2. The paradox of GenAI-enhanced task design

At first glance, GenAI appears to be an ideal companion to TBLT. One of the most persistent difficulties in implementing TBLT is the substantial labour involved in designing tasks that are meaningful, level-appropriate, and contextually authentic (Bryfonski et al., Reference Bryfonski, Ziegler and Montee2025; Huang et al., Reference Huang, Wu and Dou2024; Ouyang & Zhang, Reference Ouyang and Zhang2024). Teachers frequently report that they lack preparation time, access to needs-analysis tools, or sufficient materials to support rich task cycles (Aljohani, Reference Aljohani2025; Ellis, Reference Ellis2017; Long, Reference Long2016). GenAI seems to offer a ready solution. It can instantly generate communicative tasks, adapt them to different proficiency levels, and even provide sample performances within seconds. Activities that once required teachers to invest hours, such as designing role-plays or structuring problem-solving tasks, can now be automated, complete with sequenced pre-task and post-task activities.

This technological efficiency can easily lead educators to believe that GenAI finally ‘solves’ the labour-intensive demands of TBLT. However, the apparent convenience hides a deeper paradox. Although GenAI can rapidly generate communicative activities, it is unable to judge whether these tasks genuinely align with curricular aims, cultural norms, or the developmental needs of particular learners. For example, a GenAI-generated task asking students to ‘plan a weekend trip abroad’ might seem engaging, but it may ignore the financial or cultural realities of learners who have never travelled outside their home city. GenAI also lacks awareness of interpersonal dynamics, learners’ prior experiences, and the subtle misunderstandings that often stimulate meaningful negotiation of meaning. It cannot anticipate, for instance, that a shy learner may struggle with a debate task unless it is carefully scaffolded, or that a group with conflicting personalities may require teacher mediation to collaborate effectively. Nor can it discern whether a task resonates with the lived realities of a specific classroom or simply appears coherent in a generic, decontextualised way. In short, GenAI can generate tasks, but it cannot determine what counts as pedagogically meaningful.

Teachers, therefore, shift from being producers of tasks to critically literate evaluators of them. Their work becomes curatorial rather than purely generative. This transformation demands a new form of expertise: the capacity to scrutinize GenAI-generated content, assess its pedagogical appropriateness, and adapt it with sensitivity to local context. For example, a GenAI-generated ‘job interview’ task may require the teacher to adjust culturally inappropriate questions, remove unrealistic vocabulary demands, or modify the scenario to reflect learners’ actual career aspirations. Similarly, a GenAI-produced role-play about ‘booking a hotel room abroad’ might need to be reshaped for learners who have limited travel experience or for communities where such contexts are not culturally relevant.

Far from reducing teacher workload, responsible integration of GenAI expands the cognitive and ethical demands of task design. Teachers must not only ensure linguistic suitability but also anticipate how learners will emotionally and socially engage with GenAI-suggested activities. They must modify tasks to avoid stereotypes, align them with curricular goals, and ensure that GenAI-generated materials support, not distort, the principles of TBLT.

3. GenAI as a new participant in the task cycle

Perhaps the most transformative shift introduced by GenAI is that it effectively becomes a new ‘participant’ in the task cycle. Learners increasingly turn to GenAI at different stages of task completion (Huang et al., Reference Huang, Wu and Dou2024). For instance, they may ask for ideas, refine vocabulary choices, or test sentence structures, often receiving immediate and polished linguistic suggestions. For teachers committed to TBLT’s emphasis on meaning-focused communication, this creates a pedagogical dilemma: does GenAI provide useful scaffolding that supports communicative development, or does it function as a shortcut that undermines the very cognitive processes TBLT is designed to cultivate?

GenAI’s presence alters the cognitive work that tasks are designed to elicit. Traditional TBLT values the productive struggle involved in formulating meaning, searching for linguistic resources, negotiating ideas, and working through communicative difficulties (Ziegler, Reference Ziegler2016). These processes are central to interlanguage development. However, GenAI can instantly generate well-formed utterances, reorganize discourse structures, or supply culturally appropriate expressions. Instead of encouraging learners to draw on their developing linguistic systems, GenAI may entice them to outsource that effort to an external source.

This tension becomes particularly apparent when learners over-rely on GenAI for pre-task planning. For instance, a learner who asks a GenAI to ‘draft a discussion response for me’ bypasses the opportunity to shape their own ideas, test the limits of their linguistic repertoire, or negotiate meaning with peers. Similarly, during task performance, learners may be tempted to consult GenAI on their phones or laptops, not to seek clarification but to generate entire utterances. What was once a collaborative space for interaction risks becoming a space mediated by a silent, invisible third participant whose linguistic proficiency eclipses that of everyone in the room.

In this context, teachers are confronted with some difficult questions. Should GenAI be restricted during certain phases of the task in order to preserve opportunities for productive cognitive struggle, and if so, how should such restrictions be implemented? Should learners be allowed to use GenAI during pre-task planning but not during task performance? These questions do not lend themselves to simple answers because different learners may benefit from different forms of GenAI support at different moments in their development.

What is clear, however, is that teachers must assume new pedagogical roles. They become GenAI-use mentors who help learners differentiate between support that enhances learning and assistance that short-circuits it. This requires explicit conversations about metacognition, learner autonomy, and responsible GenAI use. Teachers must establish new norms and expectations for how, when, and why GenAI should be used during the task cycle. This involves setting guidelines that safeguard the integrity of learner-to-learner interaction while still allowing students to benefit from novel forms of scaffolding. It may also require teachers to design tasks that incorporate GenAI use deliberately, such as asking learners to critique GenAI-generated responses or compare their own ideas with those produced by GenAI, thereby transforming the technology from a shortcut into an object of analysis.

In this sense, teachers are no longer merely designers of tasks. They become architects of GenAI-mediated task ecologies, shaping environments in which human and machine interactions coexist in ways that remain faithful to the principles of TBLT. Their role is not to exclude GenAI from the classroom, but to integrate it ethically and purposefully so that it supports, rather than supplants, the cognitive and social work of language learning.

4. The emotional and ethical labour of teaching cannot be automated

An assumption in public discourse is that GenAI may eventually replace teachers (Chan & Tsi, Reference Chan and Tsi2024). Nevertheless, language learning has never been solely a matter of linguistic processing. It has always been an experience that is deeply human, emotional, and relational (Sato et al., Reference Sato, Salas, Freeborn and Tajabadi2025). TBLT, in particular, relies on group dynamics, shared experiences, and authentic communication (Ouyang & Zhang, Reference Ouyang and Zhang2024), all of which depend heavily on the emotional labour teachers perform.

Teachers support shy or anxious learners, encourage risk-taking, mediate conflicts, cultivate trust, and facilitate collaborative meaning-making. These interpersonal efforts are not peripheral; they create the very conditions in which communication can flourish. A chatbot may generate polite encouragement or neutral feedback, but it cannot perceive the discomfort of a learner who hesitates before speaking, nor can it adjust its response to soothe tension after a misunderstanding. It cannot notice when a student withdraws because of group pressure or respond to the subtle emotional cues that shape participation. Learners do not simply need accurate input, they need affirmation, recognition, and a sense of belonging (Costa & Nazari, Reference Costa and Nazari2024). These are uniquely human responsibilities, and they remain central even in GenAI-rich learning environments.

Rather than replacing teachers’ emotional labour, GenAI magnifies its significance by revealing what machines cannot do. When GenAI becomes more capable of handling surface-level linguistic tasks, the distinctively human aspects of teaching (e.g., empathy, intuition, relational judgement) stand out more clearly. Teachers do not disappear; instead, their work becomes more visible.

The ethical dimension of GenAI integration is equally critical. GenAI systems are trained on vast corpora that inevitably contain cultural, ideological, and linguistic biases (Alaqlobi et al., Reference Alaqlobi, Alduais, Qasem and Alasmari2024; Fleckenstein et al., Reference Fleckenstein, Meyer, Jansen, Keller, Köller and Möller2024; Hu, Reference Hu2024). When teachers incorporate GenAI into TBLT, they must anticipate how these biases may surface in task content or feedback. For instance, a GenAI-generated dialogue about workplace communication may subtly reproduce gendered assumptions about authority roles. A task scenario set in a ‘typical’ restaurant interaction might privilege Western norms of politeness and overlook how requests and refusals are negotiated differently in learners’ cultural contexts. Even seemingly neutral feedback may implicitly favour certain varieties of English while devaluing others (Kang & Hirschi, Reference Kang and Hirschi2025).

Such issues place significant ethical responsibility on teachers. They must protect learners’ identities, challenge biased representations, and ensure that GenAI-assisted materials support inclusive, culturally responsive pedagogy rather than homogenising or marginalising perspectives. This is not merely a matter of editing GenAI-generated texts. It requires teachers to cultivate critical GenAI literacy: the ability to recognize when GenAI output reflects stereotypical assumptions, when it erases cultural nuance, or when it reinforces unequal power relations embedded in its training data.

In this sense, teachers become ethical guardians of communicative practice. They navigate how GenAI influences learners’ linguistic development, cultural understanding, and sense of self. Their task is not to shield learners from GenAI, but to ensure that, when GenAI is present, it does not distort the values (e.g., equity, respect, authenticity) that underpin meaningful communication. The rise of GenAI does not diminish teachers’ roles; it elevates the ethical and emotional dimensions of their work to the centre of pedagogical practice.

5. GenAI and the reconfiguration of teacher professional identity

Beyond reshaping classroom practices, GenAI also compels teachers to rethink their professional identities within TBLT. Traditionally, language teachers positioned themselves as experts in linguistic knowledge, instructional design, and classroom management (Dogan et al., Reference Dogan, Nalbantoglu, Celik and Agacli Dogan2025; Oved & Alt, Reference Oved and Alt2025). Within a TBLT framework, this expertise was expressed through designing meaningful tasks, managing interaction during task performance, and guiding learners through cycles of planning, negotiation, and reflection. However, as GenAI becomes capable of generating explanations, modelling interaction, and producing instructional materials with remarkable speed, teachers may begin to question which dimensions of their expertise remain uniquely their own. This shift is not merely technical but pedagogical, and even existential, prompting teachers to reconsider the deeper foundations of their professional role in task-based teaching (Zhang et al., Reference Zhang, Lai and Gu2025).

For some teachers, GenAI’s linguistic fluency and task-generating efficiency can create a sense of displacement. When a machine effortlessly produces sample task responses, simulates role-play dialogue, or explains grammar more concisely than a teacher might, it is understandable that teachers may feel their authority eroding. Such reactions reveal a narrow conception of teacher professionalism that equates expertise with the transmission of linguistic forms. TBLT has long emphasised that teachers are not mere deliverers of language content; they are designers of learning experiences and facilitators of meaningful communication. The emergence of GenAI therefore pushes teaching towards a more expansive, human-centred definition of expertise, one already embedded in TBLT principles.

In this reconfigured landscape, teachers are valued not primarily for the language knowledge they possess but for the meaning-making opportunities they orchestrate through tasks. Their expertise lies in interpreting learners’ needs, selecting task types that support developmental readiness, and shaping interactional conditions that GenAI cannot perceive. Teachers must help learners navigate a task cycle where the abundance of GenAI-generated language requires discernment, not memorization. In a world where GenAI can instantly produce a polished task performance, teachers guide learners to understand how using language during a task shapes identity, builds relationships, and constructs belonging dimensions central to TBLT’s commitment to authentic communication.

Moreover, teachers must acquire competencies that previous generations did not require (Du et al., Reference Du, Sun, Jiang, Islam and Gu2024), particularly when integrating GenAI into task cycles. Critical GenAI literacy is becoming essential to TBLT professionalism: understanding how GenAI works, recognising its limitations, identifying bias in GenAI-generated task prompts or feedback, and guiding learners in using GenAI responsibly during pre-task planning or post-task reflection. Teachers who once focused on selecting or adapting textbook tasks must now evaluate GenAI-produced activities for pedagogical validity, design tasks that incorporate GenAI meaningfully, and safeguard learners against the uncritical use of machine-generated language. The teacher’s role thus broadens, moving from ‘task designer’ to ‘GenAI mediator’, ‘interaction architect’, and ‘ethical guide’ within the task cycle.

This shift in professional identity also reinforces the role of teachers as lifelong learners, a stance already inherent in TBLT’s reflective orientation. The rapid evolution of GenAI demands ongoing experimentation, adaptation, and critical pedagogical reflection. Teachers must learn not only how to use GenAI tools but how to understand their consequences for interaction, learner agency, and task sequencing. In this way, GenAI becomes a catalyst for sustained professional growth rather than a threat to professional relevance. It asks teachers to draw on qualities machines cannot access, such as intuition, relational sensitivity, and moral imagination, while developing new skills that enable them to steward GenAI use responsibly in task-based environments.

Ultimately, GenAI does not diminish the identity of the TBLT teacher; it prompts a reimagining of it. It shifts the centre of TBLT professionalism away from information delivery and towards meaning-making, judgement, interactional orchestration, and ethical leadership. In that sense, GenAI does not replace teachers in TBLT; it demands a more intellectually, relationally, and ethically expansive form of task-based teaching.