Artificial Intelligence and Occupational Therapy: From Emerging Occupation to Educational,Practice,and Policy Imperative

Education: Teaching With and About AI

Across occupational therapy programs, students are already using AI tools, such as ChatGPT, Grammarly, and DALL·E, to support learning. They use them to summarize dense readings, draft case documentation, prepare for practical exams, and even simulate interview questions or treatment planning conversations. For many students, especially those balancing multiple roles, neurodivergent learners, or those for whom English is an additional language, AI tools offer a sense of accessibility, efficiency, and confidence (Thacker et al., 2024).

In the absence of structured discussion or institutional guidance, however, AI use in occupational therapy education is largely opaque. Students are often unsure whether using AI is acceptable. Some describe feeling as though they are cheating, even when it supports, not replaces, their learning. Faculty views also vary: Some see AI as a valuable tool for critical thinking; others fear it promotes superficiality or dependency (Hood, 2024).

This uncertainty is understandable but increasingly untenable. Although AOTA offered early guidance (Thacker et al., 2024), most programs lack curricular competencies or ethical frameworks. As a result, students’ experiences remain inconsistent and are often shaped more by faculty discretion than shared standards. Increasingly, students also cite environmental impact as an ethical consideration, reflecting a broader awareness of AI’s societal and ecological implications.

Research conducted by Mansour and Wong (2024) illustrates the potential of intentional AI integration. In their study, occupational therapy students who engaged with ChatGPT as a study partner during fieldwork preparation reported increased confidence and perceived readiness. Faculty noted that when assignments required students to critically reflect on AI-generated content rather than simply accept it, students demonstrated deeper engagement with core occupational therapy concepts. Although Mansour and Wong observed no statistically significant shift in readiness, their design foregrounded critical-appraisal routines: Students had to interrogate bias, provenance, and accuracy for each AI output, illustrating that AI pedagogy should follow a generate → critique → revise cycle rather than one-way adoption. Recent commentary therefore urges programs to embed AI literacy and accountability frameworks across curricula so graduates become proactive partners rather than passive users of technology.

This aligns with what I have observed in my own teaching: When I position AI not as a shortcut, but as a tool for exploration and reflection, students begin to understand both its affordances and its limitations. They become more discerning thinkers. More important, they begin to see themselves as future professionals who are capable of navigating technological complexity with integrity. This approach is further elaborated in a forthcoming article in the Journal of Allied Health, where I describe the GenAI Learning Partner—a structured, theory- informed model for integrating generative AI into occupational therapy education (Jozkowski & Tabak, in press). To support the use of such models, academic programs need the following:

faculty development opportunities to explore pedagogically sound uses of AI;

Occupational therapy educators must prepare students for the clinical realities they will face. Many will enter fieldwork or practice settings where AI is embedded in documentation platforms, care planning tools, or patient engagement systems. Understanding these tools, and being able to evaluate them, is now a core professional skill.

Clinical Practice: Navigating AI in Clinical Reasoning and Documentation

AI is increasingly embedded in health care systems, including those where occupational therapists practice. Whether in the form of auto-generated documentation suggestions, machine learning diagnostic tools, or adaptive telehealth interfaces, these systems are already shaping how practitioners plan, deliver, and record care.

However, many occupational therapy practitioners are encountering these tools without adequate preparation. Unlike their peers in medicine and nursing, occupational therapists rarely receive training in algorithmic literacy, digital ethics, or how to critically assess the role of AI in clinical decision-making. Most are expected to adapt on the fly, a strategy that may suffice for basic workflows but fails to safeguard person-centered care.

The College of Occupational Therapists of Ontario (2023) provided early guidance, affirming that although AI can support documentation and assessment occupational therapists remain ethically and legally accountable for all clinical decisions. This reinforces a key point: AI must augment, not replace, clinical reasoning. For example, Rowe and Ward (2025) envisioned human–AI team members whose pattern recognition capacity lets clinicians “quickly synthesize relevant research evidence, consider alternative approaches, and expand their own clinical judgement” (p. 1), freeing saved time for authentic therapeutic interaction. As with VR interventions, occupational therapists’ person-in-context expertise can “inform the development of therapeutic … virtual worlds” (Foran, 2011, p. 195) by adapting digital environments to individual skills and goals—a design stance that translates directly to AI- supported platforms.

However, applying this in practice is far from straightforward. Many AI systems, in particular those used in documentation or decision-support software, operate as closed systems: Their algorithms are proprietary, their training data unknown, and their recommendations opaque. This raises critical questions about occupational justice: How should a therapist respond when an AI-generated care plan does not align with the client’s goals? How can bias in training data be recognized and addressed? How does one advocate for culturally responsive care when the system default is standardized and nontransparent? AI tools that privilege majority-class data can lead “to bias and discrimination against marginalised groups” (Kaelin et al., 2024, p. 3); therefore, we must demand explicit human-centered safeguards.

These are not hypothetical concerns. For example, a pediatric occupational therapist may use an assessment platform that prioritizes neurotypical motor patterns, or a community-based therapist may serve clients for whom AI-based remote monitoring is inaccessible. Without tools to critique and adapt these systems therapists risk reinforcing the very inequities occupational therapy aims to address.

At the same time, AI has significant potential to enhance occupational therapy practice. Wearable sensors can track fine-motor engagement or physical activity levels in real time. AI- enhanced home monitoring can alert caregivers to patterns of concern. Natural language processing can assist in translating client narratives into occupational profiles. These applications require occupational therapy’s voice—not just in use but in design, evaluation, and advocacy. To advance ethically grounded AI integration in practice, the profession must

develop practice guidelines that articulate when and how to use AI tools in documentation, assessment, and care planning;

Advancing these goals depends on a solid research foundation to inform clinical tools, educational strategies, and ethical safeguards.

Occupational Engagement: Expanding Occupational Therapy’s Scope of Practice

One of the most profound, and perhaps most underexplored, implications of AI for occupational therapy is this: AI use is becoming an occupation in its own right. Clients across the life course are engaging with AI tools in ways that are habitual, goal directed, and personally meaningful. Concepts borrowed from VR scholarship—immersion, presence, flow, and embodiment—offer a ready lens for analyzing AI-mediated occupations, echoing Foran’s (2011) argument that “an understanding of occupational experience can be enhanced by taking up constructs in virtual reality theory that are similar to those in occupational therapy and occupational science” (p. 190). These interactions, whether a young adult using ChatGPT to cowrite stories as a leisure activity, a neurodivergent client practicing conversational scripts with an AI chatbot, or an older adult using a voice-activated device to manage medication schedules and reduce isolation, are shaping routines, identities, and participation. In workplace contexts, AI-mediated reskilling platforms could help displaced workers learn new roles; Liu (2018) warned that the Fourth Industrial Revolution is likely to trigger an even larger wave of automation-related job losses than previous industrial shifts.

As occupational therapists, we are trained to assess and support how people do what they want and need to do. We attend to the tools, environments, and roles they engage. AI fits within this purview, yet we are not consistently treating it as such. The Canadian Association of Occupational Therapists (2023) explicitly stated that emerging technologies should be recognized as both tools and sites of occupation that shape access, agency, and engagement. Ignoring AI’s role in occupation limits our capacity to serve clients who are increasingly dependent on or affected by these systems.

Occupational science provides a conceptual framework that reinforces this position. As Laliberte Rudman (2015) described, occupation is socially and technologically situated, shaped both by individual preferences and broader sociotechnical systems that determine what forms of doing are available and valued. AI-mediated occupations represent a shift in both form and meaning: They often blend human and machine agency, require new literacies, and are enacted in digital rather than physical spaces. As Townsend and Polatajko (2007) asserted, enabling occupation in such contexts requires attention to structural, cultural, and technological barriers.

Recognizing AI use as an occupation means asking different questions in practice:

What routines involve AI use, and how does AI support or hinder participation?

Recognition of AI use as an occupation also means expanding our assessment and intervention models. Few standardized tools address AI-mediated activities. Clinicians and researchers must begin incorporating AI into assessments and help clients evaluate whether and how it supports or undermines participation.

Crucially, this also includes recognizing digital divides. Clients without reliable internet access, updated devices, or AI literacy may be excluded from occupations that are increasingly expected in school, work, and health care contexts. Occupational therapists have a role in advocating for policies related to inclusive technology and for clients’ right to access and understand the tools that help them shape their lives.

Governance, Law, and Regulation

AI governance now lags behind clinical innovation. Mohamed et al. (2025) proposed a regulatory framework that safeguards ethical AI use, tackles data governance and privacy issues, and requires practitioners to complete ongoing training in AI literacy and ethics. Their model calls for joint oversight by practitioners, developers, regulators, and service users; continual postmarket surveillance; and flexible policies that are able to “tolerate regulatory shifts and changes” as the technology evolves. Embedding such a framework into the field would complement existing accreditation requirements and provide clear guardrails for practice and education.

Accreditation: Aligning Standards With Practice Realities

As a complement to these regulatory concerns, forthcoming accreditation standards likewise underscore the need for occupational therapists to demonstrate AI literacy and ethical competence before entering practice. For example, in the 2023 Accreditation Council for Occupational Therapy Education (ACOTE®; 2023) Standards and Interpretive Guide, artificial intelligence is referenced in the definition of educational technology. Although this acknowledgment is welcome, it is insufficient. The reference is narrow, embedded only in the context of teaching strategies and digital tools. It offers no interpretive guidance for how AI intersects with broader competencies, such as critical reasoning, ethical practice, or documentation, and it leaves programs without direction for implementation.

This lack of clarity is not merely an academic issue; it affects what programs teach, what faculty prioritize, and how students are prepared. It also leads to hesitancy among educators: Is AI something to explore or avoid? How can programs document their AI-related efforts during accreditation reviews without clear standards or rubrics?

As one of the primary mechanisms through which professional values are transmitted and evaluated, accreditation must not lag behind practice trends. ACOTE has a unique opportunity to lead by clarifying how AI fits across competencies, not only as an educational tool but also as an area of professional reasoning and behavior. I recommend that ACOTE

expand the interpretive guide to map AI to standards on clinical reasoning, documentation, and professional behaviors;

This work can align the profession’s educational standards with the realities of contemporary practice while also positioning occupational therapy as a leader in health professions education innovation. As formal standards evolve, research institutions, such as the American Occupational Therapy Foundation (AOTF; 2025), are already laying groundwork for the profession’s future with AI. AOTF’s focus on technology, participation, and occupational justice positions the group as a key partner in this work.

Recommendations: A Coordinated Call to Action

To engage with AI across education, practice, and daily life, the occupational therapy profession must act in a coordinated and intentional way. Piecemeal efforts, though valuable, will not be sufficient to support widespread, ethical, and equitable integration. The following recommendations are directed to key stakeholders.

Conclusion: The Time Is Now

Occupational therapy has long adapted to technological change, policy shifts, and evolving views of health and participation. We have embraced telehealth, reimagined school and community practice, and led conversations about trauma and justice, always grounded in our commitment to enable meaningful doing, uphold equity, and reflect the full complexity of people’s lives.

AI represents a new horizon, one that is already influencing what our students need to learn, how our documentation systems function, and how clients organize their daily routines. It brings both opportunities and tensions. AI can support adaptive environments, efficient care, and new forms of engagement, but it also has the potential to exacerbate inequities, automate bias, and distance us from the relational aspects of practice that are often most therapeutic. In addition to ethical and equity concerns, emerging discourse also highlights the environmental impact of AI technologies. The energy demands of training and deploying large-scale AI models raise important questions about sustainability and responsible use. As occupational therapy embraces digital tools, we must also consider their ecological footprint and advocate for practices that align with broader commitments to planetary health and sustainable care.

These are not distant concerns; they are already reshaping our professional landscape. And although the pace of change can feel disorienting, our profession is not without preparation. Occupational therapy has always paid attention to systems, tools, and habits. We already have the conceptual frameworks to approach this with care: We understand occupation as situated, technology as shaping environments, and participation as inherently contextual and political.

What is needed now is not a reinvention of who we are but a reassertion of how we apply what we know, a deliberate, shared effort to engage with AI from within our practice, pedagogy, and scholarship. We do not need to be first adopters of every tool, but we do need to bring our lens to the table—early, critically, and with clarity about what matters most. This moment calls for action, not reaction. Let us meet it on our own terms.