Next Article in Journal
e-Learning Challenges in STEM Education
Previous Article in Journal
Are Adolescents with a Wider Vocabulary Faster at Inference Making During Reading? Evidence from Self-Paced Reading
Previous Article in Special Issue
Enhancing Academic Writing in a Linguistics Course with Generative AI: An Empirical Study in a Higher Education Institution in Hong Kong
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Exploring Autonomy in the AI Wilderness: Learner Challenges and Choices

Languages and Cultures Programme, School of Arts, University of Otago, 95 Albany Street, Dunedin 9016, New Zealand
Educ. Sci. 2024, 14(12), 1369; https://doi.org/10.3390/educsci14121369
Submission received: 19 November 2024 / Revised: 11 December 2024 / Accepted: 11 December 2024 / Published: 13 December 2024

Abstract

:
The emergence of Generative Artificial Intelligence (GenAI) raises critical questions about learner autonomy and agency. This exploratory case study examines how four university-level German language learners with diverse backgrounds developed autonomy in their learning process through engagement with AI tools. The study was conducted in early 2023 when most learners were first discovering ChatGPT’s potential for language learning. Data were collected through reflective journals, digital portfolios, and interviews during a semester-long course that scaffolded self-directed learning with AI integration. The findings reveal emerging patterns of shared agency between learners and AI tools. Learners developed distinct strategies for AI integration based on their language learning backgrounds, with heritage speakers focusing on accuracy improvement while classroom learners emphasized communication practice. Cross-case analyses identified key dimensions of autonomy development: a critical evaluation of AI output, evolving learner–AI relationships, maintaining and developing a second language (L2) voice, and the strategic integration of AI tools while preserving learner agency. These patterns suggest that autonomy in AI-mediated environments manifests through learners’ capacity to engage productively with AI while maintaining critical awareness and personal agency in their learning process.

1. Introduction

Learner autonomy, a concept central to language education since the 1980s, is concerned with the learner’s ability to take control of their own learning process. This involves making decisions about learning objectives, selecting appropriate resources, and evaluating progress. Autonomy has long been seen as a critical factor in successful language learning, as it enables learners to become self-directed, reflective, and proactive in their language development.
The sudden accessibility of Generative Artificial Intelligence (GenAI) tools has added a new dimension to this concept of autonomy. On the one hand, GenAI provides learners with access to an omnipresent study companion that assists them in a wide range of language tasks, ranging from translations and text production to interactive language practice. On the other hand, it raises important questions about the extent to which learners maintain control over their learning when such powerful tools are at their disposal. This dynamic interaction between learners and GenAI has created a need for research that explores how autonomy is exercised, supported, or potentially undermined in AI-mediated learning environments.
This study was conducted in early 2023, a time when many language learners were just beginning to discover and experiment with ChatGPT. The participants in this study, four language students enrolled in a university German course, were introduced to a range of digital tools as part of a curriculum designed to foster learner autonomy. Over the course of the semester, they engaged in a four-week learning challenge, working on self-chosen topics while documenting their progress using conventional text-based and digital resources and AI tools.
While research on GenAI in language learning has expanded rapidly since the release of ChatGPT [1], there is still little understanding of how these tools impact the development of learner autonomy. This study examines the experiences of students who were at the forefront of AI adoption in language education. It aims to explore how these learners interacted with various AI tools, including ChatGPT, how these interactions influenced their autonomous learning behaviors, and what implications this has for the future of language education.
This paper presents an exploratory case study examining how four university-level German language learners with diverse backgrounds engaged with AI tools in their language learning. The literature review provides an overview of learner autonomy and examines how the emergence of GenAI has created new opportunities and challenges for autonomous learning. The Methodology Section outlines the research design, detailing a three-phase course structure that supported students in developing autonomous learning practices. The findings are presented through detailed learner profiles, illustrating how heritage speakers and classroom learners differently approached AI integration in pursuit of their self-chosen learning goals. The Discussion Section analyzes cross-case themes, examining how learners developed and maintained agency while engaging with AI tools. This paper concludes by considering the implications of these findings for understanding autonomy in AI-mediated language learning environments.

2. Autonomy in Language Learning

2.1. Foundations of Learner Autonomy

Learner autonomy, a concept central to language education since the 1980s, has undergone major developments in both theoretical understanding and practical application. When Holec [2] introduced learner autonomy to the field of language education in 1981, he defined it as “the ability to take charge of one’s own learning” (p. 3). This foundational definition emphasizes learner agency, enabling students to set objectives, select learning methods, monitor their progress, and evaluate their outcomes independently.
Early definitions focused primarily on the individual learner’s responsibility, such as Dickinson [3], who viewed autonomy as “the situation in which the learner is totally responsible for all the decisions concerned with his learning and the implementation of those decisions” (p. 11). The understanding of autonomy expanded considerably during the 1990s. Little [4] defined it as “the capacity for detachment, critical reflection, decision making and independent action” (p. 4). He later emphasized that autonomy develops through interdependence rather than independence [5], highlighting that autonomous learning is not about learning in isolation, but about developing the capacity to make informed decisions about one’s learning through critical reflection and self-awareness within social contexts.
The development of self-regulation and metacognition is closely tied to the concept of learner autonomy. Self-regulation refers to the ability to manage one’s learning process, including setting goals, monitoring progress, and adapting strategies as needed [6]. Metacognition, on the other hand, involves the awareness and understanding of one’s own thought processes and learning strategies [7]. Research has consistently shown that self-regulation and metacognitive skills are significant predictors of autonomous language learning behaviors [8,9]. Learners with higher levels of self-regulation and metacognition demonstrate greater engagement with learning resources, more effective use of strategies, and better overall language learning outcomes [10]. As learners become more autonomous, they take greater responsibility for their learning, which requires them to develop and apply self-regulatory and metacognitive strategies [4]. Conversely, engaging in autonomous learning activities can help learners develop these skills, creating a reciprocal relationship between autonomy, self-regulation, and metacognition [11].

2.2. Autonomy in Digital Environments

The development of digital technologies in the early 21st century marked another shift in understanding autonomy. Benson [11] argued that digital environments offered unprecedented opportunities for autonomous learning by enabling learners to access resources, engage in self-directed practice, and interact with others beyond the traditional classroom. Murray [12] emphasized the importance of considering autonomy within new “spaces” for learning, recognizing that different digital environments offer different resources, forms of organization, and levels of situational freedom.
These digital spaces were further conceptualized as the “digital wilds”, referring to the vast array of informal, unstructured learning resources and interactions available online, such as social media, forums, and authentic materials [13]. Little and Thorne [14] introduced the concept of “structured unpredictability” to describe how effective autonomous learning in these digital environments requires a balance between institutional structure and the unpredictable, open-ended nature of learning opportunities. They extend this metaphor through the concept of “rewilding”, which describes the process of bringing these non-pedagogical resources and experiences into the classroom context, where learners engage with authentic language use and real-world communication within the framework of classroom activities and goals.
Digital technologies have fundamentally transformed how learners engage with language learning resources and opportunities. As Lai [15] points out, technology’s relationship with autonomy is inherently dialectical—the same features that enable autonomous learning can also constrain it. On the one hand, digital tools and platforms offer a wide range of affordances that can support autonomous learning. For example, online dictionaries, grammar checkers, and machine translation apps can help learners independently resolve language-related questions, while language exchange platforms and social media can provide opportunities for authentic interaction with L2 speakers. These tools and platforms can empower learners to take control of their learning process, set their own goals, and pursue their interests. On the other hand, the very same tools and platforms can also constrain learner autonomy in various ways. The misuse or uncritical use of tools such as machine translators or spell checkers can hinder learners’ language development. When learners rely too heavily on these tools without critically evaluating the outputs or engaging in their own problem-solving processes, they may miss opportunities to develop their language skills and deepen their understanding of the language system [16]. Another challenge posed by the digital language learning environment is information overload, particularly with the increasing prevalence of multimodal resources [17]. The sheer volume of language learning content available online, presented in various formats such as text, audio, video, and interactive media, can be overwhelming for learners. Without well-developed strategies to manage and critically evaluate these resources, learners may struggle to effectively integrate them into their learning process and may feel constrained in their ability to make informed decisions about their learning [15].
To benefit from the learning opportunities that can be found in the digital wilds, learners need to develop a range of skills and strategies. This includes the ability to critically evaluate the affordances and constraints of different digital tools and platforms, set clear learning goals, monitor their progress, and adapt their strategies in response to the challenges and opportunities they encounter [15]. Learners must also cultivate critical digital literacy skills, enabling them to assess the quality, relevance, and credibility of online resources and interactions for their learning purposes. Furthermore, learners need to develop metacognitive awareness in the digital context, allowing them to make informed decisions about the selection and integration of digital resources in their learning process [18]. This involves the ability to use digital tools purposefully and critically reflect on their impact on one’s learning process.
Recent research by Lai et al. [19] confirms these challenges, arguing that while digital tools can enhance autonomy through personalized learning experiences and increased resource access, they simultaneously require learners to develop skills in managing and critically engaging with these tools. This complex relationship between technology and autonomy now faces an even greater transformation with the emergence of AI tools, which introduce unprecedented levels of interaction and agency in the learning process.

2.3. Autonomy in the Age of AI

The popularization of AI since the launch of ChatGPT in late 2022 has added yet another dimension to learner autonomy. AI chatbots like ChatGPT provide language learners with support for vocabulary, grammar, and writing queries, offering assistance that can enhance self-directed practice. However, as Mollick [20] puts it, these AIs are akin to unreliable interns who, while trying to be helpful, may produce responses that are inaccurate or misleading.
While AI offers opportunities for learners to set individual goals, refine strategies, and explore language independently, it also necessitates that learners develop critical skills to evaluate and adapt AI-generated content meaningfully. Furthermore, as Chapelle et al. [21] warn, AI is a non-neutral mediator, with embedded algorithms that can shape learning interactions in unpredictable ways. Learners thus find themselves not only interacting with a tool but negotiating with a system that can subtly influence their learning paths.
This dynamic interaction brings forth multiple perspectives on agency and autonomy in AI-mediated learning. Godwin-Jones [22] draws on the concept of distributed agency to understand these new dynamics in language learning. Originally emerging from sociomaterial perspectives in science and technology studies, the concept of distributed agency challenges traditional notions of human autonomy by recognizing agency as shared between human and non-human actors. This theoretical approach rejects the idea of technology as merely a passive tool, instead examining how agency arises through the interaction between humans and technologies in specific contexts. In AI-mediated environments, this agency emerges through iterative interactions between learners and AI systems. As Godwin-Jones [22] explains, while both humans and AI influence the learning process, this relationship is reciprocal but not necessarily symmetrical—the human remains “in the loop”, maintaining primary control through conscious choices about how and when to engage AI support. In language learning environments, this means recognizing that both learners and AI tools actively contribute to and shape the learning process through their interactions, creating a complex relationship where learners must develop strategies to maintain meaningful control of their learning while benefiting from AI’s capabilities.
Related perspectives from different fields enrich this understanding: Mollick [20] characterizes the interaction as “co-intelligence”—emphasizing how AI can complement human capabilities while remaining fundamentally different from them, while Thorne [23] frames these dynamics within a broader pattern of human–tool co-evolution, highlighting how technological relationships have consistently shaped cognitive and communicative practices.
The significance of these changes is highlighted by Kalantzis and Cope [24], who argue that current AI developments represent a transformation as significant as the printing press, fundamentally changing how humans produce and engage with meaning. Their concept of “cyber-social literacy learning” emphasizes the reciprocal, feedback-driven nature of human–AI collaboration. In this collaboration, learners must balance the support provided by AI with their own independent judgment, making deliberate choices to ensure that AI enhances rather than limits their autonomy.
Achieving this balance requires learners to develop a metacognitive awareness that enables them to recognize the limitations of AI, critically evaluate its output, and adapt content to meet their specific goals. The presence of AI in language education thus raises fundamental questions about how learners can engage with AI while maintaining the reflective, independent decision making essential to genuine autonomy.

3. Methodology

Following Yin’s approach to exploratory case study research [25], this study investigates how ChatGPT influences and interacts with language learners’ autonomy within an intermediate-level German course. While another analysis of these data explores the ethical dimensions of AI integration [26], this study focuses specifically on learner autonomy. The exploratory case study design allows for a flexible, in-depth investigation of this under-researched phenomenon in a real-world setting, examining learners’ self-directed behaviors, strategies, and reflections as they integrate AI tools into their language learning processes. As an exploratory study, the focus is on understanding specific learner experiences rather than producing generalizable results. The study aims to establish a conceptual framework for understanding autonomy in AI-assisted learning contexts, providing a foundation for future research and broader applications.

3.1. Research Questions

The study was guided by one primary research question:
  • How do language learners with different backgrounds experience and develop autonomy within a course that supports self-directed learning and incorporates ChatGPT?
Two secondary questions helped focus the investigation on the relationship between learners’ backgrounds and their AI engagement:
  • How do learners develop and adapt strategies for self-directed learning in an AI-mediated environment?
  • How do learners develop agency while integrating AI tools into their learning process?

3.2. Course Structure

The course was structured around three phases, based on Ohashi’s [27] and Steel’s [28] learning challenge models, designed to build learner autonomy incrementally through self-reflection, goal-setting, and adaptive AI use. Each phase introduced new strategies and resources to support students’ autonomy, emphasizing both self-directed learning and AI engagement in the _target language, German.

3.2.1. Phase 1: Self-Discovery and Goal Orientation (Weeks 1–6)

The first phase focused on fostering self-awareness and strategic planning. Through a writing workshop, students were introduced to a staged writing process for their weekly journal assignments. For each entry, they first completed drafts independently and then used AI tools (DeepLWrite, LanguageTool, ChatGPT) to revise their writing. While drafts and revisions were written in German, students could choose to write their reflections on their learning process in either German or English, their first language, to facilitate critical reflection and self-expression.
Students completed five journal entries:
  • Language Learning Biography: Students reflected on past language learning experiences, motivations, and influences to establish a foundation for strategic development.
  • Self-Assessment of Language Skills: Using the CEFR scale, students assessed their proficiency across different language skills, identifying strengths and areas for improvement.
  • Strategy Inventory for Language Learning (SILL): Through an adapted SILL, students analyzed their learning strategies, identified gaps, and considered the potential of AI to support their development.
  • Experiences with AI Tools: Students documented and evaluated their initial experiences with AI, considering its effectiveness, limitations, and suitability for language learning.
  • Planning for the Language Challenge: Drawing on their previous reflections, students created personalized learning plans, outlining language goals, strategies, and AI use. These plans were refined through peer discussion.
The online curation tool Wakelet (wakelet.com, accessed on 11 December 24) served as the Learning Management System throughout this phase, providing access to course resources while familiarizing students with the digital curation skills needed for their portfolios in Phase 2.

3.2.2. Phase 2: Self-Study Period and Language Challenge (Weeks 7–11)

Building on the foundational work of Phase 1, students undertook their four-week Language Challenge, implementing their personalized learning plans. Each week, they updated their Wakelet portfolios with documents, screenshots, photos, and reflections, maintaining the revision process established in Phase 1. These updates documented their weekly activities, AI interactions, and progress toward their language learning goals.

3.2.3. Phase 3: Weeks 12–13—Presentations and Final Reflections

At the end of the Language Challenge, students shared their learning experiences in PechaKucha presentations, describing their projects, successes, and challenges. Their final journal entry provided a comprehensive reflection on their learning journey: students revisited their initial motivations for learning German, compared their current proficiency with their initial self-assessment, and evaluated their strategy development throughout the semester. They analyzed how AI tools supported or challenged their learning processes and assessed how the learning challenge influenced their autonomy, confidence, and self-regulation skills.

3.3. Participants

Four participants were purposefully selected from the German course to represent a diverse range of language-learning backgrounds, academic disciplines, and attitudes toward AI. Their language backgrounds ranged from heritage speakers to learners who began studying German at university. Academically, the participants majored in disciplines as varied as Anthropology and Global Studies, Mathematics, German and Religious Studies, and Global Studies and Politics. This diversity provided a range of perspectives on autonomy development in AI-mediated contexts, with the aim of illustrating different pathways to autonomous learning. The pseudonyms used in this study—Sienna, Gina, Cora, and Flora—match those in the other study [26] to ensure consistency across analyses.

3.4. Data Collection

Data were collected using multiple methods to provide a holistic view of participants’ learning journeys:
  • Reflective Journals: Six journal entries provided a detailed account of each learner’s reflections, strategies, and responses to AI tool use throughout the course.
  • Wakelet Portfolios: The digital portfolios provided both narrative and visual records of students’ engagement with AI tools and their strategy adaptations throughout the Language Challenge.
  • Semi-Structured Interviews: Two interviews were conducted at the end of the semester: a group interview with five students, including three of the participants, and an individual interview with one participant. Each interview lasted approximately 60 min, focusing on reflections about autonomy, AI engagement, and strategic use of ChatGPT in learning processes. Interviews were recorded and transcribed using Otter.ai.
The data collected were in both German and English due to the course structure and participants’ language proficiencies. Journal entries and Wakelet portfolio entries were primarily written in German, as part of the course requirement to practice _target language writing. The final interviews were conducted in English to allow participants to discuss their experiences with AI integration and learning strategies in depth without language constraints. All quotes from German data are presented as English translations in this paper, marked by [tr] following the quote.

3.5. Data Analysis

Consistent with the exploratory case study design [25], data were analyzed in two stages to understand how learners developed autonomy in an AI-mediated environment:
  • For the individual case analysis, each participant’s data set (journal entries, Wakelet portfolios, and interview transcripts) was analyzed chronologically to trace their development of autonomy and engagement with AI tools throughout the course. This process enabled the construction of detailed narrative profiles that captured each learner’s unique journey, challenges, and growth.
  • For the cross-case analysis, the individual profiles were then compared to identify patterns in how learners engaged with AI tools and developed autonomy. This analysis revealed key dimensions of AI-mediated autonomy, including foundational understanding, practical engagement, and critical evaluation.

3.6. Ethical Considerations

This study received ethical approval from the author’s university ethics committee. Participants provided written informed consent after receiving information about the study’s aims, data collection methods, and their right to withdraw. To ensure confidentiality, participants’ identities were protected through pseudonyms, and all data were stored securely. Participants were offered the opportunity to review their interview transcripts and case profiles before analysis. The study’s trustworthiness was enhanced through multiple data sources (reflective journals, Wakelet portfolios, and interviews). A colleague experienced in qualitative research reviewed the case analyses by cross-checking the participant data to identify consistent themes and patterns in the development of learner autonomy. This iterative process involved comparing findings across cases to ensure the analysis accurately captured both shared and individual aspects of how learners engaged with AI tools in their language learning journey.

4. Findings

4.1. Profiles

The four participants represent different pathways to German language learning: two heritage speakers—one with informal family exposure and another one who had formal German schooling; one who began learning German at university level and had prior language learning experience in other languages; and one who had learned German through classroom instruction from high school through university. These different backgrounds not only shaped their learning needs and goals but also influenced their approaches to incorporating AI tools into their language learning strategies. Their profiles illustrate how learners with different language learning histories develop unique strategies for maintaining autonomy while integrating AI support.

4.1.1. Sienna’s Profile

Language Learning Background and Self-Assessment

Sienna’s German language development began informally through her Austrian family background. Her mother created an immersive home environment through German conversations, reading, films, and songs. In her self-assessment, Sienna recognized the implications of this informal acquisition: “I think I can pronounce well, but I find grammar difficult. I need to make my German foundation stronger” [tr]. Despite confident pronunciation and natural communication abilities, she identified accuracy as her key area for improvement.

The Challenge: Improving Accuracy with Games

For her challenge, Sienna decided to organize weekly game events for all students in the department. Her goal was to create a social environment in which she could speak German with a purpose. “I decided that I needed to focus more on accuracy than vocabulary” [tr]. After an initial session trying different games, she decided to continue with Rommé, as participants particularly enjoyed it. This consistent game choice allowed players to become comfortable with game-specific vocabulary and expressions.
Sienna integrated AI tools to prepare for the sessions and review the language production during the session. Before each session, she used ChatGPT to create vocabulary lists specific to card games, deliberately choosing English for these prompts to ensure the lists were formatted appropriately for learners. During sessions, she recorded the conversations, later creating transcripts which she analyzed with ChatGPT to identify grammatical patterns and errors. Her prompts evolved to _target specific aspects of her language use: “Can you tell me why these corrections were made?” [tr] and “What are the main grammatical problems in this text?” [tr].
For her written reflections, she found DeepL Write more effective for straightforward corrections, while using ChatGPT for more detailed analysis and explanations of grammatical patterns. Her alternation between German and English prompts illustrates her ability to adapt the use of AI tools to serve different learning needs.

Reflections on Development

By the end of the challenge, Sienna had transformed what initially seemed an overwhelming task—improving accuracy—into manageable, focused learning opportunities. Her final reflection captured this breakthrough: “I believe I made the most progress in my German this semester. I knew my accuracy wasn’t good, but I didn’t know how I could improve it because it was a huge task. This challenge specifically identified what I need to work on the most” tr].
As a heritage speaker, Sienna’s approach demonstrated how AI tools could be particularly useful for addressing the specific needs of learners with strong communicative abilities but have gaps in formal language knowledge. Her differentiated use of AI tools—DeepL Write for corrections, ChatGPT for analysis and explanations—along with her strategic language choice in prompting, shows how AI tools can be effectively integrated into self-directed language learning when approached with clear purpose and reflection. The social component of her challenge provided speaking opportunities and created a supportive environment for other learners, who benefitted from these sessions in multiple ways. Beyond language practice, the game sessions fostered camaraderie and a sense of community among participants while engaging them in purposeful language use, nicely illustrating Little’s [4] point that autonomy develops through interdependence rather than independence.

4.1.2. Gina’s Profile

Language Learning Background and Self-Assessment

Gina had substantial prior language learning experience before starting her German studies at university. She studied Māori and Spanish at high school and university and also spent in Spain as an au pair, where she attended a language school in Madrid. She enjoyed reading to improve her language: “My favorite method of language practice is reading books” [tr]. She also engaged with various digital resources including videos, social media, and computer games. This preference for reading led directly to her choice of topic for the learning challenge.

The Challenge: Extensive Reading with Harry Potter

For her learning challenge, Gina chose to read Harry Potter und der Stein der Weisen, setting manageable weekly goals: “As a realistic goal, I read one chapter per week. This goal was achievable but still challenging” [tr]. To build her vocabulary, she maintained handwritten lists of unfamiliar words encountered during reading, which she documented through photographs in her portfolio. She then transferred recurring vocabulary to the vocabulary app Memrise for regular practice: “I recorded recurring words in Memrise and practice them” [tr]. She also complemented her reading with audiobook listening for pronunciation and comprehension practice.

Evolution with AI Tools

Gina’s engagement with AI developed from resistance to strategic integration. She began the semester explicitly rejecting ChatGPT: “I don’t want to use ChatGPT because it wants my phone number” [tr]. She initially relied on a combination of conventional and AI tools: Microsoft spellcheck for basic corrections, Collins Dictionary and DeepL for vocabulary, and LanguageTool for final checks. She gradually began experimenting with ChatGPT, finding it particularly useful for understanding idioms and phrases that traditional dictionaries could not adequately explain, yet frustrating when it came to identifying corrections in ChatGPT’s output. She experimented with different prompting strategies. As she explained: “This time I asked ChatGPT to mark the errors in bold. It was very helpful to see the errors” [tr]. However, she maintained critical toward AI outputs, consistently evaluating their usefulness and seeking alternative resources when needed.

Reflections on Development

Gina successfully completed her reading goal, documenting an impressive expansion of vocabulary: “I also wrote 950 new words in my notebook, although not all were unique” [tr]. More significantly, she developed confidence in her ability to continue independent language learning: “Now I believe that I can learn German without classes” [tr]. Her experience demonstrated how an experienced language learner could maintain established learning strategies while incorporating new AI tools to develop rather than replace existing practices.

4.1.3. Cora’s Profile

Language Learning Background

Cora’s journey with German began in high school, where she had initially hoped to study Spanish. However, her experience with German proved positive, largely due to an engaging teacher: “my German teacher was very cool and knew a lot about languages, history, literature, and so on, and I loved the sound of the language” [tr].
Her classroom-based learning had fostered strong written skills, as she noted: “my grammar and writing aren’t so bad” [tr]. However, with limited exposure to German outside formal instruction, she struggled with speaking confidence: “I’m not confident and easily forget words, and words don’t come quickly to mind” [tr].

The Challenge: From Cold War to Conversation

Having experienced mainly formal learning environments, Cora found self-directed learning challenging: “I always find it difficult to improve my German in my own time” [tr]. She initially wanted to focus on the Cold War, but after consulting with her instructor, she changed her topic to address her primary need—developing speaking confidence through real conversations. She arranged to meet with Julia, a German exchange student, creating both opportunity and challenge: “We will only speak in German, and I have never managed that before. I hope we won’t speak any English!” [tr].
To prepare for this conversation, Cora started using ChatGPT, using the AI as a practice partner: “I asked ChatGPT if it could behave like a twenty-year-old girl” [tr]. She used these practice conversations to build confidence and develop vocabulary for specific topics she planned to discuss with Julia, ranging from language learning experiences to life in New Zealand.
In her writing, she found ChatGPT particularly useful for finding alternative ways to express ideas: “I find it very effective to look up synonyms. I like that I can describe something and ChatGPT will give me some words” [tr]. However, she remained critical of its grammar explanations, drawing on her strong foundation from high school education to identify errors. For instance, when ChatGPT incorrectly claimed that the preposition “durch” could take both accusative and dative cases, her prior knowledge allowed her to recognize this mistake. This experience reinforced her awareness of keeping a critical stance toward AI outputs, particularly regarding grammar explanations.

Reflections on Development

Cora’s conversation with Julia proved transformative, lasting nearly three hours and covering unexpected topics. She wrote in her last journal entry: “I believe that I have achieved my SMART goals. If I meet a German person, I would feel confident enough to have a conversation with them” [tr]. Her journey from formal classroom learner to independent language user demonstrated how AI tools could be integrated to support authentic communication goals, while maintaining a critical awareness of their limitations. In particular, her motivation to engage in authentic conversation with a German speaker catalyzed her strategic use of ChatGPT, demonstrating the interplay between motivation and autonomy. Her clear vision for continued development, including pursuit of C1 certification: “I definitely want to continue my German journey. I hope that I can achieve the C1 certificate one day” [tr], suggested a strong foundation for autonomous learning beyond the classroom.

4.1.4. Flora’s Profile

Language Learning Background

Flora grew up in Singapore, with a Singaporean mother and German father. She attended a German European School in Singapore, which additionally supported her German language development. She reflected: “As a child I already had knowledge of German and in school I expanded my knowledge and improved my abilities” [tr]. Unlike Sienna, who acquired German solely through family interactions, Flora’s formal schooling gave her a foundation in written German.
A heritage speaker, Flora was fluent and confident in both spoken and written German. However, she identified listening comprehension as an area needing improvement. This was a long-standing challenge, as she explained: “I always had difficulties with listening tasks in school and sometimes found it hard to understand the entire content of conversations and audio” [tr].

Learning Challenge: Focus on Listening

To address this difficulty, she designed a challenge focused on improving her listening comprehension. She committed to listening to four podcasts per week, selecting content that covered contemporary social and political topics including body positivity, educational inequality, work–life balance, and citizenship. Towards the end, she focused on different German dialects and accents, recognizing her need for exposure to more varied forms of German.
Although initially doubtful about ChatGPT’s usefulness, it became an integral part of her weekly activities. ”I used ChatGPT for small summaries of my podcasts and correction of my texts” [tr]. Her process began with obtaining podcast transcripts through a Chrome extension. She then developed specific techniques for vocabulary development, asking ChatGPT to “pick out the hard ones” and provide “the most important words or like the hardest vocab words in this podcast summary”. When she found the AI providing overly simple vocabulary she already knew, she refined her prompts: “I noticed that it was a bit faulty in the times where. it kept giving me like simple words I already knew”.
For documenting her learning, she began with basic support: “I use DeepL just for like words I forgot… and then … if that didn’t do the trick, I’ll put it in ChatGPT”. She also used ChatGPT to check that she understood the content correctly and to correct her summaries. However, she restrained from using ChatGPT to produce perfect language, emphasizing its role as support rather than primary learning mechanisms: “For my learning challenge, I wanted to use ChatGPT as additional support, rather than relying on ChatGPT for my entire project” [tr].

Development and Final Reflections

By the end of the challenge, Flora reported increased confidence in her listening: “I find that I now have more confidence regarding my listening comprehension and can now understand more content” [tr].
Her approach to AI tools changed from initial skepticism (“I was kind of skeptical of the whole thing… I never used it”), largely due to concerns about academic integrity to purposeful implementation. As a heritage speaker, she found AI particularly useful for addressing specific language learning needs: “I still make very basic mistakes like der, die, das”. ChatGPT provided explanations that were more accessible than traditional textbook presentations of grammar.
Throughout this development, Flora maintained a clear sense of ownership over her learning. She deliberately chose not to aim for perfect accuracy in AI-corrected work, keeping her voice: “if I were to write it without ChatGPT I wouldn’t have it 100% correct. So the same thing, if I were to use it, I wouldn’t want that 100% right, because it’s not my full stuff”.
Her final reflections demonstrated a sophisticated understanding of AI tool integration: “I now have a new perspective on how I can use AI tools as support for my language learning” [tr]. This insight is particularly relevant for her as a heritage speaker, as she discovered ways to use AI tools to address specific language learning needs—like grammar explanations and vocabulary development—that traditional textbooks and formal language classes often fail to meet. Her experience shows how heritage speakers can maintain control over their learning while using AI tools strategically to support their specific language development needs.

4.2. Cross-Case Analyses

4.2.1. Development of Self-Directed Learning Strategies

Initial Approaches to AI Integration

All participants initially expressed a degree of hesitation or surprise when they learned that they would use ChatGPT as a language learning assistant in their German class. Some worried about academic integrity and technical requirements, while others had not heard about it or were not sure how to use it for language learning. However, as the semester progressed, they all developed ways of integrating it into their learning.

Development of Personalized Learning Processes

Clear differences emerged in how heritage speakers and classroom learners approached their learning challenges, particularly in their initial choices and subsequent development of learning strategies. Both heritage speakers used transcripts as a way to analyze and improve their language use. Sienna used ChatGPT to analyze transcripts from the recorded game session to improve her accuracy, and Flora used transcripts of podcasts to support vocabulary development. Their approaches emphasized accuracy improvement and formal language analysis, addressing gaps typical of informal language learners.
The classroom learners approached their challenges through more traditional academic frameworks. Cora’s first learning plan revolved around the Cold War, a typical classroom topic, while Gina identified reading as her preferred learning method based on her previous language learning experiences. Yet, their approach changed over time. Cora’s focus on addressing her real need for conversational practice, organizing a meeting with a German exchange student, and using ChatGPT to prepare for it. Gina, while maintaining her focus on reading, developed an integrated approach combining traditional reading with AI tools, audio support, and vocabulary development. Both adapted their approaches to incorporate more authentic language use while maintaining the systematic learning strategies developed through formal education. This development illustrates how learners could build upon their existing learning frameworks while embracing new tools and approaches.

AI Integration

As participants became more familiar with AI tools, their approach evolved, from viewing AI as a simple correction tool to recognizing its varied potential for language learning support. Each participant developed distinct strategies for different tasks. Sienna used DeepL Write for basic text corrections while reserving ChatGPT for grammatical analysis. Flora developed a methodical process for her podcast work, using DeepL for basic vocabulary lookups, LanguageTool for initial corrections, and ChatGPT for content summaries and more complex language analysis. Gina, despite initial reluctance, came to value ChatGPT particularly for understanding idioms and phrases that traditional dictionaries could not adequately explain, while maintaining her primary reliance on conventional tools like Collins Dictionary for basic vocabulary. Cora used ChatGPT most creatively, using it as a conversation partner for practice while using it analytically to prepare vocabulary and expressions for specific discussion topics. This strategic differentiation among tools and tasks demonstrated developing metacognitive awareness of both their learning needs and the tools’ capabilities.

4.2.2. Developing Agency and Critical Engagement

Critical Evaluation of AI Output

Participants developed approaches to evaluating AI output, moving beyond acceptance or rejection to critical engagement.
Heritage speakers primarily evaluated AI suggestions against their intuitive language knowledge. Daniela assessed grammar corrections by comparing them to her natural communication patterns, selectively incorporating changes that met her language goals while maintaining her characteristic expression. Flora evaluated AI-suggested vocabulary for her podcast work against her existing language competence, actively refining her prompts when ChatGPT provided overly simplistic suggestions that did not match her advanced level.
Classroom learners approached evaluation through their formal language knowledge. Cora’s strong grammatical foundation from high school enabled her to identify and reject incorrect AI grammar explanations, demonstrating how prior knowledge supports critical engagement with AI. Gina’s practice of cross-referencing AI-suggested vocabulary with her reading material further highlights the importance of verification skills in using AI tools effectively. This capacity to critically evaluate AI outputs is an essential element of engaging productively with generative AI in language learning.

Evolving Learner–AI Relationships

The participants’ engagement with ChatGPT evolved over time, revealing both opportunities and potential concerns. This was particularly evident in their prompting practices. Sienna’s experience highlighted a key issue around AI interaction—she noted that “ChatGPT adapted to the way I wrote my questions… eventually figured out what I actually wanted based on my previous answers”. While this adaptability can make the AI seem more helpful, it also demonstrates how AI systems can shape interactions in through their underlying algorithms, potentially influencing learning paths without users being fully aware of this effect.
This dynamic was reflected in how participants used AI for different aspects of their challenges. Some created structured dialogs with AI for specific learning purposes, while others developed systematic processes for using AI in analysis and improvement. While these varied approaches showed learners actively shaping AI interactions to serve their learning needs, they also highlight the importance of developing awareness of AI’s role as a non-neutral mediator in the learning process.

Maintaining and Developing L2 Voice

Learner agency emerged in how participants maintained or developed their L2 voice while using AI support. This manifested differently based on their language learning backgrounds. Heritage speakers seemed to focus on preserving their authentic voice while refining their language use. Flora explicitly balanced authenticity and improvement, choosing not to implement all AI corrections to maintain her natural voice [“I didn’t want everything to be 100% correct… because it’s not my full stuff”]. Similarly, Sienna used AI to analyze rather than reshape her natural German voice, using transcripts of her authentic speech as the basis for improvement. For classroom learners like Cora, AI tools served as scaffolding for developing her L2 voice. Her progression from practice conversations with AI to successful authentic interaction with a German speaker demonstrated how AI support could facilitate the development of an authentic L2 voice rather than supplanting it.
The difference in approaches reflects the different challenges faced by heritage and classroom learners in developing their L2 voice. While heritage speakers worked to refine their existing natural expression, classroom learners used AI to bridge the gap between formal language knowledge and authentic communication. In both cases, participants demonstrated awareness of the need to balance AI support with authentic language development.

Strategic Integration and Shared Agency

All participants developed clear boundaries in their AI relationships, demonstrating an understanding of shared but distinct responsibilities in the learning process. This manifested in structured processes that allowed learners to maintain control while using the AI’s capabilities. Heritage speakers tended to use AI more selectively, focusing on specific aspects of language refinement, while classroom learners integrated AI more systematically into their established learning routines.
The participants’ approaches to maintaining personal agency while using AI support suggest emerging patterns of learner autonomy in AI-mediated environments. Rather than allowing AI to direct their learning, they developed strategic frameworks for AI integration that supported their individual learning goals while preserving their role as primary agents in their language development. Their experiences demonstrate how learners can maintain autonomy while benefiting from AI support, pointing toward new models of learner agency in AI-enhanced language learning environments.

5. Discussion

5.1. Reconceptualising Autonomy

This study captures a critical moment in the evolution of learner autonomy—the integration of generative AI into language learning practices. Conducted in early 2023, the findings illustrate how learners navigated the sudden availability of powerful AI tools, reshaping traditional concepts of autonomous learning. While previous technological developments have already transformed our understanding of autonomy from individual self-direction to more socially and technologically mediated forms of learning [29], GenAI introduces a fundamentally different dynamic—one where the technology actively participates in and shapes the learning process.
The emergence of GenAI has created a new form of “digital wilds” [13], into what could be termed the “AI wilderness”. This AI wilderness represents a vast, uncharted territory of language learning possibilities, where learners can explore, experiment, and create in ways previously unimaginable. GenAI tools offer learners a wide range of language support and resources, from generating authentic text and providing dynamic feedback to engaging in immersive, interactive language experiences. However, traversing this AI wilderness also presents new challenges, as learners must develop the skills to critically evaluate the accuracy and appropriateness of AI-generated content while maintaining a sense of personal agency and creative control in their learning process.
The findings reveal early manifestations of what Godwin-Jones [22] refers to as distributed agency in language learning contexts. While participants initially expressed reservations about AI integration—ranging from technical concerns to questions about academic integrity—their hesitation quickly gave way to curious exploration. As Ranalli [30] notes, learners need to develop “calibrated trust” (p. 14) in AI systems. The data illustrate this development: learners moved from tentative experimentation to purposeful integration of AI tools that supported their individual learning objectives. The data shows evolving patterns of interaction where learners progressively refined their approaches to working with AI tools. Sienna’s observation that “ChatGPT adapted to the way I wrote my questions” points to an emerging awareness of this iterative dynamic, though the full implications of algorithmic adaptations were not yet fully apparent to learners.
The study demonstrates that distributed agency emerges through experience as a dynamic relationship rather than a fixed division of control. Learners developed processes for working with AI tools, strategically selecting and combining them for different purposes—such as using DeepL for basic corrections while reserving ChatGPT for more complex language analysis. These emerging strategies suggest that distributed agency manifests through learners’ active decisions about how and when AI contributes to their learning process.
The language learning background influenced how participants managed this distributed relationship. Heritage speakers, who often used language intuitively, used AI analytical tasks, maintaining their authentic language production. In contrast, classroom learners used AI more extensively for practice and revision. They developed strategies that enhanced rather than replaced their learning process. This differential use of AI tools reflects similar observations in the recent literature about the importance of balanced human–AI interactions in language learning [21,24].
Individual learners demonstrated varied approaches to integrating AI into their learning practices. Some focused primarily on writing support, evaluating and selecting from AI suggestions. Others used AI as a conversation partner for practice, developing prompting strategies to prepare for authentic interactions. Others integrated AI into their existing learning routines, using it alongside traditional resources and practices. These diverse approaches highlight the potential for creativity and innovation in the AI wilderness, as learners discover novel ways to leverage AI tools for language learning and self-expression.
The findings also illustrate the paradoxical relationship dynamic tension between AI and learner autonomy, reflecting the dialectical relationship between technology and autonomy discussed by Lai [15]. While AI tools offer opportunities for learners to set individual goals, refine strategies, and explore language independently, they also require the development of critical skills to meaningfully evaluate and adapt AI-generated content. This paradox highlights the need for learners to develop metacognitive awareness that enables them to recognize the limitations of AI, critically evaluate its output, and adapt content to meet their specific goals.
These patterns of engagement suggest that autonomy in AI-mediated environments requires an expanded understanding of learner capacities. The language learners in this study demonstrated emerging abilities to use AI productively for their learning needs, evaluate AI contributions, and maintain their voice in AI-mediated learning—though their approaches were necessarily exploratory given the novelty of these tools at the time.
Building the concept of distributed agency, as presented by Godwin-Jones (2024) we propose that learner autonomy in the age of AI could be reconceptualized as “the capacity to engage productively with AI while maintaining critical awareness and personal agency”. This definition acknowledges the collaborative potential of AI-mediated learning while emphasizing the learner’s essential role in directing this collaboration towards meaningful language development, recognizing the added complexities introduced by the AI wilderness.

5.2. Implications and Future Directions

The study’s findings have implications for both practice and research. For language education, the course structure proved supportive in helping learners develop approaches to developing autonomy through engaging with AI. The combination of writing workshops, regular reflection through journal entries, and opportunities to share experiences enabled learners to experiment with and critically evaluate their AI use. This structure, beginning with personal language learning histories and self-assessment, progressing through guided exploration of AI tools and preparing for individually designed learning challenges, provided learners with a structured yet flexible framework, supporting them in developing their own approaches to autonomous learning in an AI-mediated environment.
However, the study also reveals areas requiring further investigation as AI becomes more established in language learning contexts. Future research needs to examine the following factors:
  • How learners engage with AI systems that increasingly adapt to user behavior.
  • The long-term impact of AI integration on learner agency and language development.
  • How learner attitudes and practices with AI evolve over time.
  • Ways to support learner autonomy as AI functionalities expand.
  • The role of peer collaboration in developing critical approaches to AI use.
While this study identifies early patterns of distributed agency in AI-mediated language learning, it represents only the beginning of understanding how learner autonomy evolves in response to these powerful new tools. As AI systems become more pervasive, maintaining learner agency while benefiting from AI capabilities is likely to become an increasingly complex challenge that requires ongoing research.

5.3. Limitations and Concluding Reflections

Several limitations should be noted. The small-scale and specific context of this study—four university students in a German language course—suggests the need for broader investigations of how different learner populations negotiate autonomy in AI-mediated environments. This study does not account for variations in language proficiency levels, _target languages, or learner motivation, all of which might influence the development of autonomy. These factors represent important directions for future research, as they could significantly shape the interplay between motivation, agency, and autonomy.
The timing of this study, conducted during the early stages of ChatGPT adoption in early 2023, likely shaped the findings. While participants’ knowledge of their involvement in a study and the novelty of ChatGPT at the time may have influenced their behavior, the range of documented reactions—from skepticism and rejection to curiosity and strategic integration—suggests authentic engagement. If conducted today, the study might yield different results, as learners are likely to have more entrenched views (with some becoming more knowledgeable and others more reluctant), and AI tools have continued to evolve. This temporal aspect should be considered when interpreting the findings. Longitudinal studies would enhance our understanding of how learners’ relationships with AI tools evolve over time and shape their autonomous learning practices.
Nevertheless, this study demonstrates how learner autonomy evolves with technological change. The emergence of distributed agency between learners and AI tools suggests not a diminishment of learner autonomy, but rather its transformation. As language education continues to adapt to AI integration, the findings point to opportunities for reimagining teaching and learning practices. Supporting autonomy in AI-mediated environments requires both guidance and space for individual experimentation, allowing learners to forge their paths through the AI wilderness while maintaining control over their language development. This balance may help shape an educational future where technological affordances enhance rather than constrain learner agency and development.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki, and approved by the University of Otago Human Ethics Committee (Reference 23-022; Approval Date 19 February 2023).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The data are not publicly available due to privacy considerations but may be available from the corresponding author on reasonable request.

Conflicts of Interest

The author declares no conflicts of interest.

References

  1. Law, L. Application of generative artificial intelligence (GenAI) in language teaching and learning: A scoping literature review. Comput. Educ. Open 2024, 6, 100174. [Google Scholar] [CrossRef]
  2. Holec, H. Autonomy in Foreign Language Learning; Pergamon: Oxford, UK, 1981. [Google Scholar]
  3. Dickinson, L. Self-Instruction in Language Learning; Cambridge University Press: Cambridge, UK, 1987. [Google Scholar]
  4. Little, D. Learner Autonomy 1; Definitions, Issues and Problems. Authentik: San Francisco, CA, USA, 1991. [Google Scholar]
  5. Little, D. Learner autonomy: A theoretical construct and its practical application. Die Neueren Sprachen 1994, 93, 430–442. [Google Scholar]
  6. Zimmerman, B.J.; Schunk, D.H. (Eds.) Self-regulated learning and performance: An introduction and an overview. In Handbook of Self-Regulation of Learning and Performance; Routledge: Oxfordshire, UK, 2011; pp. 1–12. [Google Scholar]
  7. Flavell, J.H. Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. Am. Psychol. 1979, 34, 906–911. [Google Scholar] [CrossRef]
  8. Lai, C. A framework for developing self-directed technology use for language learning. Lang. Learn. Technol. 2013, 17, 100–122. [Google Scholar]
  9. Zheng, C.; Liang, J.C.; Li, M.; Tsai, C.C. The relationship between English language learners’ motivation and online self-regulation: A structural equation modelling approach. System 2018, 76, 144–157. [Google Scholar] [CrossRef]
  10. Oxford, R.L. Expanded perspectives on autonomous learners. Innov. Lang. Learn. Teach. 2015, 9, 58–71. [Google Scholar] [CrossRef]
  11. Benson, P. Teaching and Researching Autonomy in Language Learning, 2nd ed.; Pearson Education: London, UK, 2011. [Google Scholar]
  12. Murray, G. The social dimensions of learner autonomy and self-regulated learning. Stud. Self-Access Learn. J. 2014, 5, 320–341. [Google Scholar] [CrossRef] [PubMed]
  13. Sauro, S.; Zourou, K. What are the digital wilds? Lang. Learn. Technol. 2019, 23, 1–7. [Google Scholar]
  14. Little, D.; Thorne, S.L. From learner autonomy to rewilding: A discussion. In Learner Autonomy and Web 2.0; Cappellini, M., Lewis, T., Mompean, A.R., Eds.; Equinox: Sheffield, UK, 2017; pp. 12–35. [Google Scholar]
  15. Lai, C. Insights into Autonomy and Technology; Castledown Press: London, UK, 2023. [Google Scholar]
  16. Alm, A.; Watanabe, Y. Online machine translation for L2 writing across languages and proficiency levels. Aust. J. Appl. Linguist. 2022, 5, 135–157. [Google Scholar] [CrossRef]
  17. Kern, R. Twenty-five years of digital literacies in CALL. Lang. Learn. Technol. 2021, 25, 132–150. [Google Scholar]
  18. Teng, M.F. Autonomy, Agency, and Identity in Teaching and Learning English as a Foreign Language; Springer: New York, NY, USA, 2019. [Google Scholar]
  19. Lai, C.; Chen, Q.; Wang, Y.; Qi, X. Individual interest, self-regulation, and self-directed language learning with technology beyond the classroom. Br. J. Educ. Technol. 2024, 55, 379–397. [Google Scholar] [CrossRef]
  20. Mollick, E. Co-Intelligence: Living and Working with AI; Portfolio/Penguin: Brentford, UK, 2024. [Google Scholar]
  21. Chapelle, C.A.; Beckett, G.H.; Ranalli, J. Paths for exploring AI in applied linguistics. In Exploring Artificial Intelligence in Applied Linguistics; Chapelle, C.A., Beckett, G.H., Ranalli, J., Eds.; Iowa State University Digital Press: Ames, IA, USA, 2024; pp. 262–274. [Google Scholar]
  22. Godwin-Jones, R. Distributed agency in second language learning and teaching through generative AI. Lang. Learn. Technol. 2024, 28, 5–31. [Google Scholar]
  23. Thorne, S.L. Generative artificial intelligence, co-evolution, and language education. Mod. Lang. J. 2024, 108, 567–572. [Google Scholar] [CrossRef]
  24. Kalantzis, M.; Cope, B. Literacy in the Time of Artificial Intelligence. Read. Res. Q. 2025, 60, e591. [Google Scholar] [CrossRef]
  25. Yin, R.K. Case Study Research and Applications: Design and Methods, 6th ed.; Sage: London, UK, 2018. [Google Scholar]
  26. Alm, A. Understanding AI ethics through the student’s eyes: A study of ChatGPT integration in a University German course. In Artificial Intelligence and Academic Integrity: Navigating Ethical Challenges of AI in Education; Daniel, B., Ajji, I., Sheridan, L., Wierdak, N., Eds.; Springer: New York, NY, USA, 2025. [Google Scholar]
  27. Ohashi, L. Self-Directed Learning and the Teacher’s Role: Insights from Two Different Teaching Contexts. In Future-Proof CALL: Language Learning as Exploration and Encounters; Taalas, P., Jalkanen, J., Bradley, L., Thouësny, S., Eds.; Research-Publishing.Net: Voillans, France, 2018; pp. 236–242. [Google Scholar]
  28. Steel, C. The learning challenge: Discovering the power of digital practices. In Proceedings of the ASCILITE 2014—Annual Conference of the Australian Society for Computers in Tertiary Education Rhetoric and Reality: Critical perspectives on educational technology, Dunedin, New Zealand, 23–26 November 2014; pp. 335–347. [Google Scholar]
  29. Reinders, H.; White, C. 20 years of autonomy and technology: How far have we come and where to next? Lang. Learn. Technol. 2016, 20, 143–154. [Google Scholar] [CrossRef]
  30. Ranalli, J. L2 student engagement with automated feedback on writing: Potential for learning and issues of trust. J. Second. Lang. Writ. 2021, 52, 100816. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Alm, A. Exploring Autonomy in the AI Wilderness: Learner Challenges and Choices. Educ. Sci. 2024, 14, 1369. https://doi.org/10.3390/educsci14121369

AMA Style

Alm A. Exploring Autonomy in the AI Wilderness: Learner Challenges and Choices. Education Sciences. 2024; 14(12):1369. https://doi.org/10.3390/educsci14121369

Chicago/Turabian Style

Alm, Antonie. 2024. "Exploring Autonomy in the AI Wilderness: Learner Challenges and Choices" Education Sciences 14, no. 12: 1369. https://doi.org/10.3390/educsci14121369

APA Style

Alm, A. (2024). Exploring Autonomy in the AI Wilderness: Learner Challenges and Choices. Education Sciences, 14(12), 1369. https://doi.org/10.3390/educsci14121369

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop
  NODES
Association 2
chat 51
coding 2
Community 1
deepl 8
Frameworks 3
games 6
games 6
HOME 1
Idea 4
idea 4
innovation 3
Interesting 2
Intern 31
iOS 12
Javascript 2
languages 7
mac 28
Note 14
OOP 1
os 127
text 24
twitter 1
Users 1
visual 1
web 3