How educators talk about AI: a discourse analysis

Artificial intelligence (AI) is shaped by how it is discussed, explained, and understood in educational settings. As the main users and interpreters of AI in the classroom, educators create meaning through language by utilizing particular words, phrases, and analogies to understand its function and implications. How AI is perceived and positioned in educational practice depends on how it is discussed. Discourse takes center stage in the construction of AI's meaning. The emphasis is shifted from what AI accomplishes to how it is portrayed and understood when it is understood as a discursive phenomenon. This perspective emphasizes that meaning is created through language (Demjén and Semino, 2020).

The majority of current research on AI in education has been on observable behaviors such as adoption, usage frequency, and educational applications. Research looks at how educators use AI into their work, their acceptability levels, and the perceived advantages or difficulties (Granström and Oppi, 2025). Existing research overlooks the function of language in creating meaning by concentrating mostly on activities and results, discounting a crucial aspect of how educators interpret technological change.

The way that AI is perceived in educational settings is highly influenced by language. How educators define, categorize, and talk about AI affects how it is perceived, positioned, and eventually applied in practice (Avci et al., 2025). Positive descriptions of AI, for instance, may portray it as helpful and supportive whereas negative or morally charged descriptions suggest it is detrimental or unethical (Kim, 2024).

The purpose of this study is to analyze how educators construct AI through discourse, with a specific focus on recurring language patterns in their narratives. The study centers on how AI is described, interpreted, and represented through word choice, expressions, and linguistic structures. This study is guided by three research questions: How do educators talk about AI in education? What metaphors and frames do they use? And, how is AI constructed through language?

Discourse analysis sheds light on how underlying assumptions, attitudes, and interpretive frameworks are shaped and reflected by language. Language provides a window into both individual and group sense-making by exposing both stated opinions and implicit meanings included in commonplace utterances (Gee, 2014). Researchers can find patterns of framing, recurrent metaphors, and evaluative language that represent broader orientations by looking at how people discuss any subject (Potter and Wetherell, 1987). Such analysis is especially useful because AI lacks solid discursive conventions in contrast to established technologies. Adopting or rejecting a technology is one example of an activity that may be assessed, but the motivations behind it are often ingrained in language (Edwards and Mercer, 1987; Mhlongo et al., 2023). Narratives provide a detailed picture of interpretive processes by capturing underlying meanings in ways that can emphasize tensions, contradictions, and uncertainties (MacLure, 2003).

Language is seen as a constitutive factor that shapes understanding, creates meaning, and affects practice (Austin and Riveros, 2025). How educators discuss teaching, learning, and curriculum, language can both reflect and shape underlying views, attitudes, and assumptions (Nieto, 2001; Zheng, 2025). Examining language as a site of meaning-making is important because discursive constructs may affect how innovations are accepted, negotiated, and implemented (Christie, 2005; Phillips and Hardy, 2002).

Behavioral aspects of AI integration have been the subject of an increasing amount of research including perceived advantages and acceptance patterns (Granström and Oppi, 2025). Discourse-oriented research can show how educators interpret complicated facts by concentrating on patterns of speech, framing, and representation (Phillips and Hardy, 2002). For instance, research on classroom conversation has demonstrated that linguistic analysis can reveal conflicts, inconsistencies, and uncertainties that are difficult to identify using questionnaires or observational checklists (Bloome et al., 2005). Research in the field of educational technology has also shown that implicit assumptions about control, efficiency, and progress are carried out by policy discourses and common discussions about digital tools, which influence implementation strategies (Fahrni et al., 2025). New research has started to look at how AI is presented in institutional guidelines, policy documents, and teacher discussions (Tsao, 2025).

Changing narratives that present technical advancements have influenced the discussion around technology. Educational technologies are presented as answers to problems in teaching and learning through positive frameworks that emphasize transformation, efficiency, and advancement (Tang et al., 2025). Critical viewpoints, on the other hand, have presented technology as a source of disruption, prompting fears about its effects on equality, pedagogy, and the role of educators (Postman, 1993; Selwyn, 2015).

Narratives of invention and progress are often used to promote new technology, but as time goes on, counter discourses that emphasize risks, limitations, and unexpected consequences start to surface. For example, contrasting terminology is used to depict technologies as both improving and damaging educational processes (Wartella and Jennings, 2000; Zhou and Peng, 2025).

Metaphors help people understand complex or unfamiliar issues by linking them to familiar ideas (Cameron, 2003). Teaching, learning, and technological developments are described in educational environments using metaphors, which influence how these processes are seen and assessed (Jensen, 2006; Zheng and Song, 2010). Metaphors shape individual understanding and group discussion (Thibodeau et al., 2019). Framing complements metaphors by highlighting certain characteristics and shaping the interpretation of a problem (Polletta and Ho, 2006).

Discourse analysis serves as the main foundation for this research. Language influences how phenomena are perceived and understood (Gee, 2025). This study looked at how AI is portrayed, assessed, and positioned within educational discourse by concentrating on language use, including word choice and patterns of description. This method emphasizes how meaning is created via regular communication by moving the analytical focus from observable activities to the interpretive processes ingrained in language (Gee, 2025). This study uses framing theory in conjunction to discourse analysis to look at how AI is portrayed and understood in educator narratives. According to framing theory, an issue's description affects how it is perceived since framing emphasizes some features of a phenomenon while downplaying others (Chong and Druckman, 2007).

Discourse analysis focuses on how patterns of expression, framing, and representation create meaning by using language as the main analytical unit (Gee, 2025; Johnstone and Andrus, 2024). Discourse analysis is combined here with aspects of framing and metaphor analysis in an integrated analytical method. With an emphasis on three major aspects: first, discursive patterns in word choice and expression; second, framing AI in terms of roles and functions; and third, metaphorical representations that convey evaluative or interpretive meaning.

A mixed-methods survey was conducted in four independent school districts in 2025 in Texas, USA. After acquiring approval from the institutional review board of the Texas A&M University-Corpus Christi, the survey was conducted online through Qualtrics. Participation was voluntary and informed consent was placed at the beginning of the survey. Participants were recruited in collaboration with school leaders. A total 409 open-ended responses made up the qualitative component, in which participants discussed their opinions, experiences, and viewpoints regarding AI. A discourse-oriented coding approach (Miles et al., 2014) was used to direct the analytical process to find patterns in the language-based construction of AI. The coding method was iterative, required systematic interaction with the data to improve categories and ensure consistency (Miles et al., 2014; Saldaña, 2021). The units of analysis comprise phrases, sentences, and expressions within educator narratives.

Of the 409 respondents, most discourse about AI came from elementary and secondary teachers, with administrators and staff making up much smaller proportions. The majority of those expressing contradictory framings were teachers at the elementary level (see Table 1).

When the patterns are viewed by education level, most respondents in every group held either a bachelor's or master's degree, and these two groups accounted for the overwhelming majority of each discourse pattern (see Table 2).

Only a handful of respondents with less than a bachelor's degree appeared in any pattern, and those with doctorates made up a small share of all groups. Across all patterns, most respondents were from Generation X (born 1965–1980; aged 46–61) and Millennials (born 1981–1996; aged 30–45), who together accounted for the majority in each discourse pattern (see Table 3).

Baby Boomers (born 1946–1964; aged 61–79) and Gen Z (born 1997–2012; aged 18–28) respondents made up much smaller proportions throughout, with Gen X slightly more prominent than Millennials, especially in the “AI as a Tool” and “Shortcut or Cheating” categories. Respondents were fairly distributed by experience (see Table 4).

Every discourse pattern reflected a wide span of teaching experience. Perspectives on AI are not limited to any one stage of career. Patterns by professional development experience reveals additional differences (see Table 5).

Most respondents had not attended any prior AI literacy professional development. Those who did attend were slightly more common among the “Helper/Assistant” and “Shortcut or Cheating” groups, and made up around forty percent of each category. A detailed description of each discourse pattern is provided in the following section.

AI as a Tool is characterized by language that emphasizes function, professional control, normalization, and bounded utility. Educators frame AI as an instrument to be managed, shaped, and evaluated by human expertise. Linguistically, this pattern relies on instrumental metaphors (e.g. “tool”, leverage’, “asset”) and conditional modal constructions (e.g. “can be used”, “must analyze”).

Educators frame AI as a practical tool as one elementary teacher put it: “Being able to use the correct prompts while understanding that the answers are based off what we ask and can have some errors within it.” Another educator describes it as a source for: “Collection of data to improve your job.” For a secondary teacher: “Computer generated research and study ability is the core, the task is not mystified, but absorbed alongside other resources.” Another educator elaborates on scale: “AI is a way to leverage your work, find trends to help make decisions, and improve productivity.”

The focus on function and operational value emphasized multiple times as one secondary teacher mentioned: “There are many ways to use AI in education.” Others remarked: “Grading essays or creating assignments,” “creating individualized PLAFFPs and goals for special education,” and: “Generating outputs that improve efficiency in my work.” One administrator mentioned: “Teacher training in effective uses of AI to leverage their workload is important.” Normalization, not revolution, defines the tone. An elementary teacher acknowledges: “Absolutely loves it and [has] been showing my coworkers the benefits.” Another writes: “Where a computer generated things and grades things for you.” Another mentioned: “Assist in creating a bulletin board project to display” or “help format CSV files.”

Professional control has recurred on multiple occasions as one secondary teacher puts it: “It's a tool to be used by those who understand how to use it.” Others insist: “The human must analyze the validity of the outputs,” and: “Understanding how AI works and using it in a responsible way.” Administrators and teachers alike urge that: “Teacher training is essential,” and: “I need to be taught about AI at the basic level and then continue to learn how to best use these resources in the classroom.” Limits and cautions are acknowledged but they become professional challenges to solve: “How can it be used safely where students don't abuse it?” “Is my input saved or shared with others?” “I do not extend 100% reliance upon AI.”

There is recognition of bias, privacy, and the need for data guardianship: “Keeping student information safe,” “I do not have trust issues with AI I think it is important for people to understand the term AI is being used very generally and almost incorrectly.” Concerns are raised about creativity: “It will be very difficult for AI to know and feel what a student actually needs to succeed in instruction…” Educators also expressed AI realistically: “It has its plus and minuses… New technology should be accepted and embraced to further knowledge retention.” Others highlight: “Practice with prompting; how to use for assessment; how to incorporate with library and reader responses.” Many stretch on AI's role: “Used correctly, AI can be a valuable tool for student engagement and learning. With the help of AI, and with proper training for students and teachers, teachers can meet the needs of all of their students.” An elementary teacher sees continuity: “AI is already present in our day-to-day lives and to attempt to ignore or safeguard against it entering education is a losing battle.”

The AI as a Helper/Assistant pattern is characterized by anthropomorphic framing, relational language, positive affect, and bounded enthusiasm. Educators construct AI as a supportive partner. This pattern relies on metaphors that express both optimism and the need for responsible oversight (e.g. “assistant”, “tutor”, “colleague”). Evaluative language includes terms such as “helpful”, “exciting”, “incredible”, and conditional modal constructions include “can help”, “must be taught”, and “should keep up”.

A secondary teacher describes AI: “Very helpful when planning engaging lessons,” and an elementary teacher characterizes AI: “A computer-based assistance that helps guide you through questions and also helps you construct your ideas into clear meaning.” Some educators frame AI as: “Like having an assistant that can help students when I'm not available,” “It acts like a tutor,” and even: “A robot helping give advice to teachers for reading.” And: “Supplementing instructional presence,” and making a teacher's reach: “More immediate and accessible.”

Educators stress on AI's benefits for efficiency and individualized support: “AI is an incredible tool that can be a thought assistant and increase workflow efficiency.” An elementary teacher highlights: “I love using AI. I use it for editing messages to parents, ideas on how to handle tricky situations, and for ideas for how to help students in small groups. I love it but I am concerned about the data security.” Another educator remarked: “AI can help tailor content to individual student needs… It helps teachers reach diverse learners more effectively,” and: “AI tools can provide accommodations like text-to-speech, language translation, or simplified summaries for students with disabilities, English Language Learners, or students who need additional support.”

Positive affect permeates these accounts. Educators call AI: “Very helpful,” “exciting,” “a great benefit,” “absolutely incredible,” “mind blowing what kind of options we have,” and “I am very excited to see what the future holds with AI!” One secondary teacher stressed: “When taught and used correctly, it can help build students' confidence. They can use AI to enhance their original ideas and quickly see their vision come to life.”

Other educators also caution: “If students were able to use this feature, I fear that they would rely on it too much and lack coming up with their own creative ideas,” and another mentioned: “Time and training are the most beneficial to utilizing any new tool in education.” Issues of privacy and data security recur: “I am concerned about the data security,” “how do I keep my data secure?” and: “I want to learn where the data is stored, and if I can control what my data is used to train the AI model.”

Educators emphasize: “Professional development, hands-on practice, and curriculum guides with AI activities incorporated,” “I just need more time to play with it and explore,” and “Teachers should keep up with the technology to benefit the classroom and student learning.” Maintaining educator oversight is essential: “Helpful at creating STAAR-like questions… but students must be taught how to use it to learn, not to cheat.” One teacher articulates a central balancing act: “AI helps both students and teachers, but it needs to be used responsibly.” Educators also imagine AI as a future partner as one secondary teacher acknowledges: “AI is not going away and is here to stay. As educators, we must be willing to embrace it and teach our students how to utilize it responsibly.” Another concludes: “I believe AI is the future and whether you like it or not, you better learn to adapt.”

The AI as a Threat or Risk pattern is characterized by language of fear, distrust, loss, dystopian framing, and outright resistance. Educators construct AI not as a neutral tool or helpful assistant, but as a disruptive force that endangers authentic learning, professional identity, privacy, and student development. Linguistically, this pattern relies on strong negative evaluative language such as, “scary”, “dangerous”, “hate”, “lazy”, “bad idea”. Dystopian metaphors like “Rise of the machines,” “I, Robot”, and modality expressing impossibility or deep caution for example, “can't trust,” “there's no way to build true trust,” “nothing would convince me”.

A secondary teacher voices concern: “AI in education is truly the wrong path. I find it scary, not because I am worried about losing my position, but because it is not always correct.” The same respondent laments the loss of authentic interaction: “Certain subjects NEED the human interaction to better explain the material. Kids, today in America, NEED human interaction.” Concerns over the loss of real-world skills are echoed by an elementary teacher: “I absolutely HATE technology in the classroom. I don't like it. I won't use it. Students today have so much technology at their fingertips, but they don't want to use it to LEARN information, they want to play games.”

Concern over AI is made vivid by one secondary teacher's remark: “I think it is a lazy tool for lazy teachers.” Others call AI: “A bot [that] will give you information when you ask it a question not a source of deep learning, but of surface-level, easy answers.” An elementary teacher offers a dystopian twist: “Rise of the machines…” and another invokes fictional Hollywood sci-fi movie fears: “Have you seen I, Robot?”

Mistrust and concerns about control proliferate like: “Useful but potentially dangerous… There's no way to build true trust because it requires taking information and storing it then using it again to produce better results.” Educators shared concerns on privacy: “Confidentiality concerning FERPA laws… errors in coding that produce incorrect information that can be harmful.” An elementary teacher shared: “Students are already at risk in more ways than we can protect them. AI is another danger.”

The erosion of authentic learning and critical thinking emerges as a major concern. Educators’ caution: “AI should not be used in the classroom except for generating ideas, which are then retooled by a real person.” One warns: “They are able to get college degrees, and professional jobs using these features without even picking a book up! That's scary.” Others detect a loss of critical skills: “If AI does the thinking for them [the older students] how will they develop critical thinking skills?” There is often a sense that technology is outpacing the people meant to oversee it. “It feels like it is going faster than I can keep up with,” shares an elementary teacher while, another expressed: “I'm not interested in AI for use in school… This tech is crippling the minds of this generation.” Even interest or curiosity is couched in caution as one elementary teacher mentioned: “I am not sure. I know AI is artificial intelligence, which is already a negative thought in my mind… I would like to learn everything about it, but to be honest, I do not think anything would convince me to trust it.”

The AI as a Shortcut or Cheating pattern is characterized by moral framing, anxiety about eroded effort. Metaphors include “crutch”, and “nostalgia for authentic, human-centered learning”. Educators construct AI not as a neutral tool but as a threat to educational integrity, originality, and the very purpose of schooling. Linguistically, this pattern relies on strong moral evaluative language such as “cheating”, “plagiarism”, “lazy”, “backseat”, and “loss of integrity”. Modality of warning includes “could”, “would”, “need to”, and “should”, which together express both concerns and conditional acceptance.

Educators worry that AI: “Makes it easy to get the work done without thinking,” another explains: “Teacher could stop trying and solely rely on AI.” For one elementary teacher the concern is about: “Technology recreating something for you without you having to put much thought into it.” A secondary teacher puts it directly: “If used wisely, it will be a good tool. For the lazy, it could take the place of real teaching.”

Some educators see AI as eroding effort and independent thought. One secondary teacher describes: “My students view Gen AI as a way to cheat. Some students have the idea, ‘Why try this myself if AI will do it for me?’ Some students use AI to take a backseat in their learning.” This disengagement worries many: “Students feel they are able to ‘check out’ because the computer does the work for them, so the students become disengaged despite having an amazing set of tools in AI.”

A secondary teacher warns: “Students would use AI as a learning crutch rather than an assistive tool in learning.” Several elementary teachers extend the logic: “I worry the next generation is going to rely on AI for everything and not be able to think for themselves,” and: “I think AI will take away basic literacy. People need to learn how to read and summarize for themselves and be able to comprehend.” The moral framing here is vivid, educators describe the loss of educational integrity: “Students use it to cheat on homework, in-class activities and assignments, math problems, essays, and every thought they have. They are not using their brains because AI does everything for them.” Another educator sums up: “It means less thinking and less creative thought that people have to do on their own. Using AI is really a form of plagiarism and cheating.”

There is a real nostalgia for authentic human effort and personal interaction. One seasoned secondary teacher reflects: “Students need to have human interaction to survive and learn… students would not learn how to interact with other people [or] learn how to resolve conflict, etc.” Another educator mentioned: “Students being able to cheat on written essays and exams… Nothing can replace the human experience and caring personalities that teachers have while they are educating our youth.” Some do acknowledge that AI can be a valuable tool for teachers for planning, activities, or feedback. But strict line is drawn when it comes to student work as one secondary teacher explains: “It is great for teacher use, but not for students. Students need to learn how to use this as a productive tool, not a method of copying word for word.”

Several teachers and staff members point to broader, more systemic fears about technology's role in education. One educator observes: “All technology adoption is inevitable, but we will incur the costs of this new technology, just like we did with social media, the Internet even cars decades ago. All technology has taken a toll on basic human ability.” This sense of inevitable change is often paired with specific anxieties about creativity and authenticity, as another teacher confides: “I do worry… about how using AI could limit student creativity. I really do fear that my students will use AI to write a eulogy or their wedding vows instead of what comes from their heart because it's easier and more efficient.”

Some educators fear rapid adoption of AI is outpacing ethical and developmental safeguards. As one teacher puts it: “It's not going away so teach responsible use.” Others stress: “We need more professional development that incorporates time to use the tools and create useful products that can be used in the classroom.” But hope is also shown: “You can use AI in a way that expands learning. but the biggest issue is the time needed to ensure students engage in the learning process and not just shortcut it.”

The Contradictory Framing pattern is characterized by linguistic oscillation, ambivalence, the recurring trope of balance, expressions of uncertainty, and pragmatic hesitation. Educators construct AI not as a fixed entity but as an unsettled, contested site of meaning simultaneously promising and perilous, useful and dangerous, necessary and avoidable. Linguistically, this pattern relies on contrastive conjunctions like “but”, “however”, “yet”, and “although”. Modality of uncertainty comprises “I don't know,” “I'm not sure,” “it depends,” “I can't say for sure”, and the pervasive metaphor of “balance” as a way to negotiate competing demands.

Contradictory responses are marked by linguistic oscillation. A secondary teacher stated: “AI can be very beneficial and save teachers time… It is possible that students will use AI to do all the work without actually learning. There needs to be a balance.” Many responses begin in one register and quickly pivot: “I'm okay with it, but it needs to be a slow integration,” notes a secondary teacher, while an elementary teacher mentioned: “I am interested in implementing AI in my mathematics courses I am not afraid of it but need the time to work with the tools to gain knowledge and skills.” Others offer back-to-back statements of hope and caution: “It has massive potential to improve educational outcomes. I just can't say for sure yet.” or, “I see the possibility for individualized instruction, but not in our current classroom design.”

A common theme is the pairing of pragmatic recognition with hesitation or lack of trust. Typical claims include: “I have no idea what I should be concerned about, except plagiarism.” Or: “The more I learn about AI, the more I understand how it could be misused and potentially lead to major negative life events.” Some educators voice excitement about efficiency and support: “It helps us work smarter not harder, and I have found this to be true.” Yet insist on limits: “That they do not depend on AI to do all their critical thinking. I hope that we still live in a world full of problem solvers for generations to come.”

Contradiction is sometimes expressed within a single narrative arc. One educator noted: “I think it could really help to personalize learning and present it in a format that students like and are comfortable with,” but then pivots: “I honestly don't know what would help me use it more. More time to use and interact with AI in a meaningful way… I find myself not using any AI because it's easier for me to keep doing what I've done and what works.”

Concerns about authenticity, ethics, and unintended consequences run throughout. One secondary teacher suggested: “We need to come to an agreement on best practices. If it cannot be used on state testing, then we obviously cannot let AI write their essays. However, if it can help them create unique projects, then it should be allowed within reason.” Other educators express suspicion: “I'm not sure if there's anything that would make me completely comfortable with giving my data to AI, but I do believe it's the future and has many benefits.”

The discourse of contradictory framing also reflects a sense that the ground is shifting quickly: “It is growing so quickly it is hard to keep up with… I know with the right guidance it can only help enhance lessons and save time, but despite the advances of AI, the human interaction is crucial and cannot be replicated by a machine.” Educators often end with questions or admissions of their own uncertainty: “I don't know enough about it to use it to its full potential,” or simply, “I am still learning.”

The findings indicate that AI is constructed through multiple, and often competing, discourses. Educators draw on wide-ranging linguistic frames to describe AI, including neutral, supportive, cautionary, and moralized constructions. The presence of patterns such as AI as a tool, AI as a helper, and AI as a threat illustrates that educators engage with AI through diverse interpretive lenses. These lenses shape how AI is positioned within classroom practice, whether as a functional resource, an active support system, or a potential source of disruption. Importantly, the pattern of Contradictory Framing demonstrates that these interpretations are not mutually exclusive. This multiplicity of discourse highlights that AI is not simply experienced as a technological object but is actively constructed through language. Furthermore, the coexistence of supportive and critical language suggests that educators are engaged in an ongoing negotiation of AI's role in education.

The identified discursive patterns align with broader theoretical perspectives that conceptualize language as a central mechanism through which social and educational realities are constructed (Nieto, 2001; Zheng, 2025). In this study, educators' descriptions of AI illustrate how language constructs AI as a tool, assistant, risk, or ethical concern, thereby producing distinct meanings and interpretations (Granström and Oppi, 2025). Frames function by highlighting certain aspects of a phenomenon while downplaying others, thereby guiding perception and understanding (Polletta and Ho, 2006).

Each pattern encodes a distinct set of assumptions about purpose, trust, and risk. Recognizing this connection is essential for bridging discourse-level insights with practice-level applications. Below are specific suggestions that are closely related to our findings:

1. AI as a tool: Educators describe wanting to use AI for grading, drafting, and data-informed decision-making. But they also enquire about fundamental operational questions about data storage and safety. The instrumental frame positions educators as the evaluative gatekeepers of AI use, but the underlying data suggests this gatekeeping role is unevenly supported. While this tool pattern does not explicitly transmit educators' underlying evaluative judgments, it nonetheless fosters students' procedural competence.

2. AI as a helper/assistant: Students are likely to extend the kind of interpersonal trust that educators themselves describe extending to AI. A number of educators combined their excitement with a clear warning that students should be taught AI to use it ethically. This pattern holds educational value for students who benefit from assistive support, including text-to-speech, translation tools, and scaffolded writing strategies like interactive writing (Pletcher et al., 2026).

3. AI as a threat or risk: Fear and loss are not arbitrary but center on specific, tangible concerns including data privacy, compliance with the Family Educational Rights and Privacy Act (FERPA), and the potential decline of critical thinking. Educational institutions should address the specific conditions that provoke threat-oriented discourse about AI. Educators who mistrust AI are less likely to model transparent use, which may drive students to use AI secretly rather than refrain from it altogether (Gonsalves, 2025).

4. AI as a shortcut or cheating: While some educators are comfortable with AI for lesson planning and feedback, they remain wary of AI use in student-generated work. Through school-university partnership arrangements at the residency and clinical stages, both pre-service and in-service teachers from early on can develop the ability to articulate task-based distinctions by engaging in guided practice (Reinhardt and Shaikh, 2025).

5. Contradictory framing: Professional development need not push educators toward a settled “for” or “against” position on AI. Educators are not paralyzed by indecision; on the contrary, they are already navigating decisions about AI in their classrooms. The priority, therefore, is not the elimination of ambiguity but the provision of practical strategies for exercising sound judgment in real time (Jiang and Meng, 2025). For students, there is a distinct benefit when teachers acknowledge uncertainty regarding AI rather than reducing it to a binary of good or harmful.

Professional development cannot be limited to technical proficiency. It needs to address the frame-specific language educators use daily. Whether that language encourages trust, discourages disclosure, or draws moral limits around student work, it is the primary vehicle through which educators shape student understanding. Professional development programs should therefore treat discourse awareness as a key competency alongside technical AI literacy. Future research should further examine how different discursive frames influence educational practice, policy development, and institutional responses to AI. This includes investigating how language varies across contexts, roles, and experience levels, as well as how particular framings become dominant or contested over time.

This study demonstrates that the meaning of AI in education is not fixed or inherent but is socially constructed through teacher discourse. The ways in which educators describe AI, whether as a tool, assistant, threat, or form of shortcut, shape how it is understood and positioned within classroom contexts. These constructions are not merely descriptive but they actively influence how AI is interpreted, evaluated, and potentially integrated into educational practice.

This study reveals that educators' language around AI captures underlying tensions that are not always visible through behavioral or attitudinal measures alone. The coexistence of supportive, cautionary, and moral framings within educator narratives indicates that AI is experienced as a complex and contested phenomenon. Language provides a window into these tensions, reflecting concerns about learning quality, academic integrity, and the evolving role of technology in education. The presence of contradictory framing highlights that these tensions are not resolved but are continually negotiated through discourse.

We are thankful for the cooperation of various educators who participated in our research.