2.3.1 Image Synthesis

GenAI algorithms can generate unique works of art, such as digital paintings, sculptures, and graphic designs. These tools are handy for creating visuals to enhance lesson plans, projects, and presentations. Examples include

• DALL-E: Generates unique images based on text prompts, helping educators create visuals tailored to specific lesson themes. Midjourney: Produces high-quality, customisable images ideal for creative projects and thematic lessons.

• Adobe Firefly: Allows teachers to generate creative, culturally responsive images using prompts in multiple languages.

2.3.2 Video Generation

GenAI can produce short, animated films or realistic videos, enabling teachers to create dynamic content that engages students and explains complex concepts. Examples include:

• Sora: Converts text into realistic or imaginative videos useful for storytelling and concept visualisation.

• Runway ML: Provides video editing and creation tools, including AI-powered background removal and video synthesis.

• Synthesia: Creates studio-quality videos with AI avatars and voiceovers.

2.3.3 Music and Audio Creation

GenAI can unleash musical creativity by composing music, remixing existing songs, or creating custom sound effects. These capabilities enable students to enrich their projects, multimedia presentations, and language learning activities with personalised audio elements. Examples include:

• Amper Music: Enables educators and students to compose original music for projects or presentations.

• MuseNet: Generates music in various styles and mimics the sounds of various instruments.

• Jukebox: Provides a wide range of music, lyrics, and singing styles as input and produces new music from scratch.

2.3.4 Text-to-Speech and Speech-to-Text

Text-to-speech and speech-to-text tools can provide valuable support to teachers by reading texts aloud or transcribing speech into writing. These capabilities enhance the accessibility of lessons, catering to students with diverse learning needs. Examples include:

• Google Assistant: Offers robust text-to-speech and speech-to-text capabilities that are useful for accessibility and language practice.

• Amazon Polly: Converts text into lifelike speech, supporting audio-based learning.

2.3.5 Text Generation and Conversation

Powerful LLMs can generate human-like text, craft engaging stories, and provide personalised student feedback. Additionally, these tools power chatbots that provide instant assistance for teachers and learners. Examples include:

• ChatGPT: Generates text for lesson materials, writing prompts, and translating.

• Claude: Focuses on ethical and safe AI use, providing detailed, context-aware responses.

• Google’s Gemini: Excels at multimodal content creation, combining text, images, and audio for rich learning experiences.

2.3.6 Code Generation

GenAI can facilitate the development, writing, and debugging of programming code, making it a valuable resource for cross-disciplinary teaching. Leveraging GenAI tools can help students develop technical literacy, learn coding-related terminology in English, and practice language skills in authentic, real-world contexts.

• OpenAI Codex: Allows students to practise writing instructions or commands in English while observing how those instructions translate into code.

• GitHub Copilot: Encourages students to engage with English in specialised fields, helping them build confidence in communicating complex technical ideas.

• Amazon Code Whisperer: Provides intelligent coding suggestions for cloud-based applications, offering opportunities to practice reading and writing in a structured, goal-oriented environment.

2.3.7 Productivity and Lesson Planning

GenAI can streamline teachers’ workflows by assisting with lesson preparation, material creation, and administrative tasks. Examples of GenAI’s potential to automate and support tasks include

• Microsoft Copilot: Helps teachers create lesson plans, summarise data, and manage educational content efficiently.

• Magicschool.ai: Provides tools to generate rubrics, feedback, and newsletters tailored to classroom needs (explicitly designed for education).

• Diffit.me: Enables teachers to adapt content to suit various reading levels. Creates custom texts, quizzes, and activities by selecting the desired reading level and language.

• Perplexity: Answers questions, provides citations from web sources, and generates images and videos, making it ideal for enhancing classroom multimedia presentations.

When utilised effectively, these tools cater to second-language learners’ diverse needs while supporting teachers in delivering multimodal and personalised instruction.

Most of these tools have a specific ‘context window’ that limits the number of ‘tokens’ they can process simultaneously. Tokens represent the smallest building blocks of a prompt (e.g. part of a word, image, or video). If the input exceeds the tool’s capacity or the user interacts with it for an extended period, it may ‘forget’ some information needed to complete the task. Some tools also allow the user to set a ‘temperature’ value, which determines the randomness or creativity of the responses.

2.5.1 Ethical Concerns and Bias in GenAI

One of the most pressing challenges associated with GenAI is the potential for biased outputs, stemming from its training on large datasets that are often sourced from the internet. These datasets may perpetuate societal biases related to gender, race, culture, and socioeconomic status (Kumar, Reference Kumar2023; Maerten & Soydaner, Reference Maerten and Soydaner2023). For example, using ChatGPT to generate dialogues or instructional content such as worksheets may reinforce stereotypes or result in the creation of culturally insensitive materials. Learning materials that lack inclusivity or cultural relevance may alienate second-language learners or reinforce harmful stereotypes.

Second-language learners who expect GenAI tools to provide accurate and neutral information may encounter biased outputs that hinder their understanding of cultural norms, idiomatic expressions, or appropriate language use in real-world contexts (Godwin-Jones, Reference Godwin-Jones2024). This risk GenAI poses a significant threat to their ability to communicate effectively and confidently in intercultural situations.

2.5.2 Factual Inaccuracies and ‘Hallucinations’

GenAI tools are susceptible to ‘hallucinations’ – that is, generating plausible, yet entirely fabricated, content (Rudolph et al., Reference Rudolph, Tan and Tan2023). Faced with such inaccuracies, second-language learners might unwittingly internalise and utilise incorrect information.

GenAI tools can generate inaccurate explanations of grammar rules or word use (e.g. a GenAI tool might provide an incorrect example of the conditional tense). If the learner adopts and reinforces these errors in their writing or speaking, it can hinder their language development. Furthermore, fabricated citations or references could jeopardise the credibility of the learner’s academic work. These inaccuracies underscore the importance of human oversight in using GenAI for language learning. Teachers and learners must critically evaluate AI-generated content, treating it as a supplemental resource rather than a definitive source of truth.

2.5.3 Academic Integrity and Plagiarism

The increasing availability of GenAI tools raises significant concerns about academic integrity. ChatGPT can generate essays, translations, and assignments almost instantaneously, and it can be difficult to distinguish between human and AI-generated work (Marzuki et al., 2023). This creates the temptation for second-language users to over-rely on AI to complete tasks, potentially undermining their language development and discouraging the practising of essential skills such as organising ideas, constructing sentences, and using appropriate vocabulary.

The increasing difficulty of detecting AI-generated content poses a significant challenge to enforcing academic integrity. Learners may submit work produced by GenAI tools without proper attribution, violating academic honesty policies (Yeo, Reference Yeo2023). To address this issue, educators must emphasise the responsible use of GenAI tools and teach learners to integrate AI assistance into their work ethically while avoiding plagiarism.

2.5.4 Copyright and Intellectual Property Issues

Another critical challenge associated with GenAI involves the reproduction of copyrighted material (Chan & Hu, Reference Chan and Hu2023). Although GenAI tools generate new content, they are trained on datasets that may include copyrighted works, raising concerns about intellectual property violations. For second-language learners, legal and ethical issues may arise when they use AI-generated visuals, texts, or other materials for projects and assignments.

In language education, where creativity and originality are highly valued, unintentionally reproducing copyrighted material can undermine the authenticity of learners’ work. To mitigate this risk, teachers should guide learners in evaluating the originality of AI-generated materials and ensure compliance with copyright laws (Moorhouse et al., Reference Moorhouse, Yeo and Wan2023). Additionally, developers of GenAI tools must prioritise transparency regarding their training data sources to alleviate these concerns.

2.5.5 Accessibility and Digital Literacy

Despite the increasing availability of GenAI tools, a significant digital divide persists, resulting in unequal learning opportunities. Second-language learners from underprivileged backgrounds often have limited access to devices and reliable internet connectivity or insufficient digital literacy skills to utilise GenAI tools effectively. These factors put them at a disadvantage compared to their peers (Kamalov et al., Reference Kamalov, Calonge and Gurrib2023).

Even when learners have access to GenAI tools, they may struggle to use them effectively. They may not know how to frame effective prompts, have difficulty interpreting AI-generated feedback, or be unable to verify content accuracy. This lack of digital literacy can lead to frustration, misuse of tools, or a failure to fully benefit from their capabilities. To address this challenge, educators must integrate digital literacy training into language instruction and ensure that learners develop the skills and confidence to engage with GenAI responsibly (Moorhouse & Kohnke, Reference Moorhouse and Kohnke2024).

2.5.6 Mitigation Strategies

Using GenAI responsibly and equitably in second-language education requires careful oversight. Educators can design activities that encourage learners to critically evaluate AI-generated content and teach them to cross-check information and reflect on their learning process. Institutions must also implement AI-detection tools to prevent plagiarism and enforce academic integrity while fostering a culture of responsible AI use.

Beyond the risk of factual inaccuracies or plagiarism, over-reliance on GenAI tools may lead to the dehumanisation of the learning process and a diminished sense of learner agency. Algorithmic personalisation –the tailoring of content by AI tools based on user prompts or usage patterns –can unintentionally narrow learners’ exposure to diverse perspectives, reinforce confirmation bias, and limit opportunities for critical engagement (Selwyn, Reference Selwyn2023). As learners become passive consumers of AI-generated outputs, they may disengage from the cognitive and affective effort required in second-language acquisition: the need to grapple with ambiguity, negotiate meaning, and reflect on errors. Teachers, too, may lean on GenAI for content creation or feedback, which could reduce opportunities for meaningful interaction and empathy in language classrooms. To address these risks, language educators should foreground human-centred pedagogy, emphasise learner autonomy, and design tasks that require human interpretation, reflection, and judgment in tandem with AI use.

GenAI tool developers must strive to reduce bias, improve accuracy, and ensure compliance with copyright regulations. Collaborative efforts between educators, researchers, and policymakers are essential to establish guidelines for GenAI’s ethical use in education. These strategies will equip educators and learners to harness GenAI’s potential while minimising its risks and ensuring its effective integration into second-language education.

3.1.1 Microlearning’s Role in Second Language Acquisition

Research in second language acquisition (SLA) has provided robust evidence supporting microlearning’s effectiveness in language learning contexts. Studies have shown that microlearning enhances vocabulary retention, grammar acquisition, and listening comprehension (Kohnke, Reference Kohnke2023a; Luo & Li, Reference Luo and Li2025; Monib et al., Reference Monib, Qazi and Apong2025). For example, Rad (Reference Rad2023) discovered that microlearning helped second-language learners achieve significantly higher grammar skills than control groups using traditional methods. Opas (Reference Opas2023) incorporated TikTok as a supplementary listening tool and found it positively impacting students’ language proficiency. These findings suggest that microlearning’s concise format facilitates frequent, focused exposure to language input – a key condition for SLA (Khong & Kabilan, Reference Khong and Kabilan2022).

In addition, microlearning has proven effective in teaching speaking and pronunciation skills. Boonyabenjarit (Reference Boonyabenjarit2020) found that a microlearning approach effectively developed Thai students’ phonetics and overall pronunciation. Other studies have found video tutorials, particularly those incorporating interactive elements, allow learners to pause, rewatch, and practice at their own pace, fostering self-regulation and confidence in their language use (Dang et al., Reference Dang, Nguyen and Nga2022; Gao et al., Reference Gao, Tsai, Huang, Ma and Wu2023; Hamad et al., Reference Hamad, Metwally and Alfaruque2019). Gorham et al. Reference Gorham, Majumdar and Ogata2023 found that microlearning improved students’ peer feedback on spoken content. These studies highlight how microlearning’s learner-centred approach aligns with adult learning principles, offering autonomy, flexibility, and targeted instruction.

3.1.2 Meeting the Needs of Today’s Learners with Microlearning

Traditional educational approaches do not always meet the needs of today’s learners. The widespread use of smartphones, social media, and AI chatbots has contributed to shorter student attention spans and increased multitasking (Ng, Reference Ng2012). Modern learners often prefer fast, flexible, and personalised learning experiences over extended, lecture-based instruction. Typically digital natives, they engage with content in short bursts, expect on-demand access, and thrive in interactive, visual environments. Their learning is often informal, self-directed, and shaped by user experiences of platforms like YouTube, TikTok, and Duolingo (Kukulska-Hulme & Viberg, Reference Kukulska-Hulme and Viberg2018). They require instructional approaches that match their digital habits and cognitive preferences. Scholars have suggested that conventional teaching methods may not fully develop essential skills such as problem-solving and critical thinking (Darwin et al., Reference Darwin, Mukminatien, Suryati, Laksmi and Marzuki2024) and lack sufficient human interaction (Crompton et al., Reference Crompton, Edmett, Ichaporia and Burke2024).

Microlearning addresses these challenges by combining the most effective elements of various teaching strategies. This learner-centric method engages multiple senses and modalities, leveraging technology to enhance student participation and improve outcomes (Dolasinski & Reynolds, Reference Dolasinski and Reynolds2019). Studies suggest that microlearning’s concise and targeted format supports both knowledge retention and learner engagement. For instance, Erradei et al. (Reference Erradi, Almerekhi and Nahia2013) found that learners who engaged in gamified microlearning vocabulary lessons acquired more vocabulary than their peers who received conventional classroom lessons. The microlearning approach was particularly effective in both motivating learners and improving their performance.

By connecting with individual learners on a personal level, microlearning improves knowledge retention and increases student satisfaction. Furthermore, research on mobile microlearning indicates that learners benefit from the flexibility of accessing content on demand. Kohnke (Reference Kohnke, Corbeil, Corbeil and Khan2021a) discovered that second-language learners who could engage with microlearning lessons during commutes or idle moments strengthened their self-regulation skills. The accessibility of the lessons fostered frequent, meaningful language exposure, critical for second-language development.

3.1.3 Microlearning in Post-COVID-19 Education

The evolving educational landscape in the post-COVID-19 era has significantly altered curriculum design, pedagogical practices, and industry expectations. Academic institutions now offer classes in various modalities, including in-person, online (asynchronous and synchronous), hybrid (combining in-person and online learning), and trimodal (in-person, asynchronous online, and synchronous online). Consequently, educators across all levels and disciplines must explore innovative strategies to enhance their teaching practices and adapt them to the shifting learning environment. Microlearning’s concise and targeted nature makes it particularly well suited to ESL learning. Traditional ESL methods, often time-consuming and tedious, have become less appealing to modern learners. Today’s learners are not interested in sitting at desks, listening to lectures, taking notes, or completing long worksheets. Instead, they prefer using technologies like video, podcasts, interactive worksheets, and AI tools to learn English (Guan et al., Reference Guan, Zhang and Gu2025). Microlearning enables learners to quickly access essential language components, stimulating learning (Sankaranarayanan et al., Reference Sankaranarayanan, Leung, Abramenka-Lachheb, Seo and Lachheb2023). The convergence of shifting learner preferences and technology’s growing role has reinforced microlearning’s relevance as a teaching strategy (Rof et al., Reference Rof, Bikfalvi and Marques2024).

These studies collectively demonstrate that microlearning is not merely a theoretical concept, but a proven methodology that addresses the unique needs of language learners in diverse educational contexts. Whether through mobile apps, gamified platforms, or video-based tutorials, microlearning offers a versatile and effective approach to SLA. Educators can foster deeper engagement, improve retention, and ultimately enhance learning outcomes by focusing on targeted, digestible content.

Two recent events have underscored the need for innovative teaching practices. First, the COVID-19 pandemic, which began in December 2019, prompted many schools to shift their programmes entirely online; since then, HyFlex, blended, and online learning modalities have become the new norm (Moorhouse & Kohnke, Reference Moorhouse and Kohnke2021). Second, the advent of publicly available GenAI with ChatGPT’s release in November 2022 is likely to revolutionise education (Kohnke et al., Reference Kohnke, Moorhouse and Zou2023a). Amidst these changes, researchers and educators have begun focusing on developing students’ digital and AI literacy and investigating relevant pedagogies (Walter, Reference Walter2024). As GenAI tools have continued to evolve, AI-assisted microlearning has emerged as a significant learning modality (Kohnke, Reference Kohnke2024b).

3.2.1 Blending Multiple Delivery Methods

Microlearning is a versatile and effective approach to teaching and learning. It combines various delivery methods (in-person, hybrid, online, etc.) to maximise the impact of each component (Kohnke, Reference Kohnke2023a). Microlearning helps learners focus on crucial information by leveraging engaging formats like interactive quizzes, short video tutorials, and podcasts, combatting short attention spans and maintaining interest. This targeted and efficient learning approach addresses specific learning objectives in concise, manageable segments (Corbeil et al., Reference Corbeil, Khan and Corbeil2021; Kohnke & Moorhouse, Reference Moorhouse2024).

Microlearning’s learner-centric approach integrates different modalities (audio, video, images, etc.) with hands-on learning, both inside and outside the classroom (Kohnke, Reference Kohnke2023a). Its multisensory and multimodality design increases the likelihood that it will resonate with a broad range of students in today’s fast-paced, technology-driven world, making it a valuable tool in their language-learning arsenal. This method is particularly well-suited for today’s learners for several reasons:

• Speed and Efficiency: Microlearning enables English language learners to rapidly acquire and integrate new vocabulary, grammar structures, and language skills into their communication practice. This accelerated learning process enhances fluency and confidence in language use.

• Engagement and Retention: By offering engaging, multimedia-rich content, microlearning helps English language learners maintain interest and retain information more effectively. Examples of interactive microlearning resources include animated explanations of idioms, interactive pronunciation exercises, and short videos demonstrating real-life conversations. These resources make complex language concepts more accessible and memorable.

• Personalisation and Flexibility: Educators can customise microlearning to suit individual learning needs and preferences, enabling English language learners to select modules relevant to their immediate language-learning goals. For instance, learners preparing for a job interview might focus on modules related to professional vocabulary and interview questions.

Microlearning prioritises English language learners’ self-improvement. It encourages meaningful engagement with content by highlighting key information and mitigating the impact of diminishing attention spans.

3.2.2 Rethinking Course Content Delivery

Language teachers adopting microlearning must reconsider traditional content delivery methods. Rather than limiting the number of delivery methods, microlearning encourages the integration of multiple delivery methods within a single module. This multimodal approach leverages learning formats such as PDFs, podcasts, infographics, videos, augmented reality, virtual reality, and AI to convey essential messages or bursts of information (Kohnke, Reference Kohnke, Corbeil, Corbeil and Khan2021a). Each ‘snackable nugget’ focuses on a single idea (e.g. a specific grammar point) to minimise the risk of cognitive overload (Nikou, Reference Nikou and Graziano2019) and enhance retention (Jomah et al., Reference Jomah, Masoud, Kishore and Aurelia2016).

Examples of effective formats include:

• Infographics and Animated GIFs: These tools provide a concise visual summary of complex language concepts, such as verb tenses or sentence structure, making them ideal for quick reference and review. They break down information into easily understandable pieces, facilitating quick reference, review, and retention of key details.

• Podcasts and Audio Clips: Audio-based microlearning enables English language learners to multitask and learn while commuting or during other activities. It increases the frequency of learning opportunities throughout the day, making language acquisition more convenient and efficient.

• Videos and Screencasts: These visual tools provide step-by-step demonstrations of language skills such as techniques for carrying on conversations, which are particularly effective for demonstrating how to apply skills in practical contexts. English language learners can pause and rewatch videos, allowing them to progress at their own pace and review materials as needed.

• Interactive Tools and Chatbots: These technologies provide hands-on experience and immediate feedback, enhancing interactive learning and engagement among English language learners. They simulate real-life scenarios and conversations, offering a dynamic learning environment that adapts to the user’s responses and helps them practice language skills in context.

By incorporating these diverse formats and a multisensory, multimodal approach, microlearning engages English language learners to enhance motivation and performance (Corbeil et al., Reference Corbeil, Khan and Corbeil2021). This variety of formats caters to individual learning preferences, increasing the likelihood of appealing to diverse learners.

The choice of delivery method depends on the content. A short video lecture followed by an activity can mitigate the problem of reduced attention spans (Dolasinski & Reynold, Reference Dolasinski and Reynolds2019). Seidel (Reference Seidel2024) discovered that segmented videos of suitable length resulted in higher learning gains. Research has identified a positive correlation between video views and student satisfaction (Beatty et al., Reference Beatty, Merchant and Albert2019; Tiernan & O’Kelly, Reference Tiernan and O’Kelly2019). To reinforce content and promote short- and long-term retention, follow-up activities can incorporate visuals and infographics (VanderMolen & Spivey, Reference VanderMolen and Spivey2017), writing tasks (e.g. Google Docs) or speaking tasks (e.g. Elsa Speak, TalkPal) that require students to organise and demonstrate their learning.

3.2.3 Leveraging Social Media and AI Tools

Integrating social media platforms, apps, and basic AI tools into microlearning has created new opportunities to enhance learning experiences. Social media platforms, in particular, have proven effective in delivering microlearning content in concise, engaging formats that align with modern learners’ preferences for quick information. Platforms like Instagram, YouTube, and WhatsApp enable educators to share bite-sized videos, infographics, and interactive quizzes that learners can engage with. This flexibility allows learners to integrate learning into their daily routines – for example, commutes or breaks – without feeling overwhelmed by lengthy, traditional lessons.

The dynamic presentations by social media apps can shift learning from memorisation to comprehension and application (Keer & Frese, Reference Keer and Frese2017). For example, short videos demonstrating real-world language or infographics summarising grammar rules can make complex concepts easier to understand and retain. Social media platforms facilitate interaction and collaboration through polls, discussion threads, and live Q&As, fostering a sense of community among learners and enhancing motivation and engagement (Barrot, Reference Barrot2022).

The incorporation of streaming videos and gamified activities on devices can foster self-regulation, lifelong learning (Reinhard & Elwood, Reference Reinhardt, Elwood and Keengwe2019). Microlearning enables learners to stream concise, visually engaging videos, pause, rewatch, and review content. For instance, educators curate videos on platforms like Instagram or TikTok to demonstrate conversational scenarios, pronunciation tips, or cultural nuances in language use. These approaches make complex language concepts accessible and memorable. Small, achievable gamified milestones motivate learners to complete microlearning activities (Shamir-Inbal & Blau, Reference Shamir-Inbal and Blau2022). These milestones inspire students and nurture a sense of accomplishment, making them effective for younger learners accustomed to gaming mechanics in their digital interactions (Schöbel et al., Reference Schöbel, Saqr and Janson2021).

Learners can interact with microlearning tools regardless of time and location (Torgerson & Lannone, Reference Torgerson and Lannone2019). For example, an ESL learner can utilise a gamified app to complete vocabulary challenges, earn rewards, and track progress. Adapting lesson content based on learners’ preferences and incorporating multiple modalities facilitates self-regulated learning (Monid et al., Reference Monib, Qazi and Apong2025), increases knowledge retention (Kabudi et al., Reference Kabudi, Pappas and Olsen2021), and caters to digitally connected learners. This approach is optimal for learners accustomed to mobile devices, social media, and AI chatbots.

Basic AI tools like chatbots and adaptive algorithms enhance microlearning. Chatbots simulate practice, allowing learners to engage in dialogue and receive instant feedback (Kuhail et al., Reference Kuhail, Alturki, Alramlawi and Alhejori2023). Adaptive algorithms track learners’ progress and suggest personalised activities based on strengths and weaknesses (Kabudi et al., Reference Kabudi, Pappas and Olsen2021). For instance, if a learner struggles with verb conjugation, the system recommends targeted practice exercises or video tutorials.

Leveraging these technologies makes microlearning more engaging, interactive, and tailored to learners’ needs. Social media platforms provide familiar and flexible environments for informal learning, while AI tools personalise the learning experience without requiring technological infrastructure. These possibilities create a foundation for effective microlearning that appeals to diverse learners and supports their goals.

4.1.1 Key SLA Principles and Theories in Microlearning

Microlearning comprises several core SLA principles, making it a useful tool for language learning. For instance, retrieval practice – a process of actively recalling information – has been demonstrated to enhance retention and memory consolidation significantly (Roediger & Butler, Reference Roediger and Butler2011). Retrieval practices in microlearning reinforce memory retention by integrating techniques, such as quizzes, flashcards, and sentence reconstruction that actively involve learners with the materials. This process aligns with the testing effect, which shows that retrieving information from memory strengthens learning (Roediger & Butler, Reference Roediger and Butler2011). These design practices also resonate with cognitive load theory (Sweller, Reference Sweller2020) and the use of ‘snackable’ and ‘digestible’ content. Such approaches allow learners to revisit content at intervals, reducing cognitive overload and enhancing processing efficiency (Kohnke, Reference Kohnke, Corbeil, Corbeil and Khan2021a).

Krashen’s (Reference Krashen1982) input hypothesis asserts that learners acquire language most effectively when exposed to comprehensible input – that is, language that can be understood with the help of context, visual cues, or prior knowledge. This input should be slightly more advanced than the learner’s current proficiency level (i+1), providing an optimal challenge while remaining accessible. Such input is an essential factor in SLA. Thus, microlearning provides comprehensible input with contextualised and relevant content via short dialogues, videos, or reading passages on topics that can be tailored to learners’ needs. In addition, embedding the target language in meaningful and relevant contexts that mirror real-world communication enables students to learn the language naturally. For example, a microlearning lesson on ordering food at a restaurant could present a short video showing a speaker ordering food, followed by a quiz on key phrases and vocabulary. Swain’s (Reference Swain, Gass and Madden1985) output hypothesis also treats language production as essential to the development of language proficiency. In microlearning, opportunities for output can be offered through speaking prompts, short writing tasks, and pronunciation exercises. For instance, ELSASpeak encourages learners to practice and refine their language skills, fostering active production.

Feedback, another essential SLA component, enables learners to identify and correct errors (Long, Reference Long, Ritchie and Bhatia1996). Microlearning activities often incorporate immediate, corrective feedback during exercises, ensuring that learners can adjust their understanding and improve accuracy. This built-in feedback aligns with Schmidt’s (Reference Schmidt1990) noticing hypothesis, which asserts that learners must recognise the gap between their current output and the target language. Integrating feedback into each microlearning activity guides learners towards constant improvement of their skills. Additionally, motivation is a driving force in language learning (Dörnyei, Reference Dörnyei2001) and a cornerstone of microlearning. Features such as gamification, progress tracking, and rewards in the form of badges and leaderboards keep learners interested and motivated to learn. Furthermore, microlearning’s short, manageable lessons reduce cognitive overload and maintain learners’ motivation as they gradually build proficiency.

Microlearning’s design principles further enhance its effectiveness in language learning. Bite-sized content, lasting 2–8 minutes, reduces cognitive overload and narrows learners’ attention to specific, achievable objectives. Each microlearning unit has a clear learning goal, ensuring learners can focus on mastering one concept at a time. For example, a lesson might target using modal verbs in polite requests or pronouncing specific phonemes. Microlearning also supports just-in-time learning, allowing learners to access content when needed (Rof et al., Reference Rof, Bikfalvi and Marques2024). This immediacy fosters practical, real-world language-skill applications, such as preparing for a meeting or practising small talk before social interactions.

Most microlearning modules are designed for mobile phones or tablets. Consequently, lessons are optimised for small screens and incorporate multimedia elements, such as videos, animations, and interactive exercises. Mayer’s (Reference Mayer2021) cognitive theory of multimedia learning (CTML) suggests combining elements like text, visuals, and interactivity to enhance comprehension and retention by engaging multiple cognitive channels. The mobile-first design also accommodates learners’ preferences and habits, enabling study during commutes, breaks, or other spare moments. This accessibility further aligns with the self-determination theory, which emphasises autonomy and competence to foster intrinsic motivation (Ryan & Deci, Reference Ryan and Deci2017). Microlearning meets these needs by allowing learners to progress at their own pace, track their achievements, and engage with content that reflects their interests and goals.

Grounded in sound pedagogy and SLA theories, microlearning can be a highly effective approach to language learning. Microlearning boosts retention, engagement, and motivation by leveraging spaced retention, retrieval practice, and meaningful input. It also aligns with cognitive frameworks like CLT, CTML, and SDT. The design of microlearning (e.g. bite-sized content and mobile-first accessibility) makes it an innovative and practical solution for today’s language learners.

4.2.1 Pedagogical Alignment

The foundation of an effective microlearning activity lies in a clear and well-defined learning goal. Before designing content or selecting tools, educators should articulate what they want learners to achieve by the end of the activity. This focus ensures that all design decisions align with the intended learning outcomes to create a purposeful and impactful microlearning experience.

Well-structured pedagogical alignment requires constructive alignment (Biggs, Reference Biggs1996) of interrelated learning objectives, instructional activities, and assessments. When using GenAI tools in microlearning, educators must ensure that AI-generated content aligns with Bloom’s taxonomy (Rivers & Holland, Reference Rivers and Holland2023), guiding learners from lower-order thinking skills (recall, understanding) to higher-order cognitive processes (application, analysis, and evaluation). For example, an AI-powered microlearning module on critical thinking should progress from foundational knowledge (definitions and key concepts) to interactive problem-solving tasks that require learners to apply and evaluate information. The AI tool could initially present definitions of logical fallacies (recall) and then ask learners to identify fallacies in short arguments (application). Finally, the tool could challenge learners to analyse and evaluate the validity of complex arguments (analysis and evaluation).

Scaffolding theory (Vygotsky, Reference Vygotsky1978) suggests that learners benefit from structured guidance that gradually diminishes as they develop mastery. Microlearning activities designed with GenAI tools should incorporate adaptive scaffolding, with AI personalising content based on learners’ progress and needs (Holmes & Tuomi, Reference Holmes and Tuomi2022). This approach ensures that learning remains engaging and appropriately challenging for diverse learners. As cognitive load theory (Sweller, Reference Sweller1988) highlights the need to minimise extraneous cognitive load in microlearning environments, GenAI tools should be used to enhance clarity, summarise complex information, and generate concise, structured content that aligns with learners’ cognitive capabilities (Mayer, Reference Mayer2024). Overloading microlearning activities with excessive AI-generated content may compromise retention and engagement.

Pedagogical alignment in GenAI-driven microlearning requires a learner-centred approach to ensure relevant, structured, and cognitively appropriate content. Proper alignment fosters deeper learning, engagement, and knowledge transfer, thereby maximising the effectiveness of microlearning experiences.

4.2.2 Ethical and Responsible Use of AI in Microlearning

Integrating GenAI tools in microlearning environments requires addressing unique challenges to ensure responsible use. Although AI-driven microlearning offers personalised learning experiences, real-time feedback, and adaptive content, its output can be biased or inaccurate, and unregulated use may lead to academic dishonesty and privacy concerns (Floridi & Cowls, Reference Floridi and Cowls2019). As AI becomes an increasingly integral to microlearning, educators must adopt ethical design principles and critical oversight to ensure that AI-generated content remains accurate, fair, and pedagogically sound; AI tools must uphold academic integrity and earn learner trust.

A primary ethical concern in AI-powered microlearning is the potential for biased or misleading instructional materials. Since GenAI tools are trained on large datasets, they may inadvertently reinforce stereotypes or exclude diverse perspectives (Moorhouse, Reference Moorhouse2024). This is particularly problematic in language learning and culturally nuanced subjects; inaccuracies in AI-generated examples can hinder learner comprehension and engagement. To mitigate risks, educators must rigorously review AI-generated content before integrating, ensuring it reflects diverse perspectives and avoids stereotypes. Embedding AI-auditing practices design, including the systematic review of AI-generated materials before use, can help mitigate bias and improve content reliability (Holmes et al., Reference Holmes, Bialik and Fadel2019). For instance, educators should verify that AI-generated dialogue reflects cultural norms accurately before use in lessons on cultural etiquette.

Another concern is the risk of hallucinated or factually inaccurate content. AI models sometimes generate plausible, yet incorrect, information (Rudolph et al., Reference Rudolph, Tan and Tan2023). These inaccuracies can be detrimental to language acquisition because learners rely on precise terminology, grammar rules, and pronunciation guidance. To address this challenge, educators should cross-check AI-generated content against verified sources before integrating it into microlearning modules. Encouraging learners to critically evaluate AI outputs, rather than passively accepting them, fosters AI literacy and responsible engagement with GenAI tools.

Additionally, preventing AI over-reliance is crucial to promoting active learning. Learners may be tempted to consume AI-generated content passively. Microlearning should require learners to analyse, critique, or expand upon AI-generated responses to foster critical thinking and deep engagement. Ideally, AI should function as a support mechanism, rather than a shortcut, ensuring that microlearning remains an active and reflective process.

Many AI-powered microlearning platforms collect user data to personalise content and track progress. However, this raises concerns about data security, informed consent, and compliance with privacy regulations (e.g. GDPR, FERPA; Floridi & Cowls, Reference Floridi and Cowls2019). Educators and institutions must ensure that AI integration does not compromise learner privacy or autonomy. AI tools used in microlearning should have transparent data policies, enabling learners to understand how their data is collected and used.

4.2.3 Personalisation and Adaptive Learning

Perhaps most significantly, GenAI enhances microlearning by tailoring learning experiences to individual learners’ needs. AI-based microlearning systems can modify content in real-time to foster more efficient and engaging learning (Popenici & Kerr, Reference Popenici and Kerr2017). Adaptive learning frameworks leverage AI algorithms to assess learner interactions and modify content accordingly (Siemens & Long, Reference Siemens and Long2011). For instance, if a learner struggles with a topic, the AI tool (e.g. ChatGPT) can provide additional examples, break down explanations, or offer supplementary resources to facilitate understanding (Holmes et al., Reference Holmes, Bialik and Fadel2019). Conversely, AI can accelerate the learning trajectory for learners demonstrating competency by introducing more significant challenges or higher-order thinking tasks (Walter, Reference Walter2024). For example, an AI-driven platform in language learning can adjust vocabulary exercises based on a learner’s proficiency level, providing simpler sentences for beginners and complex, idiomatic expressions for advanced learners. This approach ensures that learners receive challenging, yet appropriate, instruction tailored to their skill levels. Furthermore, if a learner consistently struggles with a specific grammatical structure, the AI tool could provide targeted exercises and explanations focused explicitly on that structure. This approach aligns with the principle of individualised instruction, which acknowledges that learners have distinct needs.

The integration of gamification (e.g. points, challenges, instant feedback) and microlearning can optimise personalisation by providing adaptive quizzes, instant feedback, and AI-generated practice exercises (Hamari et al., Reference Hamari, Shernoff and Rowe2016). For example, a language-learning chatbot (e.g. Mizou) can simulate real-life conversations, adjusting responses based on learner proficiency and providing targeted feedback on pronunciation, grammar, and fluency.

4.2.4 Structured Examples of AI-Powered Microlearning in Language Learning

The use of GenAI in microlearning for language acquisition has transformed skill development and knowledge retention (Kohnke, Reference Kohnke2023a). The following examples illustrate how AI-driven microlearning can support language acquisition through personalised, interactive, and contextually relevant learning experiences.

5.4.1 Blended Feedback Approaches in Microlearning

To leverage the strengths of both AI and human feedback, instructors can implement a blended model of feedback that combines the efficiency of AI with the insightfulness of human comprehension. A practical strategy is to create microlearning sequences in which AI manages low-stakes, formative activities (e.g. grammar drills, vocabulary drills), and human instructors engage in more complex learning activities (e.g. writing assignments, oral presentations). For instance, a microlearning module for formal email writing can start with AI grammar correction and vocabulary suggestions. After the learners have completed their drafts, a teacher can give feedback on tone, clarity, and suitability for the target audience. This split of tasks guarantees that AI is leveraged to give feedback on mundane tasks so that teacher time is allocated for more subtle, personalised feedback.

Another strategy is layered feedback, in which students receive AI feedback first and then work with a teacher or a peer to review it. This promotes metacognition and teaches students to critically consider the quality of the AI feedback they are receiving. Teachers can use AI-generated analytics (e.g. error patterns, time on task, quiz scores) and make their own feedback more learner-specific.

The user experience is important in blended feedback. Duolingo and ELSA Speak are examples of tools that encourage learners without frustrating them by offering easy-to-use interfaces and instant feedback.

In designing feedback, teachers must take into account learners’ needs and preferences. Some students will be engaged by instant AI feedback, and others will be helped by the relational and conversational quality of human feedback. Giving students time to reflect on both sources of feedback can enable them to become more strategic and self-directed learners. By coupling the scalability of AI with the pedagogical strengths of human teaching, educators can create more interactive and adaptive microlearning experiences.

6.2.1 Leveraging AI for Enhanced Engagement

Traditional digital learning tools such as instructional videos typically enable passive content consumption. With GenAI tools, learners can interact with AI, generate responses, and receive real-time feedback. This shift supports active learning, in which learners participate, rather than being passive recipients of information (Jonassen, Reference Jonassen and Reigeluth1999; Law, Reference Law2024). For GenAI-powered microlearning to be effective and engaging, educators should

• Use AI-generated scaffolding to break complex topics into digestible, structured chunks.

• Incorporate adaptive AI-driven difficulty levels so learners receive content matched to their proficiency.

• Leverage AI chatbots and simulations to allow learners to practice real-world scenarios interactively.

Microlearning must also be optimised for mobile learning environments (e. g. Kohnke, Reference Kohnke, Corbeil, Corbeil and Khan2021a) (see Tables 6 and 7). Since many learners access learning content on their mobile devices, educators should:

• Ensure that the content is concise, structured, and easy to navigate.

• Use responsive design to accommodate different screen sizes.

• Minimise text-heavy formats, substituting AI-generated visuals, audio, and interactive elements.

6.2.2 Ensuring Ethical, High-Quality AI-Generated Content

Educators who use GenAI tools to generate materials must ensure they are accurate, contextually relevant, and aligned with pedagogical best practices. Inaccurate or biased AI-generated content increases extraneous cognitive load (Sweller, Reference Sweller1994), forcing learners to process misleading information rather than focus on meaningful learning.

One of the main risks of AI-generated microlearning is the potential for hallucinated or misleading content: plausible, but incorrect, information generated by AI tools (Ferrara, Reference Ferrara2023). To mitigate this risk, educators should

• Pre-screen AI-generated microlearning content before integrating it into lessons to ensure factual accuracy and alignment with learning objectives.

  1. ◦ Educators should cross-check AI-generated explanations against credible sources such as Cambridge Grammar for language learners or Khan Academy for technical concepts when using ChatGPT or Claude to generate materials.

  2. ◦ Educators using Quillbot or Grammarly should use established linguistic references like Merriam-Webster or Oxford Advanced Learner’s Dictionary to verify grammar explanations.

• Use AI to supplement, not replace, expert-curated microlearning materials. AI-generated explanations should be verified and refined by educators.

  1. ◦ Platforms like Duolingo and LingQ use AI to generate adaptive language exercises. Human instructors should review AI-created sentence structures and contextual examples to ensure natural, culturally appropriate phrasing.

• Leverage AI tools with built-in content verification features. Some AI platforms incorporate fact-checking and citation-based content generation.

  1. ◦ Perplexity and SciSpace are research assistants that can generate responses with citations, making them helpful for verifying information.

  2. ◦ Google’s Gemini integrates real-time search capabilities to fact-check its AI-generated responses, helping educators retrieve updated, verified content.

Another key concern is over-reliance on AI-generated feedback in microlearning. AI provides instant feedback on grammar, pronunciation, and quizzes but may be insufficient to teach critical thinking skills. Educators should

• Combine AI-driven microlearning with human feedback. For example, AI can be used to assess basic writing mechanisms while teachers provide contextual feedback on structure and argumentation.

  1. ◦ Grammarly and DeepL Write provide instant grammar corrections and stylistic suggestions but do not assess persuasiveness or logical cohesion. Teachers should review AI-suggested edits to ensure that texts remain cohesive, well-structured, and aligned with academic integrity.

  2. ◦ For pronunciation practice, ELSA Speak and Speechling provide AI-driven real-time pronunciation coaching, but human instructors should still review intonation, fluency, and cultural appropriateness.

• Encourage self-reflection and peer discussion. Learners should be prompted to critically evaluate AI-generated explanations rather than passively accepting them.

  1. ◦ AI-powered platforms like ChatGPT and Google Bard allow learners to ask complex questions and receive instant responses. However, educators should incorporate reflective activities in which learners analyse AI-generated content, identify inconsistencies, and discuss biases.

  2. ◦ Tools like Peergrade and Turnitin Draft Coach help learners critique AI-generated feedback collaboratively, allowing them to develop critical thinking skills while refining their work.

Bias in AI-generated microlearning activities can reinforce stereotypes or exclude diverse perspectives. To ensure inclusivity,

• Review AI-generated examples, dialogues, and quizzes to identify and remove cultural biases and inappropriate phrasing.

  1. ◦ When using ChatGPT or Duolingo’s AI tutor to generate language-learning dialogues, educators should review sentence structures, names, and cultural contexts to avoid reinforcing stereotypes or using outdated expressions.

  2. ◦ Summaries by Perplexity AI and Microsoft Copilot generated for use in history and social studies lessons should be cross-checked against multiple sources to ensure balanced perspectives.

• Use diverse AI training datasets when possible and supplement AI-generated content with human-curated materials that reflect global perspectives.

  1. ◦ AI-powered platforms like Canva AI and DALL-E 3 can generate educational visuals, but educators should curate imagery to ensure the representation of different cultures, ethnicities, and backgrounds.

  2. ◦ Reading comprehension exercises generated by tools like ReadTheory and CommonLit AI should be supplemented with human-selected texts from diverse authors to promote inclusivity in literature and storytelling.

By implementing these strategies for AI-powered microlearning, educators can ensure that lessons are engaging, accurate, and pedagogically sound while they benefit from AI’s ability to personalise and scale learning.

6.2.3 Supporting Diverse Learner Groups with GenAI-Powered Microlearning

GenAI technologies significantly enhance the inclusivity and accessibility of microlearning. Educators can leverage AI-powered personalisation and multimodal capabilities to address diverse learner groups’ needs (see Table 8). For example, special education learners can be provided with adaptive content personalised to their capabilities, interactive AI simulations to practice real-life scenarios safely, and accessibility features like text-to-speech or speech-to-text. Multilingual learners can leverage AI-powered real-time translations, adaptive vocabulary lessons, and AI-created culturally authentic situations to reduce cognitive load and enable deeper comprehension. Adult language learners, often motivated by practical concerns, can leverage scenario-based AI role-plays, workplace-contextualised AI-generated content, and adaptive quizzes at their proficiency levels.

These tactics align closely with established theories. Personalisation and interactive simulations for special education learners, for example, reflect constructivist theories by supporting individualised knowledge construction and the principles of self-regulated learning via instantaneous AI feedback. The reduction of cognitive load on multilingual learners via culturally relevant and adaptive AI-generated content aligns with cognitive load theory (Sweller, Reference Sweller1994). The engagement of adult learners with authentic, contextually relevant tasks designed by AI aligns with constructivist learning principles and enhances intrinsic motivation, which is consistent with self-regulated learning theory (Zimmerman, Reference Zimmerman2000).

6.2.4 Limitations and Practical Challenges of GenAI-powered Microlearning

Educators implementing GenAI-powered microlearning should be aware of real-world challenges and consider practical mitigation strategies:

• Teacher Workload: Screening and validating AI-generated content (e.g. practice questions, explanations) can increase teacher workload, given the need to check for age-appropriateness, language accuracy, and cultural sensitivity before assigning the content to students.

  1. ◦ Strategy: Form collaborative teacher teams to distribute the workload, co-develop prompt templates, and share vetted AI-generated resources through shared repositories or institutional platforms.

• Technical Barriers: Limited technical proficiency among educators or students, along with insufficient infrastructure (e.g. outdated devices or lack of reliable internet), can slow the adoption of GenAI tools. For example, a teacher unfamiliar with prompt engineering may struggle to generate effective learning materials, or a student using a low-spec phone may experience lags in AI-enabled apps.

  1. ◦ Strategy: Offer ongoing professional development focused on GenAI literacy, provide step-by-step user guides, and select intuitive, low-barrier tools that work across devices and platforms.

• Digital Divide and Equity: Unequal access to reliable internet, devices, or AI tools can create disparities in learning opportunities. Students in low-resource regions may be unable to access cloud-based platforms or streaming content. For example, a learner without stable internet cannot benefit from real-time AI feedback or voice-based interactions in mobile apps.

  1. ◦ Strategy: Ensure that microlearning experiences are accessible offline or on low-bandwidth devices and advocate for equitable access to technology and resources. Design offline-compatible activities as backup options.

• Data Privacy and Ethical Concerns: AI platforms often collect and process learner data, raising concerns about privacy, consent, and data security. Some GenAI tools may store user inputs or usage patterns without transparent data-handling policies, which can create trust issues among parents or institutions.

  1. ◦ Strategy: Select GenAI tools that are compliant with privacy regulations (e.g. GDPR, FERPA) and communicate data-handling practices transparently, and include digital literacy components in the curriculum to educate learners about responsible AI use and online safety.

7.1.1 The Need for AI-Enhanced TPD

Despite the importance of lifelong learning for teachers, traditional TPD models often fail due to various limitations:

• Lack of personalisation: Generic workshops do not address individual teachers’ pedagogical challenges, such as adapting AI tools for student-centred learning or incorporating AI-driven formative assessments in courses.

• Limited practical applications: Many TPD programmes focus on theoretical discussions without offering hands-on experience with emerging AI tools.

• Time constraints: Teachers have busy schedules and cannot always commit to lengthy training sessions.

• Technology gaps: Many teachers feel unprepared to integrate AI into their classrooms.

AI can enhance TPD to help new and experienced teachers to adapt quickly to new technologies and pedagogies. It can provide (a) adaptive learning paths tailored to teachers’ subject expertise and skill levels, (b) on-demand AI-powered assistants (e.g. ChatGPT, Perplexity) that provide real-time support, and (c) microlearning modules that allow for self-paced, bite-sized learning experiences. For example, a teacher struggling with AI-assisted assessment can receive real-time guidance from an AI tutor (e.g. Coursera Coach) that curates personalised learning paths based on teachers’ needs.

7.1.2 Microlearning’s Role in AI-Driven TPD

Microlearning TPD comprises short, focused learning sessions that help teachers to acquire new knowledge in step-by-step sessions (Kohnke, Reference Kohnke2024b) that align with their busy schedules. Unlike traditional professional development, microlearning offers instant, interactive, and flexible learning for classroom use. Moreover, microlearning engages teachers in TPD through discussions, hands-on practice, and reflection to help them to understand and retain new concepts (Borko, Reference Borko2004; Garet et al., Reference Garet, Porter, Desimone, Birman and Yoon2001).

AI-powered microlearning must be interactive and modular. Each microlearning event should be stand-alone (2–8 minutes), interactive, directly applicable to teachers’ professional needs, and allow for progressive skill development. Modular, continuous TPD can help teachers to experiment with new strategies, receive feedback, and adjust their instructional practices. These sessions can include follow-ups, coaching, and peer collaboration to reinforce a commitment to lifelong learning (e.g. Darling-Hammond et al., Reference Darling-Hammond, Hyler and Gardner2017). Table 11 presents a structured framework for designing AI-driven microlearning sessions for TPD.

Table 12 demonstrates the use of this framework to teach specific AI tools – Grammarly and Synthesia – in microlearning sessions.

Microlearning’s structured approach ensures that AI-driven TPD is focused, time-efficient, and directly applicable to classroom practices. Teachers can enhance their professional growth without feeling overwhelmed as microlearning breaks down complex technological and pedagogical concepts into manageable learning units (Shamir-Inbal & Blau, Reference Shamir-Inbal and Blau2022). This structure allows teachers to build confidence and competence progressively (Kohnke & Foung, Reference Kohnke, Foung, Tafazoli and Picard2023) and gradually integrate new skills and knowledge into their teaching (Kohnke & Moorhouse, Reference Moorhouse2024).

AI-driven microlearning TPD offers flexibility (teachers can engage with modules during breaks or at home) and reduced cognitive load (content is delivered in small units). Additionally, microlearning can be applied immediately (e.g. teachers can use the AI tools in their classrooms as soon as they learn about them). For example, microlearning modules can support teachers in using tools such as Mizou, an AI chatbot for language learning, in the classroom, bridging the gap between theory and practice (Hubbard, Reference Hubbard, Ziegler and González-Lloret2022). A microlearning module designed to introduce Mizou could include (a) a short video overview of the website and chatbot interface and (b) an infographic-based walkthrough on building an AI-generated chatbot, including how to specify learning objectives, grade levels, and format variations (quiz-based, storytelling, etc.). Microlearning bridges the gap between theory and practice. It provides teachers with immediate hands-on practice, enabling them to grasp a tool’s potential for engaging students quickly. Microlearning promotes higher motivation and creates a sustainable and enjoyable learning experience (Emersen & Berge, Reference Emerson and Berge2018; Kohnke et al., Reference Kohnke, Foung, Zou and Jiang2024b).

7.2.1 AI for Personalised Professional Learning

One of the most powerful applications of GenAI in TPD is generating personalised learning experiences based on teachers’ subject-matter expertise, interests, and professional needs. AI-powered platforms can provide on-demand TPD, curated learning materials, and real-time pedagogical guidance, helping teachers to develop new skills and use AI tools (see Tables 13 and 14).

AI-powered assistants and research tools enable on-demand professional learning by allowing teachers to access and digest research, obtain content recommendations, and refine teaching methods. These tools help translate educational theory into practical classroom strategies by offering real-time pedagogical insights. However, to maximise their utility, teachers should critically evaluate AI-generated content before applying it, verifying that the findings align with credible sources to ensure evidence-based teaching. Teacher educators can further support this process by designing microlearning tasks that encourage teachers to support this process by designing microlearning tasks that encourage teachers to use their expertise to assess and adapt AI-generated insights. While AI can be a valuable supplement, it should not replace pedagogical expertise, and all AI-generated responses must be cross-checked for validity.

7.2.2 AI for Content Creation in TPD

AI can automate time-consuming tasks, allowing teachers to focus on innovative teaching strategies rather than creating content from scratch. AI-powered tools can generate lesson plans, quizzes, interactive activities, and training materials for TPD programmes (see Table 15).

AI-driven content-generation tools save time, enhance teaching materials, and provide pre-designed templates for TPD. Teachers and teacher educators who generate lessons and quizzes automatically can focus on adjusting the content to meet professional learning goals, while ensuring AI-generated content is accurate and relevant. Best practices for implementation include the following:

• Teachers should use AI tools as brainstorming aids for their professional development, such as generating ideas for lesson plans or reflective practices.

• Teachers and teacher educators should identify potential biases in AI-generated content and adjust materials to ensure inclusivity and accuracy.

• Teacher educators should train teachers to critically evaluate AI tools and integrate them effectively into their professional learning routines.

7.2.3 AI for Augmented Feedback and Reflection

GenAI can support self-reflection and continuous improvement by providing automated feedback on teaching practices. AI-powered transcription and analytics tools can help teachers to review their lessons and identify areas for growth (see Tables 16 and 17).

AI tools that offer feedback can help teachers assess their teaching effectiveness and make data-driven improvements. They can provide feedback on teacher talking time, student engagement levels, and writing clarity. However, peer and self-reflection should accompany AI use for a holistic approach. Best practices for feedback implementation include the following:

• Teachers should adopt AI feedback as a starting point for deeper reflection.

• Teachers should combine AI-generated feedback with peer and mentor observations for a well-rounded professional development experience.

• Teacher educators should ensure that teachers understand AI-generated analytics and avoid over-reliance on automated assessments.

7.2.4 AI-Powered Coaching and Simulation

AI-driven coaching and simulations provide interactive, immersive training experiences that allow teachers to practice new skills in low-risk environments (see Tables 18 and 19).

AI-based coaching and simulations offer realistic teaching scenarios. They allow teachers to rehearse classroom management skills, improve their articulation, and fine-tune lesson delivery safely. These tools work especially well for teacher educators who are creating scenario-based microlearning. However, AI simulations should be combined with human coaching and peer feedback for maximum effectiveness. Best practices for implementation include the following:

• Teachers should complete structured reflection activities after AI-powered coaching sessions.

• Teachers must experiment with AI simulations and compare strategies before applying them in classrooms.

• Teacher educators should use AI-driven coaching tools alongside human mentorship to ensure a balanced approach to skill development.

7.3.1 Institutional AI Policies for TPD

Institutions must provide explicit instructions for the ethical pedagogical and professional use of AI in teacher training (Moorhouse et al., Reference Moorhouse, Yeo and Wan2023). Policies on AI ethics should require that AI-produced content is pedagogically sound, transparent, and unbiased. Teachers could be required to note when they have used AI tools to generate materials for their professional learning or training exercises. Institutions should also incorporate privacy measures to protect teacher and student data when AI tools are used in professional development programmes. Additionally, institutional policies should include mandatory AI training for teachers and teacher educators, consisting of hands-on workshops on AI bias and responsible AI integration.

Although AI can enhance professional development activities such as lesson planning, assessment, and feedback training, institutional policies must also highlight teacher autonomy in decision making, ensuring that AI remains a pedagogical support tool rather than a substitute for professional judgment. For example, an institution could implement an AI ethics and pedagogy policy to ensure that AI recommendations for TPD are used to supplement, rather than replace, teachers’ critical thinking and expertise. By establishing institutional AI policies focused on professional development, schools and universities can create a structured, responsible approach to AI-enhanced TPD that balances innovation with ethical considerations.

7.3.2 AI-Powered Peer-Learning Networks

AI technologies have made peer learning and data-driven professional development more manageable. AI platforms can complement traditional peer-learning communities, including professional learning networks and communities, by offering real-time analytics, discussion analysis, and personalised recommendations. AI can support peer-learning networks by providing

• curated discussions and insights. AI can search through professional discussions (e.g. Quora Poe AI and LinkedIn Learning’s AI Coach) to condense main ideas, find popular teaching topics, and recommend relevant research.

• pairings. AI can pair teachers who have the same professional growth goals, enabling the formation of data-driven and adaptive peer groups.

• feedback and coaching. AI suggestions can help teacher educators generate feedback for their peers. This allows teachers to receive individualised recommendations based on their teaching reflections and lesson recordings. In addition, AI can provide individualised feedback to teachers based on their teaching reflections and lesson recordings.

For example, a school district could employ an AI-enhanced peer-learning platform that allows (a) AI to review teacher discussions on professional forums, summarising key information; (b) AI to link teachers with peers facing similar difficulties to foster targeted peer mentoring (e.g. exchanging lesson reflections); and (c) AI-powered feedback systems to assist instructors in improving their instructional methods before sharing them with other teachers. Collaboration would become data-driven and personalised through AI-powered peer-learning networks.

7.3.3 AI-Assisted Reflective Teaching Portfolios

Reflective teaching is essential to professional development; it challenges teachers to think about how they teach. Teachers can use AI to facilitate the creation of reflective teaching portfolios by

• automating lesson analysis from AI-generated classroom transcripts to reflect on their classroom instruction, student engagement, and instructional strategies.

• using Grammarly to enhance reflection journals or TeachFX to monitor speaking time.

• monitoring long-term professional development growth to check whether and how lessons have improved over time and identify areas needing additional development.

For example, a teacher could use an AI-supported reflective portfolio that transcribes lesson recordings. These transcriptions would allow the teacher to reflect on how the students engage with each other. Over time, AI will identify patterns within the teacher’s reflections and provide targeted teaching improvement suggestions. Teachers using AI in reflective teaching portfolios will be able to monitor their progress and receive constructive feedback to enhance their teaching techniques.

8.1.1 Pedagogical Gaps: The Role of AI in Learning and Cognitive Development

Perhaps the most critical question in AI-assisted language learning is how GenAI will affect cognitive development, critical thinking, and knowledge retention. Although AI chatbots and AI-powered tutors are becoming increasingly popular and providing immediate feedback and personalised learning, empirical research addressing the long-term implications for academic performance, language acquisition, and intellectual growth is still in the early stages (e.g. Du & Daniel, Reference Du and Daniel2024; Kamalov et al., Reference Kamalov, Calonge and Gurrib2023).

Several pedagogical questions remain unanswered:

1. 1. Learning Efficacy: To what extent is GenAI-enhanced learning more effective than traditional instruction at promoting depth of understanding rather than surface-level memorisation? Studies show that AI adaptive learning platforms such as Duolingo and Grammarly enhance engagement, but their capacity to generate higher-order thinking competencies is being debated (Guan et al., Reference Guan, Zhang and Gu2025; Xu & Liu, Reference Xu and Liu2025).

2. 2. Teacher-AI Interaction: What is the right balance between instruction by AI and human-led teaching? Some teachers are concerned that depending too much on AI would undermine the value of teachers, while others believe that AI should be employed as an augmentation tool and not a substitute (Tan et al., Reference Tan, Cheng and Ling2025; Zhai et al., Reference Zhai, Wibowo and Li2024).

3. 3. Personalisation vs. Standardisation: How do different learning preferences and educational backgrounds interact with AI-generated feedback? Studies show that AI can customise learning materials for teachers and differentiate them for students. However, teachers are still concerned that AI cannot meet students’ diverse educational needs, especially in multicultural classrooms (Kohnke et al., Reference Kohnke, Zou and Xie2025b; Walter, Reference Walter2024).

4. 4. Fostering Creativity: Can GenAI stimulate original thinking and problem-solving skills, or does it primarily automate learning processes? AI models generate responses from existing knowledge and training data, but their ability to foster creativity remains uncertain (Ivcevic & Grandinetti, Reference Ivcevic and Grandinetti2024; Runco, Reference Runco2023). Further research is needed to explore how AI can enable creativity rather than merely automate learning processes.

8.1.2 Ethical and Bias-Related Gaps: Fairness, Transparency, and Plagiarism

The ethical implications of GenAI in language education present another important challenge (Giannakos et al., Reference Giannakos, Azevedo and Brusilosky2024; Hockly, Reference Hockly2023). AI models are trained on vast amounts of data that may be biased, which leads to questions of fairness, transparency, and equity.

Key ethical concerns include:

1. 1. Bias in AI-Generated Content: Research indicates that AI models tend to reflect and amplify the biases present in their training data (Ferrara, Reference Ferrara2023). This can result in cultural, gender, and linguistic bias in learning content, adversely affecting specific student groups. For example, studies have indicated that large language models are sometimes biased towards Western-centric perspectives, which might make them unsuitable for multicultural environments.

2. 2. Plagiarism and Academic Integrity: With AI-generated content becoming more sophisticated, it is becoming increasingly difficult to distinguish between human and AI-generated work (Kizilcec et al., Reference Kizilcec, Huber and Papanastasiou2024). This raises questions of academic integrity: Should students be allowed to use AI-generated content in their work? If so, how do teachers know they are learning and not simply having AI do the job?

3. 3. Understanding AI Generated Content: Many students and teachers may lack the knowledge to evaluate critically how AI models generate responses (Kasneci et al., Reference Kasneci, Sessler and Küchemann2023). Without a clear understanding of AI’s limitations, there is a risk that they will accept and disseminate misinformation as factual knowledge. Establishing guidelines on evaluating and using AI-generated content is essential to maintaining trust within the learning ecosystem.

8.1.3 Technological and Accessibility Gaps: Bridging the Digital Divide

Making AI-driven education tools available remains a key challenge, particularly in underprivileged regions. Some students have limited access to computers and the internet, resulting in disparities in the use of AI tools in education.

Important questions include:

1. 1. Digital Divide: How can the availability of GenAI learning tools be enhanced for students in low-income regions? Most AI systems need high-speed internet access and costly hardware, making them inaccessible to disadvantaged communities (Johnston & Davis, Reference Johnson and Davis2024).

2. 2. Language Limitations: AI models have been predominantly trained in English and other widely spoken languages, resulting in lower accuracy for less popular languages. These limitations negatively impact the performance of AI-supported educational software for speakers of non-dominant languages (Du & Daniel, Reference Du and Daniel2024).

3. 3. Sustainability and Energy Consumption: Large language models require enormous computational powers, making their environmental impact a cause for concern (Ren et al., Reference Ren, Tomlinson and Black2024). Future research should explore how to enable AI to operate efficiently with low energy consumption. Bridging these gaps requires collaboration among policymakers, educators, and AI developers to render AI tools effective but also ethical, inclusive, and sustainable.

8.3.1 AI-Driven Microlearning Ecosystems: The End of Traditional Language Courses

The traditional model of second language learning, characterised by scheduled, time-consuming, and structured courses, may be increasingly replaced by AI-driven microlearning ecosystems that dynamically adapt to the requirements of each learner in real time. Instead of asking learners to set aside time for language learning, AI will naturally incorporate language learning into daily life, enabling passive and contextual learning throughout the day.

Next-generation AI-powered microlearning platforms will monitor learners’ digital behaviour, speech patterns, and biometric responses to identify the best moments for learning. AI tutors will integrate microlearning exercises imperceptibly into online activities, proposing alternative sentence formulations when users are composing emails, providing instant feedback on pronunciation during informal conversations, or embedding interactive language lessons into social media websites, news stories, and entertainment. AI assistants will also initiate spontaneous voice interactions, conversing briefly with students while they perform daily tasks such as cooking, travelling, or shopping. AI-driven microlearning events will be personalised and non-intrusive, making second language acquisition as natural as developing fluency in a native language.

Smart glasses, earbuds, and neuro-adaptive interfaces will simplify microlearning delivery as wearable AI develops. AIs will know when students are mentally fatigued or most receptive to absorbing new knowledge and adjust lesson formats accordingly. Instead of overloading students with predetermined coursework, AIs will personalise their exposure to language, spacing out new vocabulary, grammar, and pronunciation practice for maximum long-term retention. It is likely that in 10–20 years, second language education will no longer be confined to formal lessons. Instead, AIs will serve as an unobtrusive linguistic collaborator, assisting learners in achieving fluency while they remain unaware that they are studying.

8.3.2 Multimodal AI and the Rise of Immersive Microlearning

Historically, language education has focused on reading, writing, listening, and speaking as separate skills (Canale & Swain, Reference Canale and Swain1980; Hinkel, Reference Hinkel2006); however, emerging AI systems are set to facilitate multimodal learning experiences that simultaneously engage all sensory modalities. Rather than having learners passively consume language lessons, they will be actively involved in AI-generated contexts that enable them to experience language within relevant situations.

Augmented reality, virtual reality, and mixed reality will be at the heart of this transformation. Learners with AI-enabled augmented reality glasses will receive real-time translations, grammar prompts, and pronunciation guides superimposed on their surroundings. For example, when walking around a foreign city, augmented reality will provide learners with instant vocabulary suggestions. This will enable them to absorb new words contextually rather than through rote memorisation. AI-generated holographic tutors will engage in discussions with students in real time, adjusting speech patterns and difficulty levels based on students’ facial expressions, speech, and comprehension.

AI will introduce tactile and gestural language learning methods as well as visual and audio learning. Haptic feedback technology will allow learners to feel the shape of written characters using wearable technology, reinforcing the connection between motor memory and linguistic retention. Gesture recognition technologies based on AI will enable sign language students to rehearse and refine gestures with real-time feedback, making language learning more embodied and interactive. Second language acquisition in ten or twenty years will be an embodied, sensory-intensive process in which the language is not only listened to and read but also sensed through artificially created environments.

8.3.3 AI-Powered Conversational Agents and Adaptive Storytelling

The most significant challenge for second-language learners is achieving conversational fluency in real life, as students generally lack access to native speakers’ authentic social interactions. Future AI-powered conversational agents driven by AI will eliminate this barrier, offering students natural-sounding, emotionally attuned dialogue partners that dynamically adjust to each student’s progress.

In contrast to today’s chatbots, future AI conversation partners will be capable of unscripted, spontaneous interaction, and students will enjoy real-world conversations. AI will analyse intonation, speech rhythms, and emotional cues and adjust responses to create a natural and engaging conversational experience. As AI advances, it will be able to replicate regional accents, cultural nuances, and humour, allowing students to acquire linguistic skills and cultural sensitivity.

AI-driven storytelling will also take second-language immersion to the next level. Instead of passively reading textbooks, students will become part of AI-generated stories, in which they will need to use the target language to clear hurdles, solve mysteries, or untangle complex situations. Whether in a simulated historical environment, a business negotiation, or a sci-fi quest in some future world, students will be active participants in an AI-generated linguistic universe in which every dialogue and decision affects the unfolding narrative. Adaptive storytelling with AI may replace language drills in twenty years, and second language acquisition will become an interactive, immersive, and highly engaging experience.

8.3.4 Ethical AI, Global Language Preservation, and the Future of Multilingualism

As AI-driven language learning becomes more sophisticated, it will be essential to ensure ethical AI development and linguistic diversity. AI must reach beyond dominant global languages to preserve endangered languages and dialects so that all linguistic communities benefit equally from AI-powered learning.

Future AI models must be trained on diverse linguistic datasets to treat underrepresented languages and widely spoken languages equally. AI-powered language preservation initiatives will digitally archive and teach endangered languages, ensuring that linguistic diversity is not eliminated in the era of AI communication. In the future, AI is likely not only to teach languages but also to act as a global linguistic equaliser, making multilingualism a common human ability instead of a privilege of education or geography. Language learning will no longer be an educational pursuit; instead, it will be an effortless, AI-driven aspect of daily life that will transform how we engage with language and communication.