1. Introduction
Since the public release of Open AI’s ChatGPT in 2022, higher education has dealt with a technological transformation that promises to fundamentally alter the nature of teaching and learning. Likened by practitioners to shifting ground (Kennedy 2023) and a tsunami (D’Agostino 2023), generative Artificial Intelligence (AI) evokes the image of a new geographical event rocking the higher education landscape. Or does it? Even in the previous two phrases, one can appreciate how discourse on technology AI easily lends itself to hyperbolic language and strong metaphors (Weller 2022).
Currently, a wide range of academic, corporate, and policy stakeholders are competing to shape the narrative of what is happening and what should be done about it (Eynon & Young 2021). A tangible manifestation of this solicitude has been the proliferation of AI literacy frameworks published by nongovernmental organizations, universities, and EdTech companies (Miao & Shiohira 2024; MacCallum et al. 2024; Bekiaridis & Attwell 2024). These documents promise to provide a roadmap or blueprint to guide faculty and third space professionals, indicating the most salient issues and defining roles for AI, students, and teachers.
Based on these premises, our paper constitutes a first attempt to synthesize the insight from this literature using idiographic metaphor analysis (Redden 2017) to unpack the roles of teacher, student, and AI within literacy frameworks. We do so through a rigorous methodology based on inductive coding of emergent language from 18 AI literacy framework documents to create metaphorical categories coded for relevance and centrality. The outcome is a hierarchy of metaphors for each of the three actors put in relation to each other to explore connections and controversial pairings.
Our analysis highlights the dominant metaphors for each actor within the studied literacy frameworks: AI-as-tool-transformer-ubiquitary-artefact-threat; student-as-analyst-citizen-creator; and teacher-as-designer-guide. Some roles are consistent with overarching educational paradigms, such as pairing AI-as-tool with student-as-creator. Others bring about metaphorical tensions that provide a means for critiquing AI literacy that does not rely on direct comparison with these prevailing models (Weller 2022). We have divided these tensions into three parts to spark conversation on directions for AI literacy development outside the techno-determinist vision of this transformation (Oliver 2011; Nemorin et al. 2023).
After this introduction, we organize the paper as follows: first, we offer a short literature review of AI literacy and metaphors in higher education. Then, we describe our research questions, methodology, case selection, and positionality as researchers. Third, we present our results with the help of three tables that capture the eight most common metaphors for AI, students, and teachers within the studied AI literacy frameworks. The fourth section then explores the connections between the roles of each actor, including some critical points for discussion. The final section reflects on the relevance of this paper and proposes suggestions for future research.
2. Literature Review
When writing about the impact of a recent technological trend, one can be tempted to stick to the most recent publications, yet novelty bias should not affect discussions of the latest transformations in our field. Education has a long tradition of exploring meaning-making processes and routinely uses metaphors to make its concepts easier to grasp, especially in periods of change (Petrie & Oshlag 1993). One instrument for the spread of new meanings and framing contests among different discourses is literacy frameworks (Rebmann 2013). This short literature review discusses metaphors and literacy, both alone and in their interaction.
Whenever geography is transformed, whether through natural, human, or artificial causes, there is a need for individual and collective sensemaking. Metaphors are an appropriate starting point for putting these new features on people’s maps (Bozkurt et al. 2024). Some even argue that our conceptual systems – how we make sense of the world and relate to other human and nonhuman actors – are fundamentally metaphorical (Lakoff & Johnson 1980). Some metaphors are structural, allowing people to understand complex topics in concrete terms. Lakoff and Johnson (1980) give the example of time-as-a-resource, highlighting how everyday experiences with money and limited resources shape how we conceptualize time. Other metaphors are imaginative or generative, allowing us to see aspects of a concept from a new angle (Schön 1979). For example, Bender et al.’s (2021) stochastic parrot metaphor for AI critically challenges the intelligence and anthropomorphization of large language models.
Metaphors have a rich history within our understanding of the roles of students and teachers (Bowman 1997; Low 2008), shifting as the overarching paradigm of education has become more student-centered (e.g., from sage on the stage to guide on the side, King 1993) and has accommodated technological change (e.g., students as navigators, Weller 2022). Closely related to metaphor, metonymy is similarly prominent in describing the roles of teachers and students in the digital age (see, e.g., Ferrari & Punie 2013; ISTE 2024). One example is the part-of-the-whole metonymy, where a term representing a part of something refers to the entire entity (Lakoff & Johnson 1980). Examples can be found within the European Union’s Digital Competence Framework for Educators (Ferrari & Punie 2013), which includes roles like Thought Leader and Pioneer, as well as the International Society for Technology in Education’s Standards for Students (ISTE 2024) with the roles of Creative Communicator and Innovative Designer. In this paper, we do not differentiate between metaphors and metonymies; we use the term “metaphor” to encompass both concepts.
Likewise, literacy has evolved beyond its traditional scope of reading and writing (Tyner 2014). Kellner and Share (2007: 5) define literacy as “gaining the skills and knowledge to read, interpret, produce texts and artifacts, and to gain the intellectual tools and capacities to fully participate in one’s culture and society.” This literary-based approach now encompasses a range of modalities, including information, media, digital, and data, also conceptualized as multiliteracies (Tyner 2014) or new literacies (Lankshear, Knobel & Curran 2013). Hence, literacy is viewed as a metaphorical tool, enabling citizens to develop knowledge and skills and participate actively in society in times of technological change (Kellner & Share 2007). Such an approach also has its detractors, who critique the excessive ubiquity of literacies across the educational and media landscape (Mason, Krutka & Heath 2021; Mishra 2024).
For this paper, a framework is a conceptual structure that defines guidelines and principles needed to achieve an educational goal. On constantly shifting ground, technological literacy frameworks represent a guide or roadmap for navigating emerging trends (Chan & Colloton 2024). Naturally, then, the public release of generative AI has seen educational stakeholders call for widespread AI literacy and develop a trove of accompanying frameworks (Miao & Shiohira 2024; Bozkurt 2024; Hibbert et al. 2024). Given the complexity of articulating what constitutes AI literacy, many frameworks employ Long and Magerko’s (2020: 2) popular definition, “a set of competencies that enables individuals to critically evaluate AI technologies; communicate and collaborate effectively with AI; and use AI as a tool online, at home, and in the workplace.” This definition’s strength lies in its simplicity, while also setting the bar high with the idea that AI-literate individuals ought to appraise its capabilities, merits, and limitations.
Crucially, conceptual metaphors complement AI literacy frameworks by organizing our perception of this complex topic and guiding our responses. Both are dynamic, evolving as our understanding of AI and literacy changes. However, metaphors and frameworks share common challenges: if too simple, they are deemed essentialist or limiting; if too complex, they lack utility1 (Lakoff & Johnson 1980; Petrie & Oshlag 1993). As Chan and Colloton (2024) elucidate, implementing AI literacy frameworks can be difficult because differences across sectors, geographies, and diverse disciplinary backgrounds easily complicate the elaboration of a definitive set of skills and capabilities. Furthermore, it can be uncomfortable for teachers and students to work according to metaphors and, by extension, frameworks that are inconsistent with their worldview and epistemology (Clarken 1997).
Despite these limitations, the intersections of technological literacy and metaphors in educational literature show how they jointly provide conceptual structure to emerging phenomena. Several recent studies measure teachers’ and students’ metaphorical perceptions of digital literacy. Dedebali (2020) examined the metaphors for digital literacy generated by teacher candidates, grouping them under five categories: vitality, integrity, guidance, complexity, and eternity. Likewise, Gündogmus (2024) organized primary school teachers’ metaphors for digital literacy into five categories: infinity, object, movement, need, and guide. In practice, metaphor generation activities are a promising critical AI literacy technique to teach about it in the classroom (Gupta et al. 2024; Vallis, Wilson & Casey 2024).
In such a robust discourse, there is no better place to look for metaphors than within AI literacy frameworks themselves. In the following sections, we employ metaphor analysis to map the most common conceptual metaphors for AI, students, and teachers within a sample of this recently proliferating field and then explore the connections and tensions that arise between them.
3. Methodology
As a work that empirically tackles the discourse on AI literacy, it is vital to adopt a methodological approach that is both rigorous and explicit. We draw upon idiographic metaphorical analysis (Redden 2017), an approach that inductively examines metaphors as they appear in text. As Redden explains, idiographic (cfr. nomothetic) metaphor analysis is useful for uncovering veiled meanings and delving into taken-for-granted assumptions without the bias of researcher prompting. Here, metaphors are embedded within literacy frameworks’ context, purpose, and values. To uncover them, we coded the language of eighteen AI literacy frameworks to capture the metaphors about AI, students, and teachers, addressing the following research questions:
R1: Which metaphors are common within AI literacy frameworks to describe AI, students, and teachers?
R2: Which metaphorical relationships between AI, students, and teachers are generated within AI literacy frameworks?
R3: What are the tensions within these metaphorical relationships and with the framework’s stated goals for AI literacy?
3.1 Case selection
Our universe of cases and potential corpus of documents was represented by AI literacy frameworks published in English before the end of 2024. We sought a diverse case selection (Curtis et al. 2000), including multiple sectors and publication formats, as we deemed these likely to influence the frames and language through which AI is discussed (Eynon & Young 2021). To ensure variation in scope, institutional context, and underlying goals, the frameworks therefore originate from higher education practice (7), higher education research (6), EdTech companies (3), nongovernmental organizations (2), and educational consultancy (1). The frameworks appear in a variety of formats: report (4), scoping review (3), research article (3), editorial (2), blog (2), conference proceedings (1), dissertation (1), and book chapter (1). Geographical diversity was also considered within the case selection, while we excluded works on AI literacy assessment scales that lacked an original framework. Table 1 provides an overview of the chosen frameworks and their main features.
Table 1
Overview of Frameworks.
| # | FRAMEWORK | AUTHORS | YEAR | ORGANIZATION | FORMAT |
|---|---|---|---|---|---|
| 1 | Deep Learn Framework | Phillips | 2024 | FH Joanneum University of Applied Sciences | Editorial |
| 2 | Supplement to the DigCompEDU Framework | Bekiaridis & Attwell | 2024 | AI Pioneers | Report |
| 3 | Framework for the Future | Becker, Parker, & Richter | 2024 | Moxie | White Paper |
| 4 | 3wAI Framework | Bozkurt | 2024 | Anadolu University | Journal Article |
| 5 | Machine Learning Education Framework | Lao | 2020 | Massachusetts Institute of Technology | Dissertation |
| 6 | Conceptualizing AI literacy: An exploratory review | Ng et al. | 2021 | University of Hong Kong | Scoping Review |
| 7 | UNESCO AI Competency Framework for Students | Miao & Shiohira | 2024 | UNESCO | Report |
| 8 | UNESCO AI Competency Framework for Teachers | Miao & Cukurova | 2024 | UNESCO | Report |
| 9 | AI Literacy Framework | Kennedy | 2023 | AI Literacy Institute | Blog |
| 10 | A Framework for AI Literacy | Hibbert et al. | 2024 | Barnard University | Editorial |
| 11 | What is AI Literacy? Competencies and Design Considerations | Long & Magerko | 2020 | Georgia Institute of Technology | Scoping Review |
| 12 | AI Across the Curriculum | Southworth et al. | 2023 | University of Florida | Journal Article |
| 13 | A systematic review of AI literacy conceptualization, constructs, and implementation and assessment efforts (2019–2023) | Almatrafi, Johri, & Lee | 2024 | King Abdulaziz University; George Mason University | Scoping Review |
| 14 | AI Literacy in Higher Education: Theory and Design | Černý | 2023 | Masaryk University | Conference Proceedings |
| 15 | The Scaffolded AI Literacy (SAIL) Framework for Education | MacCallum, Parsons & Mohaghegh | 2024 | University of Canterbury, academyEX, and AUT | Report |
| 16 | Dynamic AI Literacy Model | Chan & Colloton | 2024 | University of Hong Kong | Book Chapter |
| 17 | Critical AI Literacy | Bali | 2024 | American University in Cairo; London School of Economics | Blog |
| 18 | Rethinking the entwinement between artificial intelligence and human learning | Markauskaite et al. | 2022 | The University of Sydney (first author) | Journal Article |
3.2 Coding methodology
For this paper, we adopted a qualitative methodology using a form of coding based on inductive categories (Skjott Linneberg & Korsgaard 2019). For each object of our research – AI, student, teacher – we drafted field notes covering the use of metaphors. The second step was assigning a specific word for each metaphor used. Some of the metaphorical language was directly pulled from the text (e.g., “dragon”) while other words were semantically derived (e.g., verbs like “cultivate” and “grow” resulted in a “gardener” metaphor.) After completing this process for all frameworks, the first stage of analysis was to group all metaphors and markers with the same underlying meaning, whenever possible, and label each group. Then, we chose the eight most frequent metaphor clusters for each actor as the final qualitative coding categories, ensuring we analyzed the most dominant conceptual metaphors. To increase the reliability of this process, we coded all materials separately and discussed disagreements before reconciling them within the categories.
Lastly, we scored each AI literacy framework for each category through a point system, where ‘0’ represented the metaphor’s absence, ‘1’ represented its presence, and ‘2’ represented a central framing (see Appendix 1). For example, the citizen metaphor for student was coded as a ‘2’ in Lao (2020) and Southworth et al. (2023) because of its continued use and prominence throughout the texts. The same metaphor was coded as a ‘1’ in Miao and Shiohira (2024) and Markauskaite et al. (2022) because it appeared infrequently or only marginally. We assigned a ‘2’ to a maximum of two metaphorical categories for each actor in any framework.
3.3 Positionality statement
We recognize that metaphor analysis has a subjective component, which we embrace as part of educational research. We share our positionality here to be transparent on how our backgrounds may influence and shape our perceptions of the studied frameworks. The first author is an academic technologist at a private higher educational institution in Central Europe. Within a university community that has faced several “crises” in past years, from forced relocation to the COVID-19 pandemic, her interest in AI arises from supporting faculty and students with the critical assessment and integration of educational technologies in transitional periods. Her position on AI is engaged yet strongly critical and concerned (see Compton 2024). The second author is a professor of social sciences at a public university in Central Europe. His interest comes from his experience researching digital media use by political parties and discussing academic integrity and AI tools at his institution. At the time of writing, his position on AI is a combination of critical, open-minded, and skeptical. Our disciplinary backgrounds and roles are complementary, as the study of educational technology is inherently political (Kellner & Share 2007) and pertinent to both those who teach and those who operate within the university’s third spaces.
4. Results
A significant number of metaphors surfaced from the analysis of the three actors: AI, students, and teachers. This section will describe the eight most prominent for each, highlighting notable outliers and initial patterns that form the basis of the most common conceptual metaphors.
4.1 Metaphors for AI
We created 14 categories based on the function of AI within educational networks, derived from an initial list of 77 metaphorical terms. We then coded on our 0-1-2 scale the eight most prominent categories. After weighing the categories’ centrality, the most prominent were tool (16), transformer (16), ubiquitary (15), artefact (14), and threat (12). These categories characterize AI as a “ubiquitous artefact used as a potentially threatening tool for transforming education.” Table 2 below illustrates each metaphorical category for AI, including a brief definition and a few examples from the frameworks.
Table 2
Metaphor Categories for AI.
| METAPHOR CATEGORY & WEIGHT | CATEGORY DEFINITION | EXAMPLES |
|---|---|---|
| Tool (16) | Supports or aids people in some way | aid; scaffold; crutch; enhancement |
| Transformer (16) | Changes higher education or society in some way | (re)shaper; disruption; revolution; explosion; shifting ground |
| Ubiquitary (15) | Emphasizes that AI is everywhere and unavoidable | diffusion; influx; proliferation; saturation; permeation |
| Artefact (14) | Emphasizes that AI is a human-made object and has machine-like qualities | automation; artificial being; computational being; machine |
| Threat (12) | Poses some risk or danger to education and/ or society | predator; attentional warfare; gap widener; warped lens |
| Evolution (8) | Emphasizes that AI is developing | growth; development; emergence |
| Human (7) | Assigns humanlike characteristics to AI | brain; nervous system; emotional intelligence |
| Collaborator (6) | Assigns equal status to AI and humans working together | (thought) partner; companion; teammate |
In addition to the categories we coded, we uncovered other minor categories: novice, animal, agent, discipline/field, solution, and conduit. Many of the metaphors in these categories weren’t widespread but were generative in their creative contribution to our understanding of AI, such as a dragon (Bozkurt 2024) and cake-making (Bali 2024). Metonymy was also common in the discourse related to AI, particularly the producer-for-product construct (Lakoff & Johnson 1980), where authors substitute AI when referencing the individuals or companies responsible for AI products (see Becker, Parker & Richter 2024).
4.2 Metaphors for students
We found 59 metaphors for students, which we organized into 12 total categories. We then weighed the eight most prominent categories for centrality, with the highest being analyst (15), citizen (11), and creator (10). The presence of the student-as-analyst and student-as-creator metaphors reflects the impact of Bloom’s higher-order thinking skills in shaping the frameworks, as Ng et al. (2021) built off Bloom’s taxonomy (1956) and others expanded upon it as an early prominent model (Southworth et al. 2023; Hibbert et al. 2024). The dominance of the analyst over the creator metaphor provides a snapshot from this moment in AI literacy, as many frameworks supported the critical analysis of AI, but not all emphasized that students should be using or creating with it (see Long & Magerko 2020; Bali 2024). Table 3 offers a characterization of each metaphorical category for students, including a brief definition and a few examples.
Table 3
Metaphor Categories for Students.
| METAPHOR CATEGORY & WEIGHT | CATEGORY DEFINITION | EXAMPLES |
|---|---|---|
| Analyst (15) | A person who examines something in detail to understand more about it | judge; appraiser; critic; strategist |
| Citizen (11) | A member of society who is concerned about ethics and the wellbeing of others | advocate; (model) citizen, (global) citizen |
| Creator (10) | A person who develops AI or creates with AI | contributor; tinkerer, engineer; crafter; alchemist |
| Collaborator (9) | A person who works with AI on equal footing | partner; co-creator |
| Superior (8) | A person who holds power over AI | human at the helm, kings and queens, executor |
| Future (6) | The people of tomorrow who will run society | the future; survivor |
| Explorer (6) | A person who explores a new or unfamiliar area | navigator; rider; pioneer |
| Learner (5) | A person who is gaining knowledge, skills, and experience | gardener of one’s own mind; self-educator; transformed |
In addition to the categories we coded for prominence, we uncovered minor categories: workforce, manager, user, and expert. Many of these were generative metaphors that weren’t widespread, such as human capital (Southworth et al. 2023) or mitigator (Becker, Parker & Richter 2024), or terms that lacked strong metaphorical language, such as user.
4.3 Metaphors for teachers
Finally, 13 categories were synthesized from a list of 70 initial metaphorical terms used for teachers. We weighed the eight most prominent categories for centrality, and two emerged as the most prominent: designer (16) and guide (12). Notice how, differently from AI and students, these two metaphorical roles are not especially revolutionary or contentious. Still, their application to an educational environment transformed by technology must be scrutinized. Table 4 below offers a characterization of each metaphorical category for educators, including a brief definition and a few examples.
Table 4
Metaphor Categories for Teachers.
| METAPHOR CATEGORY & WEIGHT | CATEGORY DEFINITION | EXAMPLES |
|---|---|---|
| Designer (16) | A person who creates pedagogical experiences, with or without the use of AI | Creator; contributor; developer; engineer |
| Guide (12) | A person who shows the way to others | mentor; facilitator; coach; equipper; counselor |
| Expert (9) | A person who is very knowledgeable about or skillful in a certain field | sage; curriculum conduit; role-model |
| Analyst (9) | A person who examines something in detail to understand more about it | judge; assessor; critic; strategist |
| Superior (8) | A person who holds power over AI | human at the helm, leader, master |
| Citizen (7) | A member of society who is concerned about ethics and the wellbeing of others | advocate; equalizer; humanitarian; social justice champion |
| Explorer (6) | A person who explores a new or unfamiliar area | navigator; rider; pioneer; innovator; driver |
| Learner (5) | A person who is gaining knowledge, skills, and experience | gardener of one’s own mind; self-educator; transformed |
In addition to the categories we coded, we uncovered the following minor categories: cheerleader, collaborator, transformer, user, and controller. Some categories overlap with those for students, yet they reflect different aspects of the respective roles of teachers in higher education. For example, the citizen metaphor was common for both students and teachers. However, students were construed as the responsible or global citizens of tomorrow (Southworth et al. 2023; Miao & Shiohira 2024), whereas teachers had a substantive democratizing role in the present as equity champions (Miao & Cukurova 2024) and humanitarians (Chan & Colloton 2024). Likewise, the learner metaphor for students emphasized self-education surrounding AI topics (Cerný 2023), while for educators, it captured continuous professional development and acquiring AI skills that they can teach students (Bekiaridis & Attwell 2024). Notice how some metaphors in these frameworks apply to both teachers and students as beneficiaries of AI literacy.
5. Discussion
This discussion will delve into the relationships between each of the three actors. Given the short length of this paper, we discuss only the metaphors at the core of the overall discourse, which in practice, corresponded to those that exceeded a 10-point threshold in the final stage of coding. Figure 1 below is a simple visual representation of the interconnected system of metaphors generated by the three actors’ simultaneous presence. The following paragraphs will then highlight salient elements for discussion arising through connections and tensions between roles.
Figure 1
Metaphorical Connections between Students, Teachers, and AI.
5.1 Connections between the metaphors
In analyzing the literacy frameworks, several connections arise between the metaphors for students and AI. First, the metaphor of students-as-creators synergizes with the view of AI-as-a-tool that supports their learning process and as an artefact that can be edited, customized, and transformed. The relationship applies not only to producing or crafting with AI but also to engineering the AI tools themselves (see Lao 2020). This creator role is complementary to the other Bloom-derived metaphor of students-as-analysts, who are critical of the tools used within the creative process. For example, Long and Magerko encourage learners “to be critical consumers of AI technologies by questioning their intelligence and trustworthiness” (2020: 9). Leaning further into the educationalization paradigm (Depaepe & Smeyers 2008), students should be attuned to AI as a potential threat to society that must be appraised and evaluated to be contained.
Additionally, the conceptualization of AI-as-a-ubiquitary invokes the students-as-citizens role, as freedom from AI becomes narrower and students are asked to “drive human societies towards inclusive, environmentally sound, shared futures” (Miao & Shiohira 2024: 14). The universality of AI also involves the digital citizen metaphor, as students today are increasingly unable to opt out from technology (Nayak & Walton 2024). The stakes attached to citizenship and politics are dramatically raised in dense spaces (McFarlane 2016) – such as in cities as opposed to the countryside – just like the survival of virtual spaces depends upon a critical mass of users — think of Second Life or the Metaverse. This is reflected in the current solicitude by vendors to develop and contract educational AI products (Williamson, Molnar & Boninger 2024).
The metaphorical relationship between teachers and AI is characterized by a sense of responsibility, as educators play a critical role in shaping students’ learning experiences. This is where the role of teachers-as-designers combines with the view of AI-as-a-tool/ transformer, as teachers tap into new AI features for educational purposes. Examples include the ability to adapt instruction to learner types (Phillips 2024), provide customized feedback (Bekiaridis & Attwell 2024), and facilitate communication (Southworth et al. 2023; Chan & Colloton 2024). Moreover, the classic metaphor of teachers-as-guides takes on a new value given the inescapable and potentially dangerous character of AI-as-a-ubiquitary and a threat to learning (Williamson, Molnar & Boninger 2024). In other words, given the proliferation of AI despite its possibly negative attributes, mentoring on the teacher’s side becomes fundamental (see Long & Magerko 2020; Miao & Cukurova 2024).
Lastly, we turn to how introducing AI in learning spaces influences student and teacher relationships. On the surface, we find this essentially unchanged from the contemporary view of student-centered learning (Jones 2007), in which teachers design learning experiences that guide and support students’ analytical skills and experimentation. Moreover, the centrality of the students-as-analysts metaphor orients our understanding of teachers-as-guides who can help students correctly appraise and evaluate AI tools (Phillips 2024). Looking at other metaphorical layers adds depth to the relationship. For example, some frameworks positioned teachers and students as co-explorers, navigating a transformed technological landscape (MacCallum, Parsons & Mohaghegh 2024). This is an entry point for novel approaches to AI that place students on more equal footing with teachers as they pioneer new possibilities (Mawasi et al. 2023).
5.2 Tensions within the metaphors
In addition to the pairings consistent with prevailing educational models, tensions arose between metaphors that are hard to reconcile within the dominant discourse. These helped us identify three blind spots within AI literacy frameworks: (1) is AI literacy currently too complex to be useful?; (2) whose ends does AI literacy serve?; (3) should AI literacy be an individual and/ or collective affair? These questions can be used to develop critical dimensions of AI literacies, and we discuss them briefly in the following subsections.
A mirror bigger than the room it was placed in
The wide variety of metaphors within the studied AI literacy frameworks highlights the evolving roles of students and teachers as they navigate a tenuous technological landscape. This variation also highlights the complexities within said frameworks and, at times, overcomplexities. While nearly all include practical skills-based and ethical components, there is still little agreement on what those skills and competencies should be. Taken as a whole, the field is reminiscent of Borges’ larger-than-life map of an empire that fades into disuse; it provides clarity but also obscures due to its scale (see endnote i).
For example, teachers’ central roles as guides and designers might seem unproblematic in isolation. Yet, once paired with the complexity of some AI metaphors (transformer, threat, tool) and the social responsibility and critical thinking requested of students (citizen, analyst), their role within AI literacy can become daunting. Teachers must acquire a deep understanding of AI’s functioning to think critically about it and teach students how to use it in practical, ethical ways. While teaching is a multifaceted activity, often requiring teachers to operate under several metaphors at once (Clarken 1997), it begs the question of how these fit into the many other roles teachers enact daily.
Therefore, for utility’s sake, it may be helpful to narrow or converge these diverse models of AI literacy to reveal alternative paths for action. This does not necessitate eliminating content, per se, but rather drawing up a more systematic layering of metaphors to guide a collective response. While many frameworks liken themselves to a guide or roadmap (Miao & Shiohira 2024; Kennedy 2023), we must remember that AI literacy is not a panacea. If new AI literacy frameworks continue to proliferate, are we “wishing for a future that cannot hold our wish?2 Returning to the literacy-as-tool metaphor (Kellner & Share 2007), perhaps AI literacy should be seen as one of many tools in a cartographer’s kit rather than as the map itself. This opens space for other actions, such as refusing or opting out of AI features (The University of Greenwich 2023), designing alternatives to corporately developed technologies (The University of Edinburgh 2024), and winning political actors to a more progressive educational agenda (Jones 2019).
Whose ends does AI literacy serve?
Considering literacy as an instrument, we return briefly to the origins of literacy in reading and writing, which has functioned over time as both an enabler and detractor of democratic participation (Giroux 1986; Tyner 2014). Scribner (1984) expresses this tension in a series of metaphors that capture literacy’s intricacies, dependent upon social context, purpose, and values. Literacy-as-adaptation, referring to its survival value, imbues individuals with the practical skills to perform effectively within society. Alternatively, literacy-as-power emphasizes relationships with group and community advancement, making literacy a resource for social transformation. These metaphors are woven throughout the studied frameworks. For an example of literacy-as-adaption, see Southworth et al. (2023: 1): “the creation of an AI-ready workforce covering the essential 21st-century competencies identified as workforce and government needs worldwide.” Likewise, see Bali (2024) for an instance of literacy-as-power: “[to] be better able to resist using AI in ways that are harmful or inappropriate. Instead, they will feel empowered to use it constructively, while being aware of the limitations.” Several frameworks incorporate both metaphors (see Bekiaridis & Attwell 2024).
There is a nuanced spectrum of acceptant and resistant positions to AI within higher education institutions (Compton 2024), acknowledging the complexity of the student/teacher-as-analyst metaphor found within AI literacy. Proponents see the potential for AI to solve many of education and society’s most pressing issues (Nemorin et al. 2023). Critics worry that the choice to engage with AI, to become literate in the sense of adaptation, is in the interest of technology creators, who lobby and consolidate power within educational and political institutions (Jones 2019; Weller 2022). Kellner and Share (2007) discuss how conventional media literacy perpetuates the myth of education as politically neutral. Likewise, Ferreira et al. (2019) describe how the IT industry promotes the view of data-driven technologies as a neutral tool, often obscuring how they reproduce and deepen existing inequitable social structures. Building on critical media and data literacy, our analysis of the recurring metaphors in the studied frameworks allows us to reflect upon whose goals are being served not just by AI but also by the diffusion of AI literacy. As Bhargava et al. (2015: 14) pertinently elucidate for data literacy,
“[a]dvanced data analytics techniques, despite their potential to spur human progress, have so far worked especially well for governments and corporations. It is unclear whether and how promoting ‘data literacy’ the way it is currently conceptualized– by providing skills without […] questioning their ends and means – may reverse or repeat the history of literacy promotion.”
While the studied frameworks often evoke metaphors of power, the language is frequently casual and non-confrontational, e.g., “it will soon become imperative that everyone feel empowered living with machine learning” (Lao 2020: 28). They also provide little direction for enacting power: “empower students and teachers to shape the digital futures we want” (Miao & Shiohira 2024: 6). Some metaphorical roles, such as teachers-as-guides and students-as-citizens, are charged with collective responsibility, but are they just a conduit for social control and economic gain?
Individual vs. collective literacy
Finally, the explored metaphors usually target teachers and students at the individual level, yet they exist within higher education communities and cultures. The sector needs more collective action to shape the future of AI, as a single member’s complaint or rejection has little weight and might only damage their job perspectives. This tension evokes the famous exit, voice, and loyalty framework (Hirschman 1970). Conceived initially to explain product-specific customer behavior – do customers stop buying, complain, or stay loyal? – this framework has become one of the most powerful conceptual masks within the social sciences. Most recently, Frey and Schneider (2023) highlighted the importance of collective ‘voice’ to advocate for better institutional governance mechanisms of online communities and tools. We extend this sentiment to higher education, where students and teachers also need mechanisms to pool their ‘voice’.
In particular, students’ analyst and creator roles subscribe to an individualistic vision but also represent an institutional prerequisite for voicing personal and collective concerns. This highlights a dangerous competency hierarchy in the absence of broad public conversations. Are those who are not AI literate allowed to voice their opinions and express their concerns? Conversely, a collective perspective that implies loyalty – think of higher education institutions full of “reluctant adopters” – underlies discussions of AI-as-ubiquitary-threat-and-transformer (Williamson, Molnar & Boninger 2024). Frameworks built on hierarchies, like Bloom’s Taxonomy (1956), also imply that a better understanding of the tools can only lead to more universal adoption within higher education. After all, expecting AI literacy frameworks to discuss ‘exit’ (technological opt-out) is unrealistic, given their stake in its diffusion.3 Even without fully reconciling the individual and collective perspectives, one must be aware of this implied slippage.
Finally, the necessity of individual literacy is treated as uncontroversial, but what about a framework for institutional or group literacy? Here, the literacy frameworks and their implied technological prerequisites take on a dangerous connotation of gatekeeping for the possibility of voice among community members not directly involved in the technological transition. Instead, we may consider if all members of an institution should aim for the same breadth of AI literacy or if they can pool their diverse skills, expertise, and disciplinary backgrounds to shape AI-literate universities (see Gibson et al. 2023 for a pertinent discussion of the meso and macro-level roles of AI).
6. Conclusion
In sum, this paper’s empirical analysis of AI literacy frameworks found several widespread metaphors for each actor, with clear implications for scholars and practitioners. While the key metaphors for teachers are less controversial, the picture is more complex regarding students and AI. Overall, the frameworks tend to reinforce dominant educational paradigms and notions of good teaching and learning practice. However, the metaphorical relationships also bring to light openings that have yet to be addressed in existing frameworks, such as reclaiming terrain from corporate actors, opportunities for collective action, and an ‘exit’ option for critical actors.
While the call for literacy initiatives gives the impression of institutional activity, the current corpus of AI literacy scholarship lacks benchmarks for concrete evaluation of said action (see Knoth et al. 2024). Who can currently be considered AI literate? Is it a daily user of generative AI? Or someone who routinely reads opinion pieces about AI? What about a teacher who uses AI tools to design course materials? And what level of competency is required of students and teachers? All this confusion is convenient if the goal is exerting social control, and frameworks act as scaffolds or support for adopting these tools. Even the moderately critical elements in the studied frameworks are a weak challenge for a technological discourse with hegemonic aspirations (Eynon & Young 2021). Finally, this article’s analysis reflects a specific moment in time. The situation is in flux, and both the metaphors and the practical uses of AI will adapt as our understanding of it stabilizes (Bozkurt et al. 2024). The goal is to participate in the ongoing conversation with a certain awareness and to direct it in a way that is productive, critical, and reflective of the citizenship role emerging from the frameworks.
Lastly, we suggest the following areas for future research. The first stems from our results, which offer a conversation starter for surveys and focus groups with teachers and students. The metaphorical roles can be used in interviews or workshops, asking participants to rank them for their importance or contribute new metaphors. Another way forward lies in exploring AI literacy in contexts where institutional leadership has encouraged its diffusion. What were the goals of AI literacy initiatives? What were the challenges related to the role of creators or analysts? What does the acquisition of technological literacy skills imply for understanding this transformation? Another area that warrants continued research is approaches for coordinating collective political action. Too often in higher education, individuals shoulder the task of solving collective issues. While there is large agreement that literacy is needed in this evolving technological landscape, educational institutions must also assert their role through policy, cooperation, advocacy, or other currently unforeseen means.
Data Accessibility Statement
The data that support these findings are within the references listed in Table 1. The coding is explicit in the table provided in Appendix 1.
Appendices
Appendix 1
| FRAMEWORK# | ARTIF ICIAL INTELUGENCE IS SEEN AS A… | THE STUDENT IS SEEN AS A… | THETEACHERIS SEEN AS A… | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TOOL | TRANSFORMER | HUMAN | ARTIFACT | UBIQUITARV | COLLABORATOR | EVOLUTION | THREAT | COLLABORATOR | THE FUTURE | EXPLORER | SUPERIOR | ANALVST | CREATOR | CITIZEN | LEARNER | GUIDE | EXPERT | EXPLORER | SUPERIOR | ANALYST | DESIGNER | LEARNER | CITIZEN | |
| 1 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2 | 1 | 2 | 1 | 2 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 1 | 1 | 1 | 1 | 1 | 2 | 1 |
| 3 | 1 | 0 | 0 | 1 | 0 | 2 | 0 | 0 | 2 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 0 | 0 | 0 |
| 4 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 2 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 2 | 1 | 1 | 0 | 0 |
| 5 | 0 | 1 | 1 | 2 | 1 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 2 | 2 | 1 | 0 | 0 | 0 | 0 | 0 | 2 | 1 | 0 |
| 6 | 1 | 1 | 0 | 1 | 2 | 0 | 1 | 0 | 0 | 2 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 1 | 2 | 0 | 1 |
| 7 | 0 | 1 | 0 | 1 | 1 | 0 | 1 | 1 | 1 | 2 | 0 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 0 | 0 | 2 | 0 | 0 |
| 8 | 1 | 1 | 0 | 2 | 0 | 0 | 0 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | ||||||||
| 9 | 0 | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
| 10 | 2 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
| 11 | 0 | 1 | 1 | 2 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 2 | 0 | 0 | 1 | 0 | 1 | 0 | 0 | 0 | 2 | 0 | 0 |
| 12 | 2 | 1 | 0 | 1 | 2 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 1 | 1 | 2 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 13 | 1 | 1 | 0 | 1 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 0 |
| 14 | 1 | 1 | 1 | 0 | 0 | 0 | 1 | 0 | 1 | 0 | 1 | 0 | 1 | 1 | 0 | 1 | ||||||||
| 15 | 1 | 0 | 0 | 0 | 2 | 0 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 0 |
| 16 | 1 | 1 | 2 | 0 | 2 | 1 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 2 | 1 | 1 | 1 | 1 | 1 | 0 | 2 | 1 | 2 |
| 17 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 2 | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | 1 |
| 18 | 1 | 1 | 1 | 0 | 1 | 2 | 1 | 1 | 2 | 1 | 0 | 2 | 1 | 1 | 1 | 0 | 2 | 1 | 0 | 1 | 0 | 1 | 0 | 1 |
| 16 | 16 | 7 | 14 | 15 | 6 | 8 | 12 | 9 | 6 | 6 | 8 | 15 | 10 | 11 | 5 | 12 | 9 | 6 | 8 | 9 | 16 | 5 | 7 | |
Notes
[1] Recalling Jorge Luis Borges’ “On Exactitude in Science” (1946), in which cartographers draw a map of an empire so detailed, it becomes the size of the empire, thus rendering itself useless and forgotten by future generations.
[2] In their track ‘Mirror’, the Beautiful South (1996) mused: “Imagine a mirror bigger than the room it was placed in. Imagine my wish for a future that cannot hold my wish.”
[3] The sponsors of the US’s National AI Literacy Day on March 28, 2025 included the Valhalla Foundation, Google.org, Booz Allen Hamilton, OpenAI, and Salesforce. See https://www.ailiteracyday.org/support-us.
Competing Interests
The authors have no competing interests to declare.
Author Contributions
Both authors played a significant role throughout the entire research and writing process.