INTRODUCTION

For decades, the cornerstone of teacher preparation has been the cultivation of critical reflection which is the ability of pre-service teachers to move beyond mere descriptive accounts of their practice toward a rigorous interrogation of their assumptions, biases, and pedagogical choices. Central to this process is Schön’s (1983) concept of the ‘reflective practitioner’, which posits that professional growth occurs in the “swampy lowlands” (p.42) of messy, real-world classroom problems. Having pointed that out however, the path to critical reflection is often fraught with challenges. Pre-service teachers frequently struggle to progress beyond low-level narrative descriptions where reflection becomes more of a performative exercise in reporting what happened rather than analysing why it happened or how it could be different. The emergence of generative artificial intelligence (GenAI) offers a potential digital antidote to this pedagogical impasse. Rather than serving as a passive source, GenAI can function as a dialogic critical friend (CF). By engaging pre-service teachers in sustained and interactive dialogues with AI, it can systematically disrupt superficial reporting, prompting them to confront cognitive dissonance, unpack classroom dynamics, and transition from descriptive narratives to transformative critical reflection.

While initial academic discourse focused heavily on the risks of plagiarism and the erosion of original thought, a new body of literature is shifting toward GenAI as an instrument for collaborative reflexivity. This article explores the collaborative merging of GenAI and teacher education, specifically examining how these tools can serve as a critical friend (CF) that provides the scaffolding necessary for pre-service teachers to engage in deeper, more critical layers of reflective inquiry. By synthesizing current research on AI-mediated reflection, this article argues that the goal of integrating GenAI is not to automate the reflective process, but to re-humanize it. Through a thematic review of the literature, this paper will examine how GenAI can be leveraged to simulate classroom complexities, challenge teacher identity, and bridge the gap between theoretical knowledge and practical understanding.

THE CRITICAL FRIEND (CF) CONCEPT IN REFLECTIVE PRACTICE

Farrell (2015) posits that meaningful reflection cannot be a purely solitary, meditative act rather, it thrives on dialogic engagement and evidence-based inquiry. A pivotal concept within this framework is the “critical friend” (CF), a trusted dialogic partner who holds up a metaphorical mirror to the educator, offering empathetic yet structurally challenging feedback to unpack deeply embedded beliefs and instructional choices (Farrell, 2019). While finding an available, non-evaluative human peer during the challenging practicum experiences remains a systemic challenge for pre-service teachers, emergent technologies offer a potential solution. Generative Artificial Intelligence (GenAI), acting as a ‘critical friend’, (CF) can fill this dialogic void by serving as an active, constant collaborator in knowledge construction (Bozkurt & Bae, 2024). When pre-service teachers input their lesson plans or post-teaching reflections, GenAI tools have the potential to act as a pedagogical sounding board that actively interrogates the user’s implicit biases, classroom management assumptions, and differentiation strategies (Mishra & Heath, 2024). By posing targeted, scaffolded questions that require novice teachers to justify their instructional decisions, GenAI can transform standard, descriptive journaling into highly critical analysis, ultimately transforming reflection from a mandatory required task into an internalized, cyclical catalyst for lifelong professional growth (Farrell, 2015).

CONCEPTUALISING GenAI AS A CRITICAL FRIEND IN PRACTICE

In order to transition from passive offloading to cognitive augmentation, the operationalization of Generative Artificial Intelligence (GenAI) as a critical friend (CF) requires pre-service teachers to implement structured, deliberate prompting procedures. In Farrell’s (2015, 2019) conceptualization of dialogic reflection, the CF must intentionally disrupt a novice educator’s comfort zone by exposing their underlying pedagogical assumptions. When translating this to the human-AI interactions, pre-service teachers must avoid generic information retrieval by utilizing role-based, multi-turn prompts that explicitly instruct the GenAI tool to provide constructive critique rather than complimentary validation (Bozkurt & Bae, 2024; Mishra & Heath, 2024). The prompt examples illustrated below exemplifies how rather than asking the GenAI tool to review a lesson plan, a more context-specific prompting can potentially establish an interactive, non-evaluative critique repartee between the pre-service teachers and the GenAI tool.

 “Act as an expert ESL teacher educator and a non-evaluative critical friend. My interactive reading lesson, did not go as planned because the activities were not student-centred enough. Can you rewrite my lesson plan and come up with more engaging reading activities.“

Prompt Sample 1: Rewrite and Review Lesson Plan

“Act as an expert ESL teacher educator and a non-evaluative critical friend. I will provide a description of an interactive reading lesson I conducted that did not go as planned. Do not rewrite my lesson plan or offer immediate solutions. Instead, identify two potential blind spots or unexamined assumptions regarding student engagement in my description, and ask me two challenging, open-ended questions that would require me to justify my choice of instructional scaffolding.

Prompt Sample 2: Identify and Examine Potential Underlying Assumptions

Through this dialogic framework, the GenAI acts as a constructive critique, with structured, context-specific prompts that compel novice teachers to evaluate the complexity of a language lesson or to interrogate the efficacy of their classroom management strategies. By responding to these AI-generated interactions, pre-service teachers engage in deep, double-loop learning, moving beyond a superficial, descriptive summary of events into the highly critical “So what?” and “Now what?” phases of reflective inquiry (Farrell, 2013). This prompt-driven methodology ensures that the technology scaffolds rather than supplants cognitive dissonance and functions as a catalyst for internalized, professional growth.

THE IMPORTANCE OF PROMPT TRAINING AND ETHICAL CONSIDERATIONS IN USING GenAI AS A CRITICAL FRIEND (CF)

The integration of Generative Artificial Intelligence (GenAI) as a critical friend (CF) in teacher education is heavily dependent on the user’s capacity to navigate the dialogic exchange. Without explicit and rigorous prompt training, pre-service teachers frequently fall into the trap of superficial information-seeking, utilizing unstructured inputs that yield generic, descriptive, and often overly complimentary validation from the GenAI tool (Mishra & Heath, 2024). In Farrell’s (2015, 2019) traditional critical friendship (CF) framework, the primary value of the dialogic partner lies in their ability to disrupt complacency and push the reflective boundaries. When a teacher trainee lacks prompt literacy, this analytical friction is absent as the AI defaults to sycophantic praise or immediate, superficial solutions, which can effectively circumvent the pre-service teacher’s productive cognitive dissonance (Bozkurt & Bae, 2024). Consequently, a lack of structured prompt cultivation transforms what should be an instrument for deep and transformative learning into a performative obligatory task which reinforces a culture of superficial compliance rather than authentic professional growth.

While GenAI has the potential to present a highly feasible alternative to traditional, resource-constrained peer mentorship models, its implementation however raises ethical dilemmas that teacher educators must systematically address first. One such dilemma being, the highly responsive and conversational nature of GenAI platforms may mislead the pre-service teachers into treating a statistical pattern-recognition tool as an inherently wise and unbiased mentor (Mishra, Oster, & Henriksen, 2024). According to Daher (2025), this misplaced trust often manifests as un-critical deference, causing the pre-service teachers to accept automated critiques that may be heavily skewed by systemic algorithmic biases that may be totally devoid of localized cultural awareness. Furthermore, GenAI-mediated reflections are fundamentally limited by a lack of trust and relational depth. Because the AI has no prior knowledge of a teacher’s professional history, core educational values, or unique socio-emotional context, it cannot achieve the reciprocity and mutual understanding that characterize human critical partnerships (Hockly, 2023). Addressing this limitation requires a hybrid approach in which pre-service teachers should use GenAI only to complement human dialogue. By explicitly setting expectations for the AI, critically analyzing its outputs, and weaving in feedback from human peers or mentors, teacher educators can ensure that the deeper, trust-dependent aspects of reflection remain sustained

FINAL THOUGHTS

To fully leverage the transformative capacity of generative AI as a critical friend, teacher education programs must move toward an integrated model that pairs technological capability with critical pedagogy. This involves embedding critical AI literacy, ethical protocols, and scaffolded reflection directly into teacher training structures. The curriculum must train future teachers to master prompt design, audit AI responses for hidden biases, and intentionally synthesize automated insights with human mentorship to preserve professional autonomy (Hockly, 2023). This systemic shift relies heavily on teacher educators to model these practices transparently, guiding trainees through reflective discussions regarding the impact of automation on the teaching identity (Moorhouse and Kohnke, 2024). Ultimately, when positioned within a comprehensive educational framework, utilizing generative AI as a critical friend can significantly elevate reflective practice, acting as a powerful catalyst for pedagogical creativity, deep inquiry, and lifelong professional learning. This digital antidote is only valuable if pre-service teachers are explicitly trained to look past its polished, automated surface and engage directly with the messy, uncomfortable, and inherently human complexities of their own developing practice as teachers in training.

References

Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action.

Farrell, T. S. C. (2015). Promoting teacher reflection in second language education: A framework for TESOL professionals. Routledge.

Farrell, T. S. C. (2019). Reflective teaching. TESOL International Association.

Bozkurt, A., & Bae, H. (2024). May the force be with you JedAI: Balancing the light and dark sides of generative AI in the educational landscape. Online Learning, 28(2).

Daher, W. (2025). Ethical AI integration principles in teacher education: Navigating bias, privacy, and systemic limitations. Journal of Digital Learning in Teacher Education, 41(1).

Mishra, P., & Heath, M. K. (2024). Beyond plagiarism and productivity: Ethical evaluation of generative AI in teacher education. Journal of Technology and Teacher Education, 32(1).

Farrell, T. S. C. (2013). Reflective practice in ESL teacher development groups: From a lonely journey to a collaborative band of travellers. Palgrave Macmillan.

Mishra, P., Oster, A., & Henriksen, D. (2024). Generative AI, teacher knowledge and educational research: Bridging short- and long-term perspectives. Tech Trends, 68(2), 241-247.

Hockly, N. 2023. ‘Artificial Intelligence in English Language Teaching: The Good, the Bad and the Ugly.’ RELC Journal 54(2):445–51. 

Moorhouse, B. L. and Kohnke L. 2024. ‘The Effects of Generative AI on Initial Language Teacher Education: The Perceptions of Teacher Educators.’ System 122:103290.

Written by Shubashini Supiah