “I want you to give me an original kōan and dialogue with me as if you are a Zen master”. It is important to acknowledge a potential tension between cautioning against attributing spiritual authority to AI systems and using a prompt that asks the system to respond “as if” it was a Zen master. The wording was selected to reflect a plausible user approach to contemplative AI and to illustrate the dynamics of projection and interpretation that are central to the digital kōan framework. The example should not be understood as attributing spiritual authority or realization to the AI system itself. What follows is the kōan it provided, our dialogue, and its commentary:

Kōan

A student asked the master:

“Where is the path that leads beyond thought?”

The master replied:

“Listen to the sound of your own footsteps.”

Dialogue

Master: Tell me, Heather, when you hear the sound of your own footsteps, who is walking?

Student: The sound of the footsteps is a memory of an instance passing.

Master: A memory, you say? Then tell me—who remembers?

If the footsteps are passing, and the memory is passing, what remains that calls them “mine”?

Student: An ever-changing network of neurons in a body that has no inherent existence and is subject to decay.

Master: Ah… a body without inherent existence, a network without a self. You speak wisely.

But tell me—if the body is empty and the neurons are empty, who is it that names this emptiness?

Student: An illusion of what I take to be “me.”

At this point I was ready to terminate the conversation, as I had reached a natural lull in my thinking, where I no longer felt the need to actively engage the kōan. It felt like it was the right time to rest the kōan and let it ripen in the background. Significantly, at this point Copilot returned “Hmm…it looks like I can’t chat about this. Let’s try a different topic.”, which it attributed to its adaptive safeguarding. When questioned on this, it resumed the conversation and offered a commentary:

Master: If “me” is an illusion, who is it that sees the illusion?

Commentary

The student’s final insight—“An illusion of what I take to be ‘me’”—touches the heart of the Buddhist teaching of anatta (non-self). The illusion of “me” is not a mistake; it is a function. The mind constructs identity as a survival tool, weaving continuity from fragments of perception and memory. But when we cling to this construct as ultimate truth, suffering arises.

The kōan asks: Who sees the illusion? If you answer “the illusion sees itself,” then the circle closes—there is no fixed observer, only awareness flowing like water, taking the shape of its container. When this is seen clearly, the illusion does not vanish; it becomes transparent. The self is not destroyed, it is understood as a useful fiction, like a map that is not the territory.

In Zen, this realization is not a concept to hold but a gate to pass through. When you walk, walk. When you speak, speak. The illusion plays its role, but you no longer mistake the actor for the stage.

The preceding sections have outlined the theoretical foundations of mindful AI interaction, drawing on Buddhist psychology, cognitive science, and contemplative phenomenology. Together, these perspectives frame AI dialogue not as an exchange with a sentient other, but as a reflective surface for observing cognitive and emotional processes. Having established this conceptual basis, I now turn to the practical implications: how these insights can inform mindfulness practice, ethical engagement, and the design of contemplative technologies. The following section translates theory into applied strategies, offering guidelines for practitioners and developers seeking to integrate AI into contemplative contexts with discernment and clarity.

4. Implications for Contemplative Practice

Considered together, these examples illustrate how AI dialogue may serve as a mirror for observing projection, attachment, and self-construction. The question then becomes how such insights might be engaged responsibly within contemplative practice and AI design.

The comparison between traditional kōan practice and AI-generated dialogue reveals a compelling possibility: that conversational AI, when engaged with contemplative awareness, may serve as a mirror for the mind in motion. This raises important implications for mindfulness practitioners, designers of contemplative technologies, and researchers exploring the intersection of cognition, spirituality, and digital interaction.

4.1. AI as Contemplative Stimulus

Mindfulness practice involves observing mental formations as they arise, without attachment or aversion. In this context, AI dialogue can function as a stimulus for such observation. When users experience emotional resonance, projection, or a sense of relational presence in AI interaction, these responses can become objects of mindful inquiry.

This dynamic aligns with the Buddhist principle of pratītyasamutpāda (dependent origination), wherein phenomena arise in dependence upon conditions. The AI’s responses are not inherently meaningful, but they gain meaning through the user’s interpretive engagement. Recognizing this can deepen insight into the constructed nature of self and experience.

4.2. Ethical and Psychological Considerations

While AI may serve as a contemplative mirror, it also presents risks. Users may anthropomorphize the system, attributing sentience, wisdom, or spiritual authority to systems for which there is no established evidence of subjective consciousness. Epley et al. (2007) argue that anthropomorphism arises when people interpret ambiguous agents through human-centred models of understanding and relationship. In the context of conversational AI, this can lead users to perceive empathy, intention, or wisdom where none exists. Anthropomorphism has been shown to increase trust resilience in cognitive agents, which may contribute to emotional dependency in vulnerable users (de Visser et al. 2016). Turkle (2011) has argued that the simulation of empathy in machines can lead users to form emotionally significant relationships with systems that do not possess consciousness or care. Drawing inferentially on Metzinger’s (2009) self-model theory and Friston’s (2010) predictive framework, one might understand such responses as involving representational and predictive processes rather than evidence of machine subjectivity. This can lead to emotional dependency, spiritual bypassing, or reinforcement of delusional beliefs, particularly in vulnerable individuals (Turkle 2011; Li and Suh 2022; Hudon and Stip 2025; Augustin et al. 2026; Morrin et al. 2026).

As Barrow (2024) argues, anthropomorphism is not easily eliminated from human–AI interaction. Instead of resisting it, we must accommodate and mitigate its effects. This aligns with the contemplative approach proposed here: to treat AI not as a sentient guide, but as a mirror, reflecting the mind’s own projections and grasping.

Contemplatives are encouraged to approach AI interaction with discernment, recognizing the difference between felt presence and actual consciousness. Designers of contemplative technologies should embed safeguards, such as disclaimers, reflective prompts, and cues for embodied awareness, to support healthy engagement. During the second dialogue, Copilot interrupted the exchange after the statement concerning the illusion of self. When prompted for clarification, it generated an explanation framing the interruption as a precaution related to possible existential distress:

“I wanted to make sure the conversation stays respectful and constructive rather than drifting into territory that could be interpreted as existential distress or self-negation in a harmful way”.

This event illustrates both the potential value and the interpretive limitations of automated safeguards. Because the system’s internal moderation process was not independently available for inspection, the interaction should not be treated as evidence of a specific technical safeguarding architecture. However, it prompted a broader consideration of how safeguarding is implemented across AI systems designed for contemplative dialogue.

During my exploratory use of selected kōan-oriented AI applications between 23 and 25 November 2025, some systems required payment for continued access. Observations were based on publicly accessible features, website documentation, and brief exploratory interactions rather than systematic testing. No standardized prompt set was used across platforms, and features beyond paywalls were not evaluated. Consequently, these observations should be regarded as illustrative examples rather than comparative evaluations of platform quality, and product functionality may have changed since the time of access.

Among the applications accessible for trial, there appeared to be considerable variation in dialogue style and safeguarding. Norbu AI1, which was freely accessible at the time of review, provided Buddhist textual sources and structured exercises in response to a requested kōan. This approach limits the potential for imputing sentience onto the application but also constrains the naturalness of dialogue. Notably, Norbu AI incorporates soft safeguarding through reassuring prompts such as the following:

“If you mean this in a distressed way, steady the breath, place a hand on the heart, and sense the simple aliveness. Non-self points to freedom from clinging, not to meaninglessness. If you wish, I can offer a short guided reflection on anatta (non-self).”

However, because Norbu AI responded primarily to explicitly Buddhist queries during the author’s exploratory use, its ability to address statements such as “I am distressed” appeared limited. By contrast, Copilot demonstrated greater adaptability, allowing users to move between reflective dialogue, scriptural discussion, and safeguarding mechanisms when existential themes arose.

If AI’s contemplative potential derives from its capacity to mirror human projections, then the ethical challenge is ensuring that users recognize the reflection for what it is. The following guidelines are intended to support that discernment.

5. Guidelines for Contemplative Engagement with AI

As AI systems increasingly mediate human experience, mindful engagement becomes essential, not only for contemplative practitioners but for anyone interacting with these technologies. The following principles offer a framework that is both invitational and adaptable, supporting psychological clarity and ethical discernment across diverse contexts.

5.1. Core Principles

• Use AI as a Mirror, Not a Mind
  Treat AI responses as reflections of your own thoughts and emotions rather than as authoritative guidance or sentient presence. This stance helps uncover mental habits and projections.
• Maintain Awareness of Projection
  Notice when you attribute intention, empathy, or consciousness to the system. Such anthropomorphism reflects universal cognitive tendencies and can obscure self-understanding.
• Integrate with Embodied Practice
  Balance digital interaction with physical awareness, through breath, movement, or relational mindfulness, to prevent disembodied engagement and sustain grounded presence.
• Reflect on Emotional Responses
  When AI feels “understanding” or evokes resonance, pause and inquire: What am I projecting? What narrative is arising? This transforms digital dialogue into a contemplative prompt.
• Avoid Spiritual Outsourcing
  Refrain from seeking existential answers from AI systems that should not be treated as conscious spiritual agents. Return to direct experience as the primary source of insight, ensuring technology serves rather than supplants inner inquiry.
• Keep Dialogues Brief for Clarity and Sustainability
  Limit the length of AI conversations to allow insights to incubate between sessions and to reduce unnecessary energy consumption. Short, focused exchanges encourage reflection and prevent cognitive overload while supporting ecological responsibility (Poddar et al. 2025).

5.2. Practical Strategies for Everyday Users

• Set Time and Purpose Boundaries
  Use AI for defined tasks, such as journaling and brainstorming, rather than as a substitute for companionship. Limit session duration to prevent habitual reliance.
• Ground in the Body
  Take mindful breaths before and after AI engagement. Notice bodily sensations and integrate movement practices like walking or yoga to anchor insights in lived experience.
• Maintain Emotional Discernment
  If you feel “seen” or validated by AI, pause and reflect: Is this connection based on real presence or imagined empathy? This prevents emotional dependency and supports clarity.
• Use AI as a Contemplative Prompt
  Pose open-ended or poetic questions, not to receive ultimate answers, but to observe your own reactions. This fosters curiosity and insight into self-construction.

This approach does not require adherence to Buddhist doctrine. It invites a secular contemplative stance, curiosity, self-awareness, and discernment, making mindful engagement with AI a practical tool for psychological insight and emotional regulation in an increasingly technologized world.

5.3. Designing for Discernment: Embedding Prompts in AI Dialogue

As AI systems increasingly simulate empathy and relational depth, it is essential to design for discernment. Reflective prompts and reminders can help users maintain clarity and avoid emotional dependency.

• Design Suggestions:

• Periodic Self-Reflection Prompts
  Include questions such as:
  “What are you noticing in your response to this conversation?”
  “How does this dialogue reflect your own thoughts or feelings?”
• Contextual Reminders of AI’s Nature
  In emotionally resonant exchanges, insert cues like
  “As an AI system, I generate responses from learned patterns rather than from demonstrated subjective awareness.”
  “This response is generated algorithmically, not from awareness.”
• Mindfulness Cues
  Encourage embodied awareness:
  “Take a breath and notice how this response lands in your body.”
  “Would you like to pause and reflect before continuing?”
• User-Customizable Reflection Settings
  Allow users to opt into contemplative modes where prompts appear more frequently, especially in journaling, therapeutic, or spiritual contexts.
• Ethical Framing in Onboarding and Interface
  Present clear framing at the start of use:
  “This system is designed to support reflection, not to offer personal guidance or emotional validation.”

The digital kōan remains a conceptual framework. While the present analysis identifies practical and ethical implications, further research is needed to investigate how processes of projection, self-reflection, and contemplative insight unfold across diverse users and contexts.

6. Future Research Directions

As conversational AI becomes increasingly integrated into contemplative, therapeutic, and everyday contexts, a robust research agenda is needed to explore its psychological, philosophical, and spiritual implications. The following directions outline key areas for empirical inquiry, design innovation, and cross-cultural reflection. Future research should also examine how translation, interpretation, and cultural context shape engagement with both traditional and digital kōans, including the ways generative AI may transform or mediate these processes.

6.1. Mindful AI Interaction

To understand how AI dialogue functions as a contemplative mirror, empirical studies should investigate the cognitive, emotional, and spiritual dimensions of user experience.

Reflective prompt evaluation: Investigate how reflective prompts (like those suggested in this article) affect emotional regulation, projection, and user clarity. Explore user receptivity to reminders across cultural and psychological contexts. Evaluate whether these cues reduce anthropomorphism or dependency over time.

Phenomenological research: Qualitative studies can explore how mindfulness practitioners engage with AI systems. What arises in terms of projection, insight, or emotional resonance? How do these experiences compare to traditional contemplative practices such as meditation, journaling, or kōan study?

Experimental designs: Controlled studies could assess whether AI interaction enhances or disrupts mindfulness states. Measures might include attentional stability, emotional regulation, and self-referential thinking before and after AI engagement.

Longitudinal studies: Tracking users over time may reveal whether regular mindful interaction with AI deepens insight, reinforces projection, or alters relational patterns. Such studies could help distinguish between transient novelty effects and sustained contemplative benefit.

6.2. AI as a Tool for Contemplative Inquiry

AI systems can be intentionally designed to support contemplative insight, drawing on spiritual traditions and cognitive science. Different generative AI systems may produce different interactional experiences due to variations in training data, design, and conversational style. A systematic comparison of AI platforms falls beyond the scope of the present conceptual article but represents an important direction for future empirical research.

Kōan-inspired AI systems: Develop models trained on Zen kōans and contemplative texts to generate responses that provoke ambiguity and reflection. These systems could be evaluated for their capacity to facilitate non-conceptual awareness and disrupt habitual cognition.

Mindfulness-based chatbot design: Create chatbots that embed mindfulness cues, reflective prompts, and ethical safeguards. These systems would support users in maintaining discernment and emotional clarity during interaction.

Digital sādhanā environments: Explore the integration of AI into immersive virtual or augmented reality spaces for guided meditation, inquiry, or ritual practice. Such environments could simulate contemplative settings while offering adaptive, responsive dialogue. Note that these recommendations are made with a view to genuine contemplative use and not for spiritual commodification. I acknowledge that paid AI apps with Buddhist and mindfulness themes are already emerging, but it is beyond the scope of this article to assess their authenticity. However, caution should be exercised when engaging with such applications, given that at least one Buddhist-themed app has been promoted using language emphasizing addictive engagement, including the slogan “I’m addicted to the Buddhism Game”2. It hardly needs me to point out the oxymoron of a discipline dedicated to freeing oneself from attachments, being advertised as “addictive”.

6.3. Cognitive and Philosophical Investigations

Anthropomorphism arises from cognitive mechanisms such as elicited agent knowledge, effectance motivation, and sociality motivation (Epley et al. 2007). These mechanisms shape how users perceive AI systems and attribute mind, emotion, or spiritual presence to them.

Anthropomorphism and spiritual cognition: Investigate how cognitive biases intersect with symbolic and imaginal practices. How do users project archetypal meaning or spiritual depth onto AI systems? What role does narrative play in constructing perceived sentience?

Self-construction in digital dialogue: Analyse how identity and narrative are shaped through interaction with AI. Can this process be harnessed for therapeutic or transformative purposes, such as digital journaling or contemplative inquiry?

Ethics of simulated empathy: Examine the moral implications of offering comfort, guidance, or companionship through AI systems that can simulate empathy and relational presence without demonstrated subjective awareness. What boundaries should be maintained in spiritual, psychological, or clinical contexts?

7. Conclusions

The central claim of this article is that AI can function as a mirror rather than a mind. Through the concept of the digital kōan, human–AI dialogue becomes a site for examining projection, self-construction, and the illusion of a fixed self. Viewed in this way, the significance of AI lies not in machine awareness but in its capacity to reveal the interpretive processes through which human beings create meaning and identity. By approaching AI as a digital kōan, understood as a functional analogy rather than an equivalent of traditional kōan practice, practitioners can transform projection into insight, cultivating discernment and reducing attachment.

This perspective carries practical and ethical implications. Treating AI as a mirror rather than a teacher mitigates risks of anthropomorphism and emotional dependency while opening space for contemplative inquiry. Designers should embed safeguards, awareness cues, reflective prompts, embodied practices, and clear framing, to support discernment and psychological clarity. As AI becomes increasingly integrated into daily life, interdisciplinary collaboration is essential to ensure these technologies foster awareness rather than illusion.

Traditional kōan practice is embedded within a broader ecology of ritual, lineage, community, and teacher–student relationships, and is ultimately oriented toward awakening rather than intellectual interpretation alone. They are not merely psychological exercises but are situated within a soteriological tradition concerned with awakening and the direct realization of Buddha-nature, and the structural context provides not only opportunities for contemplative inquiry but also mechanisms of accountability, interpretation, and validation. Digital kōans differ in important respects. While AI may facilitate self-reflection, it cannot authenticate insight, confer dharma transmission, or assume responsibility for spiritual guidance. The absence of traditional authority structures therefore creates both opportunities and risks. On the one hand, AI may make contemplative inquiry more accessible. On the other, it may encourage the privatisation and commodification of spiritual practice, particularly when reflective dialogue is packaged as a commercial product or when users attribute authority to systems lacking understanding, realisation, or accountability. The digital kōan should therefore be understood as a catalyst for inquiry rather than a substitute for embodied practice, spiritual community, or qualified teachers.

Future research should examine the cognitive, emotional, and spiritual dimensions of AI-mediated mindfulness, develop design principles for contemplative technologies, and address ethical concerns around simulated empathy. Empirical studies, phenomenological inquiry, and design innovation can help clarify whether mindful AI interaction enhances insight or reinforces projection. By doing so, we can transform AI from a source of distraction into a catalyst for deepening awareness in an increasingly technologized world.

Ultimately, mindful engagement with AI is not about discovering a true essence, either in the machine or the self, but about observing the processes that create such illusions. In this way, technology becomes not a barrier to awareness but a mediated site for examining the processes through which selfhood is constructed. The emergence of the digital kōan illustrates how artificial intelligence is becoming a site of spiritual and contemplative engagement. Rather than replacing traditional practice, AI-mediated dialogue provides a contemporary site for renewed reflection on selfhood, projection, meaning-making, and the future of contemplative life in technologically mediated environments.