Abstract
English language teaching is an area that faces many changes during the digital transformation of education. The current trajectory in the professional development (PD) models is that of progressively more independent and technologically aided PD models. The current study focused on the perspectives and practices of the in-service English language teachers in Türkiye regarding the artificial intelligence (AI)-based tools to inform their self-regulated (SRL) and their PD. With the advent of AI platforms such as ChatGPT, and Gemini, the agents of professional learning, personal learning, reflection, and co-creation of content have seen a revolution in the way teachers interact with these agents now. However, despite the increasing international zeal over using AI in education, there is scant empirical literature that broadly covers the adoption of AI among in-service language teachers in practice in Türkiye. Using a unified conceptual framework of SRL, a qualitative research design was employed to explore the motivations, perceived benefits including the impediments and sociocultural dynamics that influence in-service English language teachers’ SRL with AI tools. The data, collected through semi-structured interviews with 10 teachers, were analyzed using the reflexive thematic analysis method. Synthesized findings presented in clusters and core dimensions indicated: (a) AI as a scaffolding tool for metacognitive regulation, (b) efficiency versus over-reliance, and (c) the role of informal peer networks as a mediator in the absence of formal institutional support. The findings generated within the study add to the emerging understanding of the ecology of teacher learning and implications of emerging technologies in local contexts of ELT.
Plain Language Summary
Abstract Introduction: The field of English language teaching is undergoing significant change as it navigates the digital transformation of education. The current trend in professional development (PD) models highlights a move towards more autonomous and technology-facilitated forms of teacher learning. The emergence of artificial intelligence (AI) platforms, such as ChatGPT and Twee, has revolutionised how teachers engage in professional learning, personal reflection, and the co-creation of content for others. However, despite the global enthusiasm for AI in education, there is limited empirical research that explores the adoption of AI by in-service English language teachers in Türkiye. Aim: This study aimed to explore how in-service English language teachers in Türkiye perceive and engage with AI tools in relation to their self-regulated learning (SRL) for professional development. Method: A qualitative research design was adopted. Data collection focused on teachers’ motivations, perceived benefits, challenges, and the sociocultural dynamics that shape their SDL practices with AI tools. Results: The findings revealed that teachers engaged with AI tools both as agents of professional learning and as resources for reflection and content creation. Teachers valued the autonomy and immediacy of AI support but also reported challenges related to reliability, ethical concerns, and the need for contextual adaptation. Their practices were shaped by sociocultural factors, including institutional expectations and peer collaboration. Conclusion: The study contributes to the emerging understanding of the ecology of teacher learning in the digital age. It highlights the role of AI tools in supporting professional development and points to implications for future practice, including fostering teacher agency, addressing challenges of responsible use, and integrating emerging technologies into the local contexts of ELT.
Introduction
Professional development (PD) refers to any activity, strategy or program meant to alter teachers’ beliefs and practice with an aim to enhance student learning outcomes (Guskey, 2002). In addition, PD is also a need for language teachers to ensure that their practices are relevantly pedagogical, a need which has been elicited in a rapidly changing world, more with the positive impact of technology on the teaching of the English language (Richards & Farrell, 2010). With the spread of digital technologies, numerous online resources have made SRL a convenient and easily accessible form of self-regulated professional development (SRPD; Bates, 2022; Botero et al., 2018; Hubbard & Schulze, 2025). Although the pedagogy of English language teaching in Türkiye has seen vast growth and development over the years, rapid technological advancements in technology, especially in artificial intelligence (AI), have led to demands for an overhaul of English language teacher PD programs. The traditional PD programs in Türkiye have typically been top-down and prescriptive models of PD (Kırkgöz, 2008; Uysal, 2012), but there is growing acknowledgment that PD should be under more autonomous and flexible learning models.
SRL (Self-regulated learning) in this context becomes a critical mechanism for teacher agency. Since SRL is a deliberate plan where the learner takes major responsibility for planning, implementing, and evaluating their learning (Merriam & Bierema, 2014). This decision-making process enables the teachers to take an independent initiative to identify their learning needs, the learning resources required, the learning strategies which are to be employed and the outcome is evaluated (Zimmerman, 2001, 2008). Recent studies emphasize the prospects of SDL for agency enhancement of teachers, PD of teachers, and sustainability of PD experience (Calvert, 2016; Mercer et al., 2022).
In this regard, a variety of digital tools and products (webinars, MOOCs, educational websites, social media, etc.) are used in teachers’ SDL journeys, as they increasingly rely on online environments for their developmental needs (Castaño-Muñoz et al., 2018). In this evolving ecosystem, more recently, AI-based tools that include ChatGPT, Grammarly, QuillBot, and text-to-speech tools have emerged as new ecosystem components with affordances for planning, scaffolding, content-creation, assessment, and language bolstering (Barnes & Tour, 2025; Kılıçkaya & Kic-Drgas, 2026; Kic-Drgas & Kılıçkaya, 2024; Ou et al., 2024). Recent studies regarding first reactions and impressions of ChatGPT show an overall view on a global scale on the usage of this technology in schools and demonstrate how much potential it has to change the perception of education formulations (Ravšelj et al., 2025). For the purposes of this study, AI tools are described as Generative AI (GenAI) platforms (e.g., ChatGPT) and AI writing assistants (e.g., Grammarly) that rely on machine learning to process natural language and generate or rectify content.
While AI has been globally explored in education, few know to what extent in-service language teachers in Türkiye interact with AI applications regarding SDL. Most of the research has focused on institutional implementations or preservice teacher training hence leading to a crucial gap of focusing on every day, informal, and situated practice of in-service teachers. To fill this gap, the present study investigates the dynamics of AI tool adoption among in-service English language teaching professionals in Türkiye, examining their motivations, perceived benefits and challenges, and the institutional and cultural contexts that influence their engagement. Using Zimmerman’s (2000) model of Self-Regulated Learning, the below-mentioned research questions are answered:
How do in-service EFL teachers utilize AI tools to scaffold the forethought, performance, and reflection phases of their self-regulated professional learning?
What perceived benefits and challenges do teachers encounter when integrating AI into their self-regulated learning practices?
How do institutional and sociocultural factors influence teachers’ adoption of AI for professional development?
These three questions are conceptually integrated through an ecological perspective of teacher development. RQ1 establishes the internal cognitive process of how teachers’ self-regulation manifests through AI scaffolding. However, because this process is not value-neutral, RQ2 examines the evaluative “friction,” the perceived benefits, and challenges, that dictates whether these self-regulated practices are sustained. Finally, since neither process nor perception occurs in a vacuum, RQ3 situates these individual experiences within the broader institutional and sociocultural structures that either catalyze or constrain AI adoption. Together, they provide a holistic view of the AI-mediated professional learning ecosystem.
Literature Review
The purpose of this literature review is to establish a conceptual and empirical rationale for examining how in-service EFL teachers use AI tools within SRL. Specifically, the review (a) clarifies the conceptual distinctions among SDL, SRL, and adult learning/teacher agency to define the learning processes addressed in RQ1; (b) synthesizes emerging evidence on AI in teacher education and professional learning, with particular attention to reported affordances and risks, to motivate RQ2; and (c) examines institutional and sociocultural conditions shaping teachers’ technology adoption, especially informal professional learning networks and policy contexts, to frame RQ3. Taken together, these strands foreground an unresolved empirical gap: while AI-enabled professional learning is rapidly expanding, we still know relatively little about teachers’ day-to-day, informal, self-regulated uses of AI tools within local ELT ecologies (such as Türkiye), and about how institutional structures and peer networks mediate these practices.
Theoretical Framework and Professional Development
Distinguishing SDL, SRL, and Professional Development
To provide conceptual clarity, this study distinguishes between SDL and SRL, anchored in Adult Learning Theory. While SDL refers to the macro-level design of the learning trajectory where the learner decides what and why to learn, SRL describes the micro-level cognitive processes (planning, monitoring, reflecting) involved in executing that learning (Zimmerman, 2000). This approach aligns with Knowles’ (1984) Andragogy, which posits that as individuals mature, their self-concept moves from dependency toward self-direction, driven by internal motivation and the need to solve real-life problems. Teacher agency functions as the catalyst within these constructs, empowering teachers to initiate SDL and sustain SRL despite systemic constraints.
Rossner (2017) defines PD as “The professional growth that teachers achieve in the process of gaining experience and knowledge and reflecting on their teaching” (p. 169). This development can be achieved through various approaches, the very first of which is the traditional approach, which is characterized as one-shot “sit-and-get” programs with a top-down approach to disseminating knowledge. Major drawbacks of this approach include considering teachers “as consumer modes of teacher learning” (Borg, 2015, p. 5), providing a one-size-fits-all view without considering individual needs, strengths, and weaknesses, leaving no responsibility to teachers so that they can monitor their own learning. This traditional view “give[s] way to alternative PD structures that allow for self-directed, collaborative, inquiry-based learning that is directly relevant to teachers’ classroom lives” (Johnson, 2006, p. 243). These might include teacher inquiry seminars, teacher study groups, professional learning communities, and collaboration and cooperation among teachers. Unlike this traditional view, via SDL, also known as teacher agency, “the teacher controls and takes ownership of their own PD journey, in contrast to prescribed, non-voluntary forms of PD perhaps mandated by institutions or local educational authorities” (Mercer et al., 2022). This can be attributed to the fact that teachers know best their professional needs, interests and learning preferences.
AI in Teacher Education and Professional Learning
Since the public release of ChatGPT in late 2022, research on generative AI in language education has expanded rapidly. Recent syntheses map emerging themes such as classroom uses, learner-facing outcomes, and persistent concerns about reliability, bias, and academic integrity, while also noting that teacher learning and PD remain comparatively under-examined within this literature. Systematic reviews of early work on ChatGPT and language education highlight that much of the initial wave prioritized learner tasks and assessment debates, leaving important gaps regarding how teachers themselves learn to use GenAI tools responsibly, critically, and strategically in their professional growth (e.g., Li et al., 2024). This study responds to that gap by focusing on in-service EFL teachers’ self- SRL-informed professional learning with AI tools in Türkiye. In ELT specifically, scholars emphasize that GenAI may offer efficiency and feedback affordances, yet it also raises domain-relevant risks (e.g., plausible but inaccurate language explanations, overconfident feedback, and blurred authorship), which makes teacher judgment and AI literacy central (e.g., Hockly, 2023).
The use of AI tools by teachers is rapidly evolving, fundamentally reshaping PD practices and how educators engage with their ongoing growth. Recent literature emphasizes that AI tools provide unique affordances for personalization and immediate feedback, which are critical for effective adult learning (Kong & Yang, 2024; Younas et al., 2025). The evolution that uniquely pushes new technological affordances, simultaneously shifting pedagogical possibilities through researcher-enabled personalized, flexible and situated experiences (Younas et al., 2025). Through intelligent tutoring systems and chatbots, AI facilitates a continuous loop of professional learning. As noted by Reinders et al. (2023), these tools support teachers in designing, facilitating, and reflecting on their instructional methods, thereby enhancing self-regulation. Specifically, AI-powered chatbots such as TeacherGAIA scaffold self-assessment and metacognitive reflection by simulating reflective dialogue (Ali et al., 2023). This capacity for “on-demand” scaffolding allows teachers to engage in deep reflective practice without the immediate presence of a human mentor. As Kong and Yang (2024) and Gotavade (2024) point out, AI-enhanced SRL spaces have been shown to provide teachers with greater confidence in using AI in their practice, as well as enhanced creativity around implementing AI into their pedagogical practice.
Importantly, language teacher education research is now beginning to examine how GenAI reshapes initial and continuing professional preparation as well as acceptance and use of AI tools (e.g., Acikgul & Sad, 2026). Teacher educators anticipate substantial impacts on curriculum, pedagogy, and assessment, while also reporting limited confidence and competence to address GenAI implications systematically. Complementing this, intervention research has proposed and empirically tested training models for developing language teachers’ professional GenAI competence, showing that explicit instruction can strengthen pedagogical and critical awareness, while also identifying areas that remain difficult to develop such as guiding learners’ responsible use (e.g., Moorhouse & Kohnke, 2024). These developments indicate that language teacher education is shifting from speculative debate toward competence-based preparation, which is an important backdrop for interpreting in-service teachers’ self-regulated AI adoption.
Benefits of AI for PD
Frameworks and tools driven by AI can provide an incredibly individualized way of engaging in professional learning because of the ways in which a learning experience can be aligned with the needs, interests, and situation of the individual teacher (Kong & Yang, 2024). Individualization can occur in several ways through generative AI tools for lesson planning, differentiation for use in instruction and real-time formative feedback for teachers. These means of affordance to teachers can alleviate planning pressure and let teachers engage in learning about certain areas of growth while enhancing the time and efforts they expended on professional learning in other contexts. Moreover, because the individualized context of AI support means teachers are using content about uniquely their needs and challenges, this should lead to improved motivation and efficacy (Gotavade, 2024; Younas et al., 2025). In addition to this, AI tools that provide immediate feedback allow teachers to iteratively reshape their lesson materials and instructional modalities, and engage in a continual improvement process with the support of technology (Cope et al., 2026).
AI-driven tools are also instrumental in enabling reflective practice, which is a crucial aspect of effective teacher PD. Educators can gain access to tailored resources and assessment practices through intelligent tutoring systems and AI chatbots to support effective prompt design and metacognitive reflection on one’s pedagogical practice; (Ali et al., 2023; Kong & Yang, 2024; Younas et al., 2025). For example, AI-facilitated chatbots such as TeacherGAIA (https://teachergaia.rdc.nie.edu.sg/v2) are designed to help teachers engage in their own learning by simulating reflective dialogue and prompting critical thinking with respect to instructional decisions (Ali et al., 2023). This continuous reflective process can help teachers to change and refine their strategies based on the immediate issues faced in the classroom, thus improving the pedagogy flexibility and response. PD programs with integrated use of AI may provide experiences for conceptualizing their knowledge of AI and pedagogical uses of the technology. Besides strengthening technical capacities of teachers, the understanding will enable them to design AI-enabled courses that would allow for better student engagement and satisfaction (e.g., Kong & Yang, 2024).
One of the most crucial benefits of AI tools in the support of teachers’ SRL is personalization. AI tool s might allow for tailoring paths and content to the unique PD needs, interests and context of various teachers. This makes professionals in the education field apply their energies toward the core focus areas that hit the most meaningful and most impactful touch points within their professional practice, making growth in their professional life more efficient (Gotavade, 2024; Younas et al., 2025). In addition, AI offers a dynamically adapting learning experience, which in turn allows teachers to respond to the education landscape demands more flexibly.
Another important advantage of AI tools is feedback immediacy. In contrast with other learning processes, AI enables verification and assessment of the learning program in real time, so that skills may be acquired at a likewise pace, and critical reflection as an action may take even lesser time than before. The immediacy will afford a teacher to identify in their mind quickly and gauge the to-do list or the top identified area for improvement, with iterative course corrections made even quicker to improve on their teaching methodology. These timely feedback loops help enrich the SRL process to realize continuous, teacher-pace, and response PD (Ali et al., 2023; Younas et al., 2025).
Moreover, teacher involvement with AI tools has been noted to entail more engagement with teachers undertaking SRL. Inquisitiveness studies in AI have noted that education is more stimulating, and educators can remain only stimulated, if they are employing apparatus which is conversational, individualized, and versatile. Such engagement is critical in sustaining the momentum for lifelong learning, which forms an intrinsic motivational component of acceptance of emerging technologies in education (Ali et al., 2023; Gotavade, 2024; Kong & Yang, 2024). Reported affordances within language education include support for planning lessons, adapting materials, and quickly generating feedback, as well as potential for reflective dialogue and iterative development of ideas. However, accompanying caveats hold that these are dependent on teachers’ ability to interrogate outputs, evaluate linguistic/pedagogical appropriacy, and be transparent about authorship and decision-making (Barrot, 2023; Moorhouse & Wong, 2025).
Challenges of Using AI in PD
Alongside the benefits mentioned above, it is crucial to point out challenges or barriers. Major possible challenges are lack of technological infrastructure in some educational contexts, lack of digital literacy skills of teaching staff and restrictions on use of AI tools owing to its perceived complexity, ambiguity, and responsibility or worry about how use of AI tools might threaten formal teaching professionals (e.g., Bowen & Watson, 2026; Moorhouse & Wong, 2025). Such challenges bear on the success of AI’s implementation in self-regulated professional learning and can only be circumnavigated through PD, institutional commitment, and policies development that ensures equitable and fairness use of AI technologies (Mkhasibe & Ajani, 2024; Younas et al., 2025).
In the language-teaching literature, concerns about GenAI are frequently framed as issues of epistemic trust and pedagogical risk: outputs may be fluent yet misleading, language explanations may be overgeneralized, and feedback may appear authoritative without being grounded in sound SLA-informed principles. These risks are compounded by academic integrity and authorship concerns that are brought to the fore in language teaching contexts, where writing support tools are already in full swing. Consequently, recent ELT scholarship makes a case for effective use as inseparable from teachers’ critical AI literacy, the ability to verify claims, calibrate reliance, and articulate ethical boundaries for use (Barrot, 2023; Hockly, 2023).
The existing literature points at the SDL/SRL frameworks as providing a strong perspective of understanding teacher-led professional learning, and the potential for AI tools to scaffold planning, monitoring, reflection, and content generation. At the same time, the literature outlines that there are lingering concerns revolving around epistemic trust, over-reliance, ethical ambiguity, and unequal access/support infrastructure. However, most empirical accounts focus chiefly on institutional initiatives, preservice contexts, or evaluation of tools rather than focusing on lived, routinized, and socially mediated appropriation practices of in-service language teachers harnessing AI for SPRD. This study, therefore, fills in the gap by exploring (a) the AI technologies teachers incorporate in each phase of self-regulated learning in their SRL process (RQ1), (b) the benefits and challenges teachers encounter (RQ2), and (c) institutional and sociocultural conditions affecting the adoption (RQ3). While these three strands, SRL processes, AI affordances, and institutional contexts, are often treated in isolation, this study argues they are inextricably linked within local ELT ecologies. The empirical gap addressed here is not merely the lack of local data in Türkiye, but the lack of understanding regarding how the internal mechanics of teacher learning (RQ1) are appraised through professional judgment (RQ2) and mediated by the surrounding policy and peer environments (RQ3).
Methodology
Research Design and Rationale
This study uses a qualitative descriptive research design to explore the adoption of AI tools for SRL among in-service EFL teachers. A qualitative approach was selected to study teachers’ lived experiences and other intangible dimensions more closely. Although the term “case study” is widely used, this study is a qualitative interview case study on the phenomenon of AI adoption with the bounded group of Turkish public school EFL teachers. Considering the exploratory nature of the research, and the relatively new nature of AI Tools in teacher development, semi-structured interviews were best suited to generate delicate data that depends on the context. This study was also conducted under ethical research rules and followed ethics outlined by the Helsinki declaration and the European Code of Conduct for Research Integrity, ensuring transparency, informed consent, and protection of test subjects. All received answers were fully anonymized and processed according to data protection requirements. The study design limited the risk of harm by ensuring all interviews were conducted in a private, non-evaluative online setting where participants could speak freely without institutional oversight. Potential benefits, namely, the enhancement of teacher agency and the refinement of PD models, outweighed the minimal risks associated with professional reflection. Formal informed consent was obtained digitally prior to each interview, with participants explicitly briefed on their right to withdraw without penalty.
Research Context
English is a compulsory foreign language provided to most students at the Turkish public schools across the formal educational chain, from primary to secondary and tertiary schools. Thus, there are a lot of new national EFL teachers entering the education system every year, and there is a need for consistent and systemic support mechanisms across these teachers’ professional careers within the system from regular teacher education programs to their retirement. Although there are several in-service training opportunities available to EFL teachers, they are often externally initiated and top-down design of INSET programs, and the lecture-based nature of the formats with little room for active teacher participation (Kırkgöz, 2008; Korkmazgil, 2015). Other criticisms include a lack of expertise by lecturers and a lack of continuous guidance or follow-up support mechanisms, as noted in previous research (Uysal, 2012).
Participants and Sampling
The study involved 10 in-service English language teachers working across diverse regions and school types in Türkiye. A purposive sampling strategy was used to select participants who reported prior or current use of AI tools (e.g., ChatGPT, Grammarly, QuillBot, TTS software) in their professional learning. The inclusion criteria were: (a) currently employed as an EFL teacher in a public school; (b) having used at least one AI tool for professional purposes; and (c) willingness to discuss their SRL habits. Maximum variation was sought regarding teaching experience, school level, gender, and technological familiarity (Table 1). Though the sample size of 10 interviewees appears small, it is deemed sufficient for this exploratory qualitative inquiry to have reached data saturation, operationalized as the point of no new major thematic patterns regarding the respondents’ AI usage patterns (Creswell & Plano Clark, 2017). Maximum variation was desired in teaching experience, school level, gender, and technological familiarity. In order to foreground the participants as social actors rather than decontextualized cases, the teachers were referred to by pseudonyms in the presentation of the findings. This approach hopes to balance the need for confidentiality with interpretive richness by placing excerpts within the participants’ professional trajectory and their teaching context.
Participant Demographics.
All participants held teaching positions in public schools, with access to the internet and a range of digital tools. Some had previous exposure to technology integration through MoNE-led initiatives or participation in international teacher communities.
Data Collection
Data were collected using in-depth semi-structured interviews (Appendix A). The interview protocol was divided into 6 thematic sections corresponding to the research questions: teaching context, PD preferences, use of AI tools, benefits, challenges and concerns, and future outlook. To ensure that the questions remained clear, the protocol was piloted on two teachers who were not part of the main study. The interviews were conducted online using GoogleMeet, took 40 to 60 min, and were conducted in English. The researcher attempted to establish a rapport with the interviewee by assuring him anonymity and by presenting the interview as a joint effort to describe and understand PD and change.
Data Analysis
Interview data were thematically analyzed based on Braun and Clarke’s (2006) six-phase method. This included: (a) familiarization of data through repeated reading of data, (b) Generating initial codes such as efficiency, skepticism, prompt engineering, (c) searching for themes by grouping the codes, (d) reviewing themes in relation to the data set, (e) defining and naming themes, and (f) producing the report. Inductive coding was used as the basis of theme development was procured from the participant’s narrative. To increase analytic transparency, an audit-oriented summary of the coding-to-theme pathway (themes, illustrative codes, and exemplar extracts) is provided in Table 2.
Thematic Analysis of Teacher Interviews on AI Tool Adoption for Self-Regulated Learning.
Trustworthiness and Ethics
To ensure trustworthiness, member checking was used whereby participants were given a chance to go through their interview transcripts to ascertain accuracy. In addition, a reflexivity journal was kept by the researcher to track possible biases related to technology adoption. This study followed the ethical guidelines mandated by the host university’s Institutional Review Board (IRB). Informed consent was obtained from all the participants and pseudonyms were used to conceal their identity. Interpretive quality was enhanced through (a) an explicit audit trail of coding decisions, code revisions and theme definitions over successive analytic cycles; (b) analytic memoing including interpretive summaries, disconfirming evidence, and connections to SRL constructs; and (c) conscious searches for divergent or deviant cases to advance the boundaries of emerging themes and avoid overly homogeneous interpretations. Where feasible, preliminary themes were discussed with a qualitative methods colleague to test coherence and challenge researcher assumptions. These procedures align with reflexive thematic analysis as they value transparency, reflexivity, and interpretation over a mechanical reliability index.
Findings
The findings are presented in relation to the three research questions, highlighting three interconnected dimensions of AI use in paritcipants’ self-regulated professional learning. Regarding RQ1, participants reported using AI tools across multiple phases of self-regulated learning (SRL), particularly performance (e.g., conceptual clarification and resource use) and reflection (e.g., post-lesson evaluation), with forethought-oriented planning mentioned less consistently. Across the dataset, SRL-oriented uses were described most frequently in the performance phase (n = 9/10) and reflection phase (n = 8/10), while explicit forethought-oriented planning was described by 5 participants. For RQ2, participants emphasized perceived benefits, especially efficiency/time-saving, personalization, and metacognitive support, while also reporting challenges associated with epistemic trust, ethical ambiguity, and uneven digital literacy. Efficiency/time-saving was mentioned by 9 participants and personalization by 8; metacognitive/reflective benefits by 6 and confidence-related benefits by 5 participants. Challenges most concerned accuracy and the need to cross-check (n = 6) and ethical concerns related to “cheating,” authorship, or originality (n = 6). For RQ3, participants consistently reported limited institutional recognition or guidance regarding AI-enabled professional learning (n = 8/10) and described relying on informal peer networks and online communities to discover tools and practices (n = 8). To enhance transparency in this qualitative inquiry, the manuscript provides the number of participants (n) who articulated each recurring perception or practice; these counts indicate recurrence within this dataset and are not intended as statistical generalization beyond the study sample.
RQ1: AI Scaffolding of Forethought, Performance, and Reflection
This section defines the SRL constructs used to interpret how participants describe planning, strategy use, monitoring, and reflection with AI tools, which directly frames RQ1. The analysis indicated that participants’ engagement with AI tools for SRL-informed professional learning was multifaceted and varied in intensity across participants. A self-regulated professional identity, describing themselves as independent learners who seek resources beyond formal in-service training, was articulated by 6 participants. Relatedly, 7 participants described initiating AI use through informal exploration (e.g., curiosity-driven experimentation) or peer/online exposure rather than through institutionally organized training. Overall, participants reported using AI tools across SRL phases, including forethought (planning), performance (strategy use and conceptual scaffolding), and self-reflection, although the specific emphasis and sequencing of these phases differed by teacher and context.
With reference to Forethought and Planning, Some participants (n = 4) wrote that they saw themselves as independent and following a self-regulated path in professional learning. For instance, Fatma remarked, “I use AI to professionally chart my goals for the semester, such as producing CLIL material.” This is an example of using AI in the forethought phase to organize a learning trajectory. As for Performance and Usage, participants used a plethora of AI-supported SRL practices which went beyond the conventional PD sessions. Many (n = 7) used ChatGPT to “clarify educational terminology, summarize academic papers, generate alternative lesson ideas for review, or engage in pedagogical role-play dialogues.” Meltem said, “When I read an article, I ask ChatGPT to help me with complex sentences or to bring it to my teaching context, which is conceptual scaffolding for me.” For Reflection, AI tools were also instrumental in reflective practice. Nuray’s description is “I put lesson transcripts into ChatGPT and ask for suggestions on how to improve it – a feedback loop of my own making.”
These instances display their SRL readiness which is showcased through intrinsic motivation, and high context autonomy over their learning. Based on the way participants articulated their reflections, it is evident that their SRL is not a series of episodic events but rather a product of accumulated experience resulting from a lifetime of habitually engaging in established processes of self-inquiry and reflection with others. Their readiness to consider an AI tool in their self-learning process means that the AI tools meet their rising expectations of professional learning in rapidly changing educational contexts.
Initial exposure to AI tools varied across participants but was predominantly informal. Most participants (n = 8) reported learning about tools like ChatGPT, Grammarly, or QuillBot through social media platforms (e.g., X (previously Twitter); Instagram), YouTube, or peer conversations. Curiosity often sparked the first interaction, but sustained use followed from recognizing the tool’s value in improving one’s own knowledge and practice. These entry points demonstrate that SRL using AI begins with awareness and experimentation, often shaped by the teacher’s immediate learning goals and contextual opportunities. Participants described a range of AI-supported SRL practices that extended beyond traditional PD activities. Many (n = 6) used ChatGPT to clarify educational terminology, summarize academic papers, generate alternative lesson ideas for review, or role-play pedagogical dialogues. Others (n = 4) employed Grammarly or QuillBot to revise their reflective writing and teaching portfolios. Use frequency varied by the teacher’s workload and SRL habits and the participants indicated they used AI tools occasionally, up to daily. It was also common for participants to use different tools for different purposes. For example, they would watch a tutorial on a particular concept, and then they would ask ChatGPT to elaborate on the concept, and then revise their response in Grammarly. This multimodal, tool-combining behaviors indicates that AI is becoming a distributed cognitive partner within their SRL ecosystem.
RQ2: Perceived Benefits and Challenges
Participants reported several perceived benefits of AI tools for their SRL. The most frequently mentioned benefits were (a) improved efficiency and time-saving (n = 9), (b) personalization of learning resources to match individual needs and proficiency (n = 8), and (c) enhanced reflective or metacognitive engagement with professional learning (n = 6). Emre highlighted this metacognitive benefit: “AI helps me think about how I learn, not just what I do. I now plan my learning more intentionally.” In addition, 5 participants reported increased confidence in professional communication (e.g., reflections, proposals, applications), and described using AI to test ideas in a low-risk environment (e.g., simulating pedagogical scenarios or receiving “mentor-like” feedback). Taken together, these accounts suggest that AI was perceived not only as an informational resource but also as a metacognitive scaffold supporting goal setting, strategy use, and self-evaluation.
In addition to perceived benefits, participants outlined challenges that constrained sustained or confident use of AI. Limitations in digital literacy and uncertainty about harnessing tools were reported by 7 participants, while the accuracy of content and the need to cross-check when using AI was reported by 6 participants. Barış specified a critical evaluation skill: “I cannot always judge the truth of the content generated by AI. I need to verify it first, which wastes time.” Time constraints and inability to maintain consistent routines (e.g., during exam periods or when there is a lot of administrative work) were each described by 4 participants. Ethical concerns (e.g., authorship, originality, or that AI support is akin to “cheating”) were mentioned by 6 participants. Also, Kerem shared an ethical dilemma: “It took a while before I was mentally comfortable with the idea that I wouldn’t be “cheating” by using AI as a tool to aid my learning process.” These “deviant” or critical voices show that adoption is not an all embraced solution but rather a problem filled with hesitation. These accounts show that participants’ engagement was not wholly positive but rather marked by tensions between the utility of the tools and concerns related to trust, ethics, and sustainable practice.
RQ3: Institutional and Sociocultural Influences
The section describes institutional policy, recognition, and informal networks of peers as mediators for technology uptake to provide a lens for RQ3. The ability of participants to adopt AI into their SRL practices was also influenced by the wider institutional discourse. However, most participants (n = 7) identified their schools being neutral or having no interest in the use of AI in developing teacher capacity. Umut, for instance, commented: “There is no one who watches over my education, so I have total freedom, but I would like there to be recommendations.”
In instances where no explicit formal policy or advice was available, participants relied on informal communities such as Telegram groups or online forums where peers shared prompt ideas, possible use cases, and guidance on how to navigate difficulties. Lack of acknowledgment or incentive from MoNE was a repeated theme, as participants (n = 3) described the use of AI-supported learning as routinely invisible in official PD modes. However, peer support and community learning acted as a compensator for the institutional context and strengthened that SRL is socially embedded. Participants (n = 8) indicated emphatically the need for clear PD resources that address the needs of their professional learning as participants. While the participants (n = 5) valued their self-regulated exploration of the tools, they asked for more formalized guidance in the form of learning pathways (e.g., curated video playlists, an interactive course, or an AI-powered PD module with progress tracking). In addition to generating content, some participants (n = 5) also imagined using AI in a future learning endeavor. All the participants indicated the desire to have AI use as a coaching agent for suggesting goals, tracking the progress of players, and giving feedback about the PD tasks. This aspirational view marks a shift from the situational use of such tools to a shift of integration via AI to the SRL practices.
Discussion
This study explored how in-service EFL teachers adopt AI tools for SRL. In direct relation to RQ1–RQ3, the discussion below interprets the findings by (a) linking each interpretive claim to the empirical patterns reported in the findings (themes, recurring practices, and illustrative excerpts), and (b) indicating recurrence through participant counts (n).
The findings demonstrate that the three research questions are functionally interdependent. The “metacognitive scaffolding” identified in RQ1 is directly contingent upon the resolution of “epistemic trust” issues identified in RQ2. Furthermore, the transition from “informal exploration” to “sustained practice” is governed by the institutional “neutrality” or peer-based “horizontal networks” explored in RQ3. This interaction confirms that AI-supported SRL is not a purely cognitive event, but a socially and institutionally situated practice. The findings also suggest that AI might act as a “metacognitive scaffold” (Zimmerman, 2000), supporting teachers not just in content creation but in the regulation of their professional growth. This claim is grounded in participants’ reported use of AI for planning, monitoring, and reflective evaluation, specifically, accounts of goal-setting and planning activities (forethought; n = 5), conceptual clarification and strategy enactment (performance; n = 9) and post-lesson review or improvement planning (reflection; n = 8). This interpretation aligns with emerging language-education syntheses showing that GenAI is increasingly discussed not only as a content generator but also as a dialogic resource that can prompt planning, monitoring, and reflection, while simultaneously requiring participants to exercise evaluative judgment to manage trust and accuracy (Li et al., 2024).
The findings also indicate a multifaceted understanding of the manner in which English language teachers in Türkiye assimilate a contemporary AI tool in support of their SRL, with participants making reference to motivation, autonomy and digital affordances as a catalyst for informing contemporary teacher learning and development. That “complex picture” empirically manifests in the co-occurrence of (a) perceived benefits such as efficiency/personalization (n = 9) and reflective prompting (n = 6) and (b) constraints such as accuracy concern and the need to cross-check (n = 6) and ethical ambiguity (n = 6) which shows that adoption was shaped by enablement and frictions rather than unilaterally positive experiences. Teacher autonomy with AI is representative of the shift toward informal, collective, and self-regulated professional learning that is consistent with adult learning theory (Merriam & Bierema, 2014) and contemporary models of SRL (Panadero, 2017; Zimmerman, 2001). This pedagogical shift was apparent in the way participants described starting and maintaining their learning through informal (e.g., self-initiated exploration and peer/online discovery) rather than formally organized (i.e., continued professional development [CPD] arranged by institutions) means (n = 7), supporting the interpretation that AI use was integrated in larger schemes of teacher-led learning.
The focus on teacher identities and SRL orientation exemplifies the critical roles agency plays in technology adoption. Participants articulated identity as autonomous learners who often find their own routes to learning without institutional directives. In this dataset, an explicitly self-regulated professional identity was articulated by 6 participants, and this recurrence provides the empirical basis for interpreting “agency” as a salient mediating factor rather than a speculative assumption. Such types of dispositions posit that SRL is not a reactionary or restorative process but rather it is a process of strategic learning in the continuum in which AI tools increasingly become cognitive, metacognitive, and even social-emotional learning scaffolds (Hadwin et al., 2018). To avoid overclaiming, “social-emotional supports” are interpreted as perceived functions (“confidence, reassurance, professional validation, etc.,” reported by the participant) rather than validated psychological outcomes; affect-related benefits yielded 5 participants mentioning them, while foregrounding caution and uncertainty, stipulating meaningful heterogeneity in experience (n = 6). The context is notable in a climate in which organized CPD) often lags behind technological change but implies recognition and validation of teacher initiated SRL as a component of PD, growth and agency. This implication was primed by participants’ recurring narration of limited institutional guidance/recognition (n = 7) against a backdrop of reliance on self-initiated learning practices indicating that teacher-initiated SRL acted as a redress in the local professional learning ecologies.
The findings are in accordance with Zimmerman’s SRL phases. During the forethought phase, the participants used AI to establish goals. In the performance phase, tools such as ChatGPT served as “more knowledgeable others,” scaffolding complex concepts (Vygotsky, 1978). In the reflection phase, AI provided the immediate feedback that traditional PD often lacks. Crucially, the empirical basis for this alignment lies in (a) recurrent forethought-oriented planning accounts (n = 5), (b) repeated descriptions of performance-phase conceptual scaffolding and resource mediation (n = 9), and (c) participants’ accounts of reflective use (n = 8). Where participants did not describe a given phase, this absence is treated as an analytically informative boundary condition rather than forced into the SRL frame.
The importance of networked and peer-based discovery has echoes of situated learning theory and a community of practice lens (Lave & Wenger, 1991), where institutional context is key in accessing the potential of innovation. Participants described learning about AI tools through Twitter, Telegram groups, and YouTube, which indicates that while CPD has emerged to support formalized approaches to professional digital literacy learning and teaching, informal, socially situated mechanisms, are still the primary methods of developing digital literacy. This conclusion is empirically supported by the high recurrence of informal discovery routes (n = 8) including social media and peer-network channels, contrasted with sparse mention of institutional training or guidance (n = 7 reporting its absence). This dynamic upholds the notion that the meaningful mixing of AI does not start in the training rooms or the policy documents but dialogic spaces where participants share experiences and try out tools non-theoretically and contextually (Trust et al., 2016). The claim is reframed as a pattern identified from the dataset to cater to the reviewer’s request for claim–evidence coherence: Informal dialogic spaces were delineated as the primary venues for learning and experimentation by 8 participants, while a minority (n = 4) described more guarded or limited approaches to learning there, with reference to cumulative workload, uncertainty, as well as inaccurate/ethical content prompting such an approach.
The patterns of AI use for SRL uncovered in this study demonstrate a sophisticated and layered engagement with digital tools. To make this claim evidentially explicit, we define “layered engagement” operationally as multi-step and/or multi-tool use (e.g., combining different tools for ideation, clarification, drafting, revision, and reflection) and the recursive movement between these steps over time. In our findings, multi-tool engagement was evidenced qualitatively through accounts of using different tools for different purposes (e.g., ChatGPT for clarification and idea development and Grammarly/QuillBot for revision), including descriptions of tool-combining sequences (tutorial → ChatGPT → Grammarly). Because the interviews did not systematically elicit frequency for multi-tool sequencing, we treat this as a recurrent qualitative pattern rather than a quantified category.
Rather than using AI in an instrumental or piecemeal way, participants also described blended, recursive processes of first exploring an idea with one tool, then revising their understanding by consulting another tool, and finally reflecting on the implications for practice by a third tool.
This interpretation is based on participants’ descriptions of distributed tool use and iterative refinement and is qualified by countervailing evidence in which constraints limited sustained engagement (accuracy/verification concerns: n = 6; time constraints: n = 4; ethical ambiguity: n = 6). Such practices align with the cyclical and dynamic nature of SRL models, which stress the intertwined nature of goal-setting, strategic action, and self-reflective phases (Panadero, 2017; Zimmerman, 2008). Notably, AI tools seem to scaffold all three phases of the said cycle, implying that their role is not limited to content delivery, but pedagogical reasoning too. “Deeper pedagogical reasoning” was operationalized to be supported when participants explicitly indicated that they used AI to interrogate pedagogical choices, make better instructional choices, or assess classroom implications (performance-phase conceptual scaffolding/resource mediation: n = 9; reflection-phase use: n = 8).
The clearly perceived advantages of AI-supported self-regulated learning included time-saving, personalized learning products, and amplified learning of the reflective practice, mirroring similar research on AI in the education sector (Holmes et al., 2019; Kovalenko & Baranivska, 2024). Foregrounding these benefits is empirical in the sense that time-saving/efficiency is stated by 9 participants, personalization by 8 participants, and reflective/metacognitive by 6 participants in the dataset. That this study has focused on the affective dimensions of the perceived benefits is worthy of note. Participants communicated that they felt more confident, intellectually aroused, and professionally validated through their interaction with AI, especially in situations where there was little or no feedback from the institutions. This affective dimension is also present with explicit reporting of increased confidence and/or professional validation (n = 5), usually juxtaposed to school neutrality/uninterest (institutional affective dimension: school neutrality/uninterest: n = 7; MoNE invisibility in official PD: n = 3). At the same time, participants did not report purely affective gains, as shown by constraints such as digital literacy uncertainty (n = 7) and trust and verification demands (n = 6), which point to the contingencies of AI-mediated affect and motivation. The willingness to try, the development of new knowledge, developing greater independence in their thinking, feeling more confident, and feeling validated could define SRL not as purely cognitive, but socio-cognitive and socio-emotional, SRL is deeply emotional and motivational (Boekaerts, 2011). To maintain interpretive discipline, we frame this as: the findings suggest SRL in AI-supported professional learning is experienced by many participants as simultaneously cognitive and motivational/affective (n = 5), rather than asserting that SRL is redefined universally by AI.
The study also identified significant barriers to the continued use of AI as a learning tool. Major barriers to developing a sustainable pattern of use of AI include digital literacy difficulties, trusting what content takes from online, time constraints, and the uncertain ethical aspects of the AI tool’s use. Empirically, digital literacy/know-how challenges were reported by 7 participants, time constraints by 4 participants, accuracy/trust, and the need to verify outputs by 6 participants and ethical ambiguity (e.g., authorship) by 6 participants. These barriers reflect broader concerns from the field around AI systems trustworthiness and transparency (Zawacki-Richter et al., 2019). From an SRL perspective, the barriers appear more concerning as they may indicate the potential for breakdowns in the monitoring or evaluation parts of the cycle of learning-critical phases for regulating one’s learning journey. Specifically, recurrent cross-checking demands may shift cognitive load from learning to verification, potentially constraining sustained engagement for participants with limited time; this interpretation is grounded in participants’ accounts that verification slows down use and reduces perceived efficiency (n = 6).
The institutional/policy context was the most explicit mediating factor in this study. Despite participants’ testimonies to an increased use of AI for personal and PD-related purposes, participants overwhelmingly recounted a lack of support. To avoid ambiguity, recurrence is specified, and lack of support/guidance was reported for the school level (7) and visibility/acknowledgement of MoNE in official PD (3). This discrepancy is a sign of a systemic failure to scaffold personal and professional innovation. In the absence of formal PD, teachers rely on what Trust et al. (2016) call “horizontal” networks. This study validates that AI adoption in Türkiye is a teacher-led movement rather than a policy-driven one, with the combination of informal discovery routes (n = 8) and the little facilitation/recognition at institutional level (school level: n = 7, MoNE invisibility theme: n = 3). Recent language teacher education research similarly reports that teacher educators anticipate major GenAI impacts but often feel underprepared, reinforcing the need for explicit competence development rather than leaving uptake to informal trial-and-error alone (Moorhouse & Kohnke, 2024).
Though participants indicated rising use of AI for personal and PD purposes, they overwhelmingly recounted a lack of support or facilitation, recognition, or formal guidelines from schools or the influential authority of national authorities such as MoNE. Again, to strengthen the claim–evidence link, we report recurrence: 7 participants described lack of school-level guidance/recognition, and 3 participants explicitly referenced the absence of national-level guidance/policy clarity (where applicable in the interviews). Given the barriers identified (accuracy verification, ethical uncertainty, uneven digital literacy), a policy response would ideally include (a) guidance for responsible use (authorship, transparency, ethical boundaries), (b) structured opportunities for teachers to develop evaluative judgment and verification practices, and (c) community-based professional learning designs that recognize and leverage existing peer networks rather than attempting to replace them. Participants’ visions for AI-supported learning point to a realistic yet optimistic perspective. Participants envision AI not just as a provider of information, but a personal learning assistant; a kind of feedback; to monitor progress, recommend personalized learning plans. In this dataset, such future-oriented expectations were articulated by 5 participants and frequently positioned as addressing perceived gaps in feedback and personalized support. The usefulness of educators responding to unwelcome new learning environments appears almost as an expectation for adaptive learning to meet the needs of language teachers, and noteworthy as a direction toward a personalized PD future that includes AI, in education (Aguilar-Cruz & Salas-Pilco, 2025). However, educators will require not only the tools, but pedagogical frameworks, raise professional ethical considerations, and communities of practice that can mediate the ways to engage their tools meaningfully. This implication directly follows from the empirical tensions documented in the findings: participants’ desire for personalization (n = 8) and feedback (10/10) co-existed with concerns about trustworthiness (n = 6) and ethics (n = 6), indicating that “meaningful engagement” depends on frameworks that support verification, transparency, and reflective use rather than purely instrumental adoption. These concerns mirror persistent cautions in ELT-focused scholarship that fluent outputs can mask inaccuracies and that writing-support affordances intensify authorship and integrity dilemmas, making verification practices and ethical boundary-setting central professional skills, in line with the discussions by Barrot (2023).
This line of research underscores the role of self-regulation, motivation, and mediating context on the uptake of AI tools among language teachers for professional learning. With the rise of the use of AI in education into the future, it will be critical to explore the role of AI not only as an educational instrument, but also as a partner in teachers’ learning. To aid in this relationship, the policies in the institutions have to be formulated not in a reactive manner but in a way to create collaborative models of PD, which takes into account the indeterminacy, contextuality, and autonomous nature of 21st century professional learning for teachers.
Limitations and Further Research
This study is faced with several limitations. On the one hand, the sample size of 10 might limit the generalizability of the findings to the wider population of Turkish teachers. Second, the data was self-reported, which introduces social desirability bias-teachers might have over-reported their “successful” use of AI and under-reported struggles. Finally, the rapid evolution of AI tools means that the particular tools mentioned (e.g., ChatGPT) may change, even though the underlying patterns of self-regulation are likely to be still relevant. Future studies need to look at the longitudinal effects of AI on teachers’ efficacy. Using quantitative studies, a larger population could be surveyed to validate the relationships between digital literacy and AI adoption identified in this paper. In addition, observational studies could verify the AI use self-reporting and the classroom application.
Implications and Conclusion
This study adds to the literature of AI adoption in language education by focusing on teachers’ realities in Turkish public-school language classroom contexts as well as incorporating their experience and perception with the technology. The thematic analysis shows that although the participants possess a high level of motivation with adopting SPRD AI tools and for innovating strategies inside the classroom, there are several factors identified as influencing individual, social, and institutional adoption processes. The findings highlight the role of informal learning networks and digital agency as critical components social influence and facilitating conditions in the initial acceptance and continued use of AI tools. However, issues related to digital literacy, ethics, and lack of institutional support present great challenges toward achieving equitable and effective integration. Policymakers and educational leaders should seek to design holistic CPD programs that practically train teachers on AI integration, issue formal endorsements and codes of conduct regarding the ethics of AI use, and create an enabling environment that fosters innovation while safeguarding educational values. Finally, participants view AI as a foreseeable change in future language teaching and a change that needs to be prepared for accordingly using a blended approach to professionalization. These are relevant training of teachers, at the nation level and also in international discourse on how best to harness the potential of AI in a pedagogically meaningful and ethically sensitive manner to the teaching and learning of languages. For the policymakers, this study means that they should go beyond “policing” AI to “scaffolding” the use of AI in PD frameworks. For the teacher educators, the need for “AI Literacy” is evidently called for within the INSET programs, with a focus on prompt engineering being a reflective skill.
Footnotes
Appendix A
Acknowledgements
The manuscript has been revised and linguistically refined with the assistance of digital tools, including Grammarly and Google’s Gemini, to enhance clarity and coherence. The authors assume full responsibility for the content, interpretations, and any remaining errors in the final version of the manuscript. The second author would like to express his sincere gratitude to the Polish National Agency for Academic Exchange (NAWA) for their support under the Ulam NAWA Program.
Ethical Considerations
The research was conducted in strict alignment with established ethical principles for studies involving human participants, particularly those outlined in the Declaration of Helsinki. All methods employed were non-invasive and carried no risk of physical harm, psychological distress, or social disadvantage.
Consent to Participate
Participation was completely voluntary, and individuals retained the right to discontinue their involvement at any stage without facing any repercussions. Prior to data collection, informed consent was secured through a comprehensive consent form that clearly explained the study’s objectives, procedures, possible risks and benefits, and measures for safeguarding data. To ensure confidentiality, all responses were anonymized, and any personally identifiable details were removed during both analysis and reporting.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors received financial support from Adam Mickiewicz University in Poznan, Poland for the open access publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
Data is available upon request.