Teaching in the Age of Generative AI: A Qualitative Analysis of Faculty Identity, Emotion, and Assessment Concerns

A.IDENTITY AND PROFESSIONAL SELF-PERCEPTION

Erikson’s [14] identity theory focused on the development of a coherent sense of self and the negotiation of roles within societal contexts. Faculty identity in higher education is closely tied to pedagogical authority, assessment design, and intellectual sovereignty. These dimensions shape how faculty perceive themselves as professionals entrusted with cultivating critical thinking, creativity, and academic integrity. A study [15] emphasized that teachers’ professional identities are shaped by their roles as knowledge providers and assessors. The introduction of AI tools that can perform these tasks has led to concerns about diminished professional relevance.

Recent empirical studies illustrate growing concern among instructors regarding generative AI’s potential impact on their professional identity. For example, a mixed-methods study [16] with 178 faculty members at a US university found that although instructors generally recognized the utility of generative AI, many expressed anxiety that AI might erode their distinct human-centered teaching capacities—such as instructional nuance and critical responsiveness—even if they were not outright replaced.

Similarly, [17] surveyed faculty across 76 countries and highlighted tensions between perceived pedagogical benefits and ethical unease; instructors acknowledged GenAI’s efficiency but worried it could dilute pedagogical ownership and undermine their sense of authorship. Studies using technology acceptance frameworks further contextualize these identity concerns. By using a UTAUT2 model, [18] report that although performance expectancy and ease of use predicted intentions to adopt AI tools, policy mandates and the need to redesign teaching practices significantly increased stress and undermined faculty confidence in their professional efficacy. In parallel, education scholars have shown that anxiety over changing disciplinary norms and assessment methods can heighten faculty’s sense of professional vulnerability, even where job security is stable [9].

Collectively, these recent studies reveal a dual dynamic faculty view generative AI as a potentially supportive tool, yet simultaneously perceive it as a force that may destabilize traditional dimensions of pedagogical authority and intellectual autonomy. These tensions are manifested in emotional experiences such as professional apprehension, diminished sense of ownership over curriculum, and ethical uncertainty. This underscores the importance of exploring how faculty negotiate their professional self-concept amidst AI-induced pedagogical shifts.

B.EMOTIONAL RESPONSES TO AI INTEGRATION

Affect theory emphasizes the pre-conscious, embodied responses individuals have to change, highlighting subtle shifts in emotional experience [19]. In the context of higher education, the introduction of generative AI tools has elicited a range of emotional reactions from teachers. A Spanish study titled “Are University Teachers Ready for Generative Artificial Intelligence? Unpacking Faculty Anxiety in the ChatGPT Era” [20] examines specific sources of faculty anxiety related to AI adoption. It identifies three primary anxiety domains: concern about student misuse (e.g., plagiarism), fear of inaccurate AI-generated content, and anxiety over professional role change. The findings reveal that anxiety about student misuse and AI inaccuracies significantly reduces teachers’ intention to adopt AI tools, mediated by lower performance expectancy. However, anxiety pertaining to broader professional change (e.g., the devaluation of teaching roles) did not significantly affect adoption intention—likely due to high job security in Spanish public universities. Similarly, [21] observed that educators reported feelings of inadequacy and fear of obsolescence as AI began to encroach upon traditional teaching roles. These emotional responses underscore the relevance of affect theory in understanding how teachers process and react to technological disruptions in their professional environments.

These affective tensions are exacerbated by the speed at which AI technologies are being integrated into university systems, often with little consultation or pedagogical support. Emotions are not private states but are socially and institutionally mediated [22]. Instructors’ anxieties about AI are shaped by broader narratives of progress, efficiency, and innovation—discourses that frequently sideline the emotional and relational dimensions of teaching [23]. As such, the growing emphasis on “digital transformation” in education risks further alienating teachers from their core pedagogical values and practices [10].

C.COGNITIVE DISSONANCE IN TEACHING PRACTICES

Cognitive dissonance theory [24] posits that individuals experience psychological discomfort when confronted with conflicting beliefs or behaviors, motivating them to resolve the inconsistency. In the realm of higher education, the rise of AI presents such conflicts for educators. A study [25] highlighted that teachers experienced cognitive dissonance when their established pedagogical beliefs were challenged by the capabilities of AI, leading to a reevaluation of their teaching methods. Similarly, [3] reported that educators struggled to reconcile their traditional assessment practices with the ease of AI-generated student work, prompting a reassessment of academic integrity standards.

Conceptual study [8] introduced the notion of GenAI-induced cognitive dissonance, where faculty and students experience psychological tension when using AI tools that conflict with their core pedagogical values like originality, academic effort, and intellectual ownership. The study explores how reliance on AI for writing, feedback, or lesson design can trigger dissonance between professional identity beliefs and pragmatic efficiencies. It recommends strategies such as reflective pedagogy, transparent AI use, and task redesign to mitigate this dissonance. The framework explicitly connects cognitive dissonance with emotional disruption that manifest as discomfort, ambivalence, or resistance among educators adapting to AI integration. These instances illustrate how cognitive dissonance theory can elucidate the internal conflicts faced by educators as they navigate the integration of AI into their teaching practices.

D.ASSESSMENT CHALLENGES IN THE AI ERA

Assessment theory [26] conceptualize assessment as a central element of the learning process, shaping both instructional practices and student outcomes. In the context of emerging AI technologies, this framework underscores the urgency of rethinking conventional assessment practices in higher education. Their work suggests that traditional approaches often fail to capture the complex forms of learning fostered by AI tools, thereby strengthening the case for adopting more authentic and formative assessment strategies.

A study [27] highlighted that teachers are concerned about the authenticity of student work when AI-generated content is prevalent, leading to challenges in maintaining academic integrity. These concerns align with Boud and Falchikov’s assertion that assessment practices must evolve to remain relevant and effective in the face of technological advancements.

A study [28] surveyed 358 university teachers across the Middle East, exploring their views on using GenAI for student assessment. While many faculty adopted AI tools to streamline grading and feedback, the study revealed deep concerns about academic integrity, fairness, and the erosion of disciplinary judgment. Distrust in AI-mediated assessment arose particularly when institutional policies were unclear or ill-defined, exacerbating anxiety about delegating evaluative authority (performance expectancy, effort expectancy). Faculty underscored the need for ethical guidelines and professional development to mitigate these affective tensions.

In a mixed-methods study [3] of 178 US-based instructors, Lyu et al. highlight that faculty often report simultaneous trust and distrust toward GenAI: they recognize its potential but also fear misrepresentation, hallucination, and over automation of assessment tasks. This ambivalence fuels anxiety around the reliability of AI-driven graded feedback and raises concerns about losing control over evaluative criteria. Their qualitative analysis illustrates how assessment distrust contributes to emotional disruption, prompting some instructors to limit AI usage or modify assessment design strategies.

A survey [29] of faculty and students in Vietnam and Singapore found persistent skepticism toward AI-generated feedback, particularly regarding grading reliability and fairness. While AI support was generally accepted when paired with human oversight, fully automated grading tools were met with distrust. The study further highlighted assessment-related anxieties tied to unclear policies, opaque detection systems, and fears of misgrading or bias, pointing to the importance of institutional frameworks for building confidence in AI-assisted assessment.

Across these studies, a consistent pattern emerges: educators experience emotional disruption when AI integration forces a reevaluation of trusted pedagogical practices and professional roles. Cognitive dissonance is especially salient, as faculty navigate the tension between their educational values and the technological affordances of immediate efficiency or automation. Finally, assessment-related anxiety surfaces strongly in contexts where AI is used for grading or feedback without institutional clarity, raising concerns about fairness and academic integrity.

E.GAPS AND CONTRIBUTION

Current research on faculty experiences with GenAI—mostly quantitative or mixed-methods—focuses on their intentions, satisfaction, and perceived usefulness. However, few studies delve into the deeper emotional and identity-related aspects. We don’t have much qualitative insight into how faculty truly feel—the frustration of seeing their teaching identity shift, the stress over grading and assessment, or the internal conflict between their educational values and the pressure to work more efficiently. Additionally, most studies don’t incorporate frameworks like emotional labor or affect theory to make sense of these experiences.

In short, this research highlights AI’s expanding role in higher education and the mixed feelings faculty have about it, where hope clashes with emotional unease. While existing studies acknowledge this tension, they often neglect the deeper human dimensions, as few explicitly draw on emotion-centered theories—such as affect theory, or identity theory—to examine how faculty experience AI integration in the classroom.

A.RESEARCH DESIGN

This study adopts a qualitative exploratory design to investigate faculty members’ affective barriers to adopting AI in higher education. This approach is suitable for exploring complex emotional and psychological responses that may not be fully captured through quantitative. Semi-structured interviews were conducted to understand how university educators emotionally and cognitively navigate the rise of generative AI in their teaching practice. Each session lasted approximately 35 minutes and was recorded with participants’ consent. Thematic analysis was used to code transcripts, and emerging patterns were reviewed iteratively.

B.PARTICIPANTS

The participants for this study were recruited through the official WhatsApp alumni group of Saint Joseph University of Beirut (USJ). An invitation message describing the purpose of the study and the voluntary nature of participation was sent to all 364 members of the group. The message included information about the study topic, confidentiality procedures, and an invitation to express interest in joining an online interview (Appendix A).

A total of 55 members responded to the initial invitation, and each was screened based on predefined inclusion and exclusion criteria. The inclusion criteria required that participants be currently teaching in a higher education institution, regardless of discipline, contract type, or geographic location. The exclusion criteria eliminated retired instructors.

After screening the 55 volunteers, 20 participants met the inclusion criteria and were invited to take part in the study. The sample size of 20 participants was deemed appropriate for achieving thematic saturation in qualitative research. All participants were actively teaching at either the undergraduate or postgraduate level. Fourteen out of the 20 participants were actively teaching in Lebanon and 6 in the gulf region. Those who agreed received a unique Zoom meeting link for scheduling the interview. All interviews were conducted online, recorded with participant consent, and subsequently transcribed. For consistency in analysis, interviews originally conducted in Arabic or French were translated into English prior to coding and thematic analysis.

The final sample represented a diverse group of university instructors from different disciplines, academic ranks, and teaching contexts (Table I). This diversity enhanced the richness of the data and allowed for a more comprehensive understanding of faculty affective barriers to AI adoption in higher education.

Table I. Participants’ codes and disciplines

Table II. NVivo-style thematic coding framework

Table III. Samples of teachers’ dissonance and conflicting cognition

Table IV. Assessment distrust linked to assessment theory of Boud & Falchikov

Table V. Relating fear of cheating and surveillance to Foucauldian theory

Table VI. Summary of affective patterns

C.DATA COLLECTION AND ANALYSIS

Data were collected through semi-structured interviews conducted via Zoom or in-person between March and June 2025 and ranged from 20 to 35 minutes. Questions (Appendix B) focused on:

• 1.Erosion of Pedagogical Identity
• 2.Emotional Disruption and Cognitive Dissonance
• 3.Assessment Anxiety and Distrust

Interviews recorded per consent of participants were transcribed and analyzed using thematic analysis [5]. It follows six-phase framework: (1) familiarization with data, (2) generation of initial codes, (3) searching for themes, (4) reviewing themes, (5) defining and naming themes, and (6) producing the report.

Thematic saturation was monitored continuously throughout the analysis of faculty interviews on affective barriers to AI adoption in higher education. Each interview was transcribed and coded immediately, and a saturation tracking table was used to record the emergence of new codes related to emotional, psychological, and attitudinal barriers toward AI. During the first 12 interviews, core themes—such as anxiety about AI replacing pedagogical roles, lack of confidence in using AI tools, fear of losing academic integrity, and uncertainty about institutional expectations—were identified. After interview 13, no new themes emerged, indicating that theme saturation had been reached. The remaining interviews (14–20) provided additional depth, nuance, and variation within the established affective barriers, contributing to meaning saturation (Fig. 1). This approach follows the recommendations of [30,31] who distinguish between theme saturation (no new themes) and meaning saturation (full understanding of theme complexity). Therefore, the final sample of 20 interviews was judged sufficient to capture both the breadth and depth of faculty affective barriers to adopting AI in higher education.

Fig. 1. Thematic saturation.

NVivo 14 software was used to assist with data organization and coding NVivo is a qualitative data analysis (QDA) software used to organize, code, and analyze unstructured data in this place we mean interviews. Thematic coding in NVivo which refers to the process of systematically identifying, categorizing, and interpreting recurring themes (patterns of meaning) in data was used.

Initial codes were generated inductively from the transcripts and then grouped into themes reflecting emotional and psychological barriers. Each theme was grounded in multiple participant excerpts and validated through both inter-coder agreement and member checking. Interviews were audio-recorded, transcribed by turbo scribe application, and anonymized. Inter-coder agreement was established by a second researcher independently coding 20% of the transcripts, with a 91% agreement rate after discussion. (Appendix C).

Reflexive journaling was used to monitor researcher subjectivity, and member checking was conducted with six participants to validate key themes (Appendix D).

Reliability was ensured through:

• •Inter-coder agreement (a second coder will analyze 20% of transcripts)
• •Reflexive journaling by the researcher
• •Member checking to validate interpretations

A.THEME 1: EROSION OF PEDAGOGICAL IDENTITY

Many participants expressed a sense of diminishing authority and confidence in their pedagogical roles. For example, one literature professor noted, ““I used to feel like the source of knowledge. Now I feel redundant.” (Participant 20). Others voiced concern that AI tools undermine the intellectual labor that teaching seeks to cultivate. This erosion of confidence often manifested in self-doubt, with several participants questioning whether their assessments or teaching strategies were still “meaningful.” Participants described feeling emotionally depleted by the constant need to reinterpret their role. The rapid pace of AI developments, combined with inconsistent institutional guidance, led to a sense of pedagogical whiplash. A history professor shared, “Every month it feels like the rules have changed again. What am I even assessing now—knowledge, creativity, or prompt engineering?” This ambiguity left many teachers feeling isolated and unsupported.

The most frequently coded theme reflected a deep sense of ontological disruption among faculty members. Many described the sudden emergence of LLMs as not just a technical challenge but a threat to their professional identity. Participants spoke of feeling “displaced,” “less essential,” or even “substitutable” in the wake of AI’s ability to generate coherent, sometimes excellent responses to student assignments.

Two core subthemes emerged:

• •Loss of Professional Purpose: Teachers reported struggling to define their role when students can bypass critical thinking tasks with a prompt.
• •Shift from Expert to Facilitator: Many reflected on a reconfiguration of their identity—from knowledge transmitters to moderators of process or ethics.

This theme intersects with identity theory [3] and role strain theory [32], suggesting that faculty are navigating competing role demands between educator, evaluator, and now, AI ethicist. Erikson’s identity theory [3] frames identity as a lifelong, dynamic process shaped by psychosocial challenges. In times of major disruption, like the rise of AI in education, individuals may experience identity crises, prompting re-evaluation of their roles, values, and sense of purpose. This theory is highly applicable to the emotional and psychological challenges university teachers face today. For example: “If I’m no longer the expert in the room, in other words, if AI can ‘teach’ content then what’s my role? That’s an identity question, not just a pedagogical one.”

As for role strain theory [32], emotional and cognitive stress occurs when expectations within a single role are too demanding or contradictory. It explains emotional disruption, burnout, and resistance among educators by revealing:

• •Why many feel overwhelmed despite being committed to their profession.
• •How changing technologies introduce intra-role conflict (within the role of ‘teacher’).
• •The psychological cost of institutional expectations that may not align with everyday realities.

B.THEME 2: EMOTIONAL DISRUPTION AND COGNITIVE DISSONANCE

This theme captured the emotional ambivalence and internal dissonance many faculty members experience in response to LLMs. While a few embraced the “potential” of generative AI as a tool, the majority described a disorienting mix of fascination, fear, skepticism, and guilt. This tension produced what several participants described as “ethical fatigue,” wherein educators felt torn between institutional pressures to innovate and their own professional values “It’s exciting, but I also wake up at night worrying what this means for my job.” (Participant 12)

Two primary subthemes were identified:

• •Conflicted Emotions Toward AI: Faculty found themselves intellectually intrigued but emotionally destabilized.
• •Disorientation and Unease: A majority expressed feeling like they were operating without a “moral compass” or “instruction manual.”

These emotional responses resonate with cognitive dissonance theory [24], wherein teachers’ long-standing pedagogical values now clash with a new technological reality. Cognitive dissonance theory is a foundational psychological theory that explains the emotional and mental discomfort people feel when they hold two or more contradictory beliefs, values, or attitudes, or when their behavior is inconsistent with their beliefs. This discomfort—called dissonance—motivates people to restore consistency, often by changing their beliefs, justifying their behavior, or minimizing the conflict. I use ChatGPT to help with marking rubrics, but I still feel uneasy when students use it for essays. It’s like I’m part of the problem I’m trying to avoid.” (Participant 16).

C.THEME 3: ASSESSMENT ANXIETY AND DISTRUST

This theme revealed a deep distrust of traditional assessment methods and anxiety about authentic student engagement. Instructors voiced concern that LLMs compromise the integrity of student work, making it difficult to distinguish genuine learning from algorithmic mimicry. “I can’t tell anymore if students actually understand the content or just prompt ChatGPT well.” (Participant 03)

Subthemes included:

• •Fear of Cheating and Surveillance Pressure: Some faculty began contemplating more invasive surveillance tools or oral exams.
• •Devaluation of Academic Work: Others reported a demoralizing sense that “essays no longer matter” or that “grading feels futile.”

This theme aligns with assessment theory [26] and connects with surveillance culture [33], as teachers feel both complicit in and victimized by a system that pushes them toward punitive monitoring rather than trust-based pedagogy. Assessment theory [26] focuses on rethinking the role of assessment in higher education, shifting from a teacher-centered, summative model toward a student-centered, learning-oriented approach. Their work is central to contemporary debates about assessment for learning, particularly in light of challenges posed by digital technologies like AI.

In addition, surveillance culture [33] states that The Birth of the Prison refers to a societal condition in which individuals regulate their own behavior because they are aware they might be watched. It’s not just about being surveilled—it’s about internalizing that surveillance. The following theory is closely reflected by several participants. For example, Grading used to be about assessing understanding. Now it feels like a constant investigation. I’m Googling phrases, checking metadata, looking for signs of AI. It’s draining.” (Participant 14, History)

The findings reveal profound emotional turbulence among faculty as they navigate AI integration, with three dominant themes shaping their affective experiences. First, the erosion of professional identity evokes loss, disorientation, and fear, as educators grapple with a destabilized sense of purpose. This emotional strain directly undermines their self-efficacy, leaving some questioning their role in an AI-augmented classroom. Second, emotional disruption manifests as ambivalence and anxiety, reflecting the tension between recognizing AI’s potential and fearing its implications. This unresolved conflict contributes to psychological fatigue, as faculty expend mental energy reconciling competing emotions. Finally, assessment anxiety—driven by distrust and frustration toward AI’s reliability—triggers a shift toward defensive pedagogy, where educators adopt rigid or overly controlled teaching practices to reassert authority. Collectively, these affective responses highlight a critical gap in AI implementation strategies: while technical training is prioritized, the emotional labor of adaptation remains unaddressed, risking burnout and resistance.