Original Research

Impostor phenomenon in human resource management within the era of artificial intelligence

Faraaz Omar, Tendency Beretu

Received: 02 Oct. 2025; Accepted: 24 Feb. 2026; Published: 03 July 2026

Copyright: © 2026. The Authors. Licensee: AOSIS.
This work is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license (https://creativecommons.org/licenses/by/4.0/).

Abstract

Orientation: The global adoption of artificial intelligence (AI) is transforming organisations, including in South Africa, where human resource management (HRM) professionals increasingly use AI-enabled systems. While AI offers efficiency, it creates psychological challenges, notably the impostor phenomenon (IP), where professionals doubt their competence and credit success to external factors like AI.

Research purpose: This study explored how AI adoption is perceived by HRM practitioners to relate to experiences of the IP and examined the implications of IP for practitioners’ perceived professional confidence in AI-driven contexts.

Motivation for the study: Despite growing research on AI-in-HRM, its psychological impact remains understudied. Addressing this gap is vital to ensure AI adoption enhances performance without undermining professional confidence.

Research approach/design and method: A qualitative design was employed. Semi-structured online interviews were conducted with HR managers in South Africa (n = 15). Data were transcribed verbatim and analysed using thematic analysis to generate key themes.

Main findings: Participant accounts suggest that AI adoption heightened feelings of self-doubt among HRM professionals, with many attributing workplace achievements to AI systems, luck or external support. Participants described IP as eroding their confidence, reducing ownership of achievements and affecting work engagement.

Practical/managerial implications: HR leaders should embed AI training within human resource development (HRD) programmes, implement performance appraisal systems that recognise both human and AI contributions, and design interventions to strengthen self-efficacy.

Contribution/value-add: This study offers interpretive insights into how IP is experienced in AI-driven contexts and offers actionable insights for fostering resilience and adaptability among HRM professionals in South Africa.

Keywords: artificial intelligence; impostor phenomenon; self-efficacy; human resource management; South Africa.

Introduction

The rapid adoption of artificial intelligence (AI) across industries has transformed the way organisations manage processes, talent and decision-making. Human resource management (HRM) has not been immune to this transformation, with AI increasingly deployed in recruitment, workforce planning and performance management (Meijerink et al., 2021). While AI offers efficiency and accuracy, it also introduces unintended psychological challenges, particularly for human resource (HR) professionals whose roles and identities are being reshaped (Huang & Rust, 2021; Tambe et al., 2019). This transformation extends beyond task automation to a fundamental redefinition of professional value and agency. AI systems in HRM often function as black boxes delivering potent outcomes such as identifying top talent or predicting turnover without transparent rationale (Benabou et al., 2024). This opacity creates an attribution ambiguity where it becomes difficult to discern whether success stems from the algorithm’s sophistication or the practitioner’s skilled interpretation and ethical oversight. Consequently, HR professionals may increasingly view themselves as mere operators of intelligent systems rather than as expert diagnosticians and strategists (Pandey et al., 2024). This shift risks marginalising the very human skills such as empathy, contextual judgement and ethical reasoning that are becoming paramount as AI handles analytical tasks (Huang & Rust, 2021). The psychological impact of this role redefinition, particularly in a high-stakes, people-centric function such as HRM, remains critically underexplored.

One underexplored psychological challenge within this technological upheaval is the impostor phenomenon (IP). First conceptualised by Clance and Imes (1978), IP describes an internal experience of intellectual fraudulence where high-achieving individuals, despite external evidence of competence, doubt their accomplishments and fear exposure as a fraud. Contemporary research confirms that IP is not a rare affliction but a prevalent experience among professionals, linked to perfectionism, fear of failure and environments with high performance demands (Bravata et al., 2020; Farahany, 2023). It manifests through chronic self-doubt and a tendency to attribute success to external, unstable factors such as luck, timing or the help of others, while internalising failure (Brauer & Proyer, 2022). Recent research within the South African HR context suggests these experiences are widespread and vary across demographic groups, underscoring the need for targeted interventions (Omar & Beretu, 2026). The consequences are far from trivial, contributing to heightened anxiety, burnout, diminished job satisfaction and the avoidance of career-advancing opportunities (Neureiter & Traut-Mattausch, 2016).

In AI-enabled HRM, managers may perceive their achievements as products of AI systems rather than their own expertise, undermining confidence and self-efficacy (Xiao et al., 2025). We theorise that AI acts as a powerful, novel antecedent to IP through three interconnected mechanisms. Firstly, the attribution problem: the ambiguity in credit assignment between human and machine can fuel a persistent external attribution style, a core cognitive marker of IP. Secondly, the pace of obsolescence: the relentless evolution of AI tools creates a moving goalpost for competence, fostering a perpetual sense of inadequacy and the fear of being found out as technically outdated. Thirdly, the professional identity challenge: as AI subsumes traditional analytical HR tasks, practitioners face an existential uncertainty about their unique value proposition, exacerbating feelings of being an impostor in their own evolving profession.

This dynamic risks eroding professional agency and weakening HR practitioners’ strategic influence at a time when their guidance on the human implications of AI is most needed. Although IP has gained attention in psychology and management research (Bravata et al., 2020), its specific intersection with AI adoption in HRM remains largely unstudied. Initial studies on technology-related self-doubt have not isolated the distinct, intelligent and autonomous nature of modern AI systems or focused on the HR profession, where the stakes of human judgement are particularly high (Safadel, 2025; Zhao et al., 2019).

In South Africa, where organisations are rapidly digitalising while grappling with skills shortages and high unemployment, this gap is particularly pressing (Beretu & Charles, 2025; Modise, 2025). For HR professionals, this context layers additional pressure as they must navigate their own technological adaptation while also managing the fair and ethical implementation of AI in the workforce, often without clear frameworks or support. If HR managers themselves experience intensified impostor feelings in AI contexts, their capacity to foster employee well-being, ensure equitable AI deployment and contribute strategically to organisational effectiveness may be severely compromised.

Therefore, this research sought to address the following questions: (1) How do HRM professionals in South Africa perceive the relationship between AI adoption and their experiences of the IP? (2) In what ways does the IP influence HRM practitioners’ perceived professional efficacy and strategic confidence in AI-driven contexts? Through addressing these questions, this study aims to contribute to HRM scholarship by providing a nuanced, contextually grounded interpretive account of a hidden psychological dimension of technological change, and offers practical considerations for supporting resilience and self-efficacy among HR professionals navigating the era of AI.

Literature review

This review synthesises existing scholarship to build a conceptual framework for understanding the IP within the specific, transformative context of AI-driven HRM. It moves from establishing the core psychological construct to examining the technological disruption of the HR profession, culminating in a critical theorisation of their intersection.

The impostor phenomenon: Conceptual foundations, contemporary antecedents and impact

The IP, first conceptualised by Clance and Imes (1978), describes an internalised experience of intellectual phoniness where high-achieving individuals, despite objective evidence of success, remain convinced they are frauds destined to be exposed. Core characteristics include chronic self-doubt, a tendency to attribute success to external, unstable factors (such as luck, timing or having deceived others), and a pervasive fear of evaluation (Brauer & Proyer, 2022; Bravata et al., 2020).

Contemporary research has significantly expanded our understanding of IP beyond its initial conceptualisation. It is now recognised as a prevalent experience among professionals, not a rare pathology, with recent meta-analyses indicating high prevalence rates in knowledge-work sectors (Bravata et al., 2020). Antecedents are multifaceted, spanning individual factors such as perfectionism and neuroticism (Kaur & Jain, 2022), socialisation experiences that condition self-worth to achievement (Langford & Clance, 1993), and critically, organisational and environmental triggers. Workplaces characterised by high competition, unrealistic expectations, a lack of constructive feedback or cultures that stigmatise failure are fertile ground for IP (Bielenberg et al., 2024; Ménard & Chittle, 2023).

The consequences extend far beyond personal discomfort, constituting a significant human capital and organisational development issue. IP is robustly linked to increased anxiety, stress, emotional exhaustion and burnout (Neureiter & Traut-Mattausch, 2016). Perhaps more insidiously, it leads to behavioural strategies such as over-preparation or procrastination, and the avoidance of new challenges or promotions because of an intense fear of failure and exposure (Kensbock & Stöckmann, 2025). This can stifle innovation, career progression, and ultimately, organisational contribution. Therefore, IP represents not merely a private psychological state but a professional vulnerability that can be activated or exacerbated by specific workplace conditions.

The transformation of human resource management by artificial intelligence: Efficiency gains and existential shifts

The HRM function is undergoing a profound metamorphosis driven by the integration of AI technologies. AI’s applications are now deeply embedded across the employee lifecycle: from algorithmic CV screening and predictive analytics in talent acquisition, to AI-powered chatbots for employee services, and sophisticated data analytics for performance management, engagement monitoring and attrition prediction (Meijerink et al., 2021; Van den Broek et al., 2021). These tools promise unprecedented gains in efficiency, scale and data-driven insight, positioning HR to contribute more strategically to business outcomes.

However, this technological shift precipitates a fundamental renegotiation of the HR professional’s role and identity. As Huang and Rust (2021) theorise, as machines increasingly master analytical and routine tasks, the comparative advantage of human workers must shift towards skills that are intrinsically human specifically, empathy, ethical judgement, complex relationship management and strategic interpretation. This evolution from administrative expert to strategic partner and ethical guardian is not automatic or seamless. The transition creates a potential competency gap and an identity lag, where practitioners may feel their historical expertise is devalued before a new, confident professional identity is fully formed or organisationally recognised (Parent-Rocheleau & Parker, 2022).

This challenge is magnified within the South African context. The nation’s rapid digitalisation agenda coexists with persistent socio-economic challenges, including stark skills disparities, high unemployment and a legacy of workplace inequality (Modise, 2025; Mulalo & Morukhu, 2025). For South African HR professionals, adopting AI is not just a technical upgrade but a complex exercise in navigating fairness, mitigating bias in algorithms and managing workforce transitions while contending with their own adaptation pressures. This unique confluence of technological and social complexity forms a critical backdrop for examining psychological workplace phenomena.

The intersection of artificial intelligence and the impostor phenomenon: Theorising a critical research gap

The convergence of the IP literature and the narrative of AI-driven HR transformation reveals a critical, unexamined space ripe for scholarly inquiry. While technology-induced stress is observed (Tarafdar et al., 2019), the specific, intelligent and autonomous nature of modern AI presents novel psychological challenges that may directly catalyse the distinct cognitive and affective patterns of IP. We theorise that AI may act as a powerful trigger for IP through three interconnected mechanisms, while acknowledging that these may interact with pre-existing individual tendencies.

The attribution problem: Systemic ambiguity and the erosion of agency

The first mechanism centres on the problem of attribution in human-AI collaboration. Many advanced AI systems function as black boxes, producing powerful recommendations such as a top candidate shortlist with limited transparency into their underlying logic (Kellogg et al., 2020). This creates fundamental ambiguity: is a successful hire because of the algorithm’s design, the data scientist who trained it or the HR practitioner’s final, context-informed judgement? Drawing from attribution theory (Weiner, 1985), a consistent pattern of attributing positive outcomes externally like in this case, to the AI system, which systematically inhibits the internalisation of success, a cornerstone of self-efficacy and a key deficit in IP (McDowell et al., 2015). Recent studies on algorithmic management suggest this can lead to a diminished sense of personal agency and skill atrophy, as workers may feel they are riding the algorithm’s coattails rather than exercising expert judgement (Parent-Rocheleau & Parker, 2022). This theorising, however, carries an assumption that AI’s black box nature is universally perceived as opaque and alienating. It is crucial to problematise this: for some practitioners, AI may be seen as a reliable partner that enhances, rather than replaces, their judgement. This study must therefore explore not only the presence of attributional ambiguity but also how it is subjectively experienced and interpreted by HR professionals.

The pace of obsolescence: Technostress and the perpetual becoming

The second mechanism arises from the relentless pace of technological obsolescence. AI tools evolve rapidly, rendering specific technical proficiencies transient. The resulting imperative for continuous upskilling creates a moving goalpost for competence. Professionals may consequently inhabit a state of perpetual becoming, where they feel never fully proficient or legitimately expert as a classic cognitive trigger for impostor feelings (Neureiter & Traut-Mattausch, 2016). This aligns with literature on technostress, which identifies the constant pressure to learn and adapt to new digital systems as a major workplace stressor (Tarafdar et al., 2019). The fusion of technostress with the core IP fear of being ‘found out’ as incompetent creates a potent psychological threat. A critical edge must be applied here: this theorising assumes a linear, deficit-based view of skills. It risks overlooking how organisations and individuals might frame continuous learning as an empowering opportunity for growth rather than a threatening demand. The research must therefore investigate whether the pace of change is experienced as a motivator or a primary source of fraudulent feeling.

The identity challenge: Role ambiguity and the search for value

The third and perhaps most profound mechanism is the challenge to professional identity. As AI assimilates analytical and administrative tasks traditionally central to HR, the profession’s core identity is in flux (Tambe et al., 2019). The mandate to pivot towards uniquely human skills such as empathy and ethical stewardship (Huang & Rust, 2021), while intellectually compelling, can be experientially disorienting. If these softer skills are less tangible, poorly measured or not visibly valued in organisational metrics and reward systems, practitioners may face an identity vacuum (Petriglieri & Petriglieri, 2020). This existential uncertainty of ‘Who am I as an HR professional in an AI-driven world?’ can directly exacerbate feelings of being a fraud or impostor in one’s own field. This theorising critically depends on the premise that a clear, valued new identity has not yet been cohered. It raises important questions: Are organisations actively redefining and valuing the new HR role? How do practitioners narrate their evolving identity? This study must explore the narratives of identity work in progress, not assume a static state of crisis.

While previous studies have found generalised self-doubt in technological contexts (Safadel, 2025; Zhao et al., 2019), none have systematically explored how the specific attributes of AI adoption as well as its autonomous decision-support, rapid evolution and identity-disrupting potential, which cultivate the distinct patterns of IP within a specific profession such as HRM. This gap is especially significant in a transforming society such as South Africa, where high unemployment, skills mismatches, and rapid technological change continue to shape workplace experiences. This study aims to address this gap by critically exploring how the theorised pathways linking AI adoption and employee wellbeing are experienced and narrated by HR professionals, thereby providing a nuanced interpretive understanding of the psychological implications of the AI revolution in contemporary workplaces.

Research design

Research approach and method

This study employed a qualitative, exploratory research design, informed by a constructivist paradigm. This approach was deemed most suitable because the research aims to understand the subjective, lived experiences and meaning-making processes of HR professionals as they interact with AI (Clarke & Braun, 2013). A constructivist paradigm acknowledges that multiple realities are constructed through social interaction and personal interpretation, making it ideal for exploring how individuals ascribe meaning to technological change and internalise feelings of impostorism. Unlike a positivist approach seeking a single objective truth, this design allows for a nuanced exploration of the complex psychosocial phenomena at the intersection of professional identity and technological adoption.

Participants and sampling strategy

A purposive sampling technique was used to recruit participants who could provide rich, information-laden insights into the phenomenon (Patton, 2015). The target cohort was senior HR managers in South Africa who were members of the South African Board for People Practices (SABPP), ensuring a baseline of professional credibility and exposure to strategic HR issues. The specific criteria were: (1) holding a senior HR title (e.g. HR Manager, Head of HR, HR Director); (2) having direct, hands-on experience with AI tools in their HR function (e.g. for recruitment, analytics or employee engagement); and (3) being employed in a large organisation (workforce > 150) to ensure substantial organisational complexity and likely investment in digital HR systems.

Sampling process

The SABPP, acting as a gatekeeper, distributed an initial invitation email to its national database of senior practitioner members. This email outlined the study’s purpose and inclusion criteria. Of approximately 200 invitations, 32 individuals expressed interest. A follow-up screening questionnaire confirmed that 22 met all criteria. Following scheduling coordination and two withdrawals because of time constraints, 15 participants were finally secured for interviews. This represents a final participation rate of ~47% from the eligible, interested pool. The SABPP’s role was strictly limited to the initial, anonymised dissemination of the invitation; all subsequent communication, informed consent and data collection were managed directly by the research team to ensure ethical independence. Participants’ demographics are summarised in Table 1.

Data collection

Data were collected via semi-structured online interviews conducted on Microsoft Teams, each lasting 45–70 min. An interview guide was developed not as a rigid script but as a flexible framework to ensure key theoretical domains were explored. The guide was constructed by adapting core psychological constructs from the validated Clance Impostor Phenomenon Scale (CIPS) and contextualising them within the AI-in-HRM landscape. This ensured questions probed the relevant dimensions of IP such as fear of evaluation, attribution style, sense of fraudulence and fear of failure within the specific technological context theorised in the literature review.

To substantiate the link between the guide and our research problems, examples of probe questions include:

• Attribution problem: ‘Can you walk me through a recent work success where AI played a role? How did you explain that success to yourself and what part did you feel was your contribution versus the tool’s?’
• Pace of obsolescence and fear of exposure: ‘How do you keep up with new AI developments in HR? What feelings arise when you hear about a new tool or feature you’re not yet familiar with?’
• Identity challenge: ‘How has using AI in your daily work changed how you view your role as an HR professional? What feels most valuable about your work now compared to before AI?’

All interviews were video-recorded with consent and transcribed verbatim by a professional service, followed by a meticulous accuracy check by a research team member. Transcripts were anonymised, with all identifying details removed.

Data analysis

The two researchers engaged in collaborative sense-making, comparing their initial codes and discussing interpretations until a shared understanding was reached. This process was not intended as a reliability measure in a positivist sense, but rather as a reflexive dialogue to enrich interpretation and challenge individual assumptions, consistent with the reflexive thematic analysis approach (Clarke & Braun, 2013):

• Phase 1–2: Familiarisation and initial coding: Two researchers independently immersed themselves in the data by reading and re-reading transcripts. Initial coding was conducted manually using NVivo software to manage the data. We employed a hybrid approach: primarily inductive, allowing codes to emerge directly from the data (e.g. crediting ChatGPT, feeling like a middleman), but sensitised by deductive theoretical concepts from our framework (e.g. external attribution, identity dissonance).
• Phase 3–4: Generating and reviewing themes: The two researchers compared their initial codes, discussing discrepancies until consensus was reached, enhancing intercoder reliability. Through an iterative process of sorting, clustering and mapping codes, five preliminary themes were identified. These were rigorously reviewed against the entire dataset such as checking if they worked for all transcripts and if the coded data formed a coherent pattern. This led to refinement; for instance, initial separate codes on pressure to perform and fear of mistakes with AI were consolidated into the richer theme of ‘Performance perfectionism’.
• Phase 5–6: Defining themes and producing the report: The essence of each theme was finalised and named. Critically, the analysis did not stop at identifying discrete themes. Through constant comparison and collaborative discussion, we examined the relationships between themes. For example, we observed how data coded for external attribution often directly preceded expressions of eroded self-efficacy, and how social comparison intensified feelings captured in fear of exposure. Mapping these relationships on whiteboards allowed us to conceptualise the interconnected thematic structure that later formed the basis of the theoretical model (Figure 1). This model is thus an interpretive synthesis of the thematic analysis, representing our understanding of how these experiences coalesce to foster IP.

FIGURE 1: Thematic map illustrating the interpretive synthesis of participant accounts of artificial intelligence-related impostor feelings in human resource management.

Trustworthiness

To ensure trustworthiness within the constructivist paradigm, we adhered to the following enhanced principles:

• Credibility: Beyond prolonged engagement, we employed peer debriefing sessions with academic colleagues outside the core team to challenge interpretations. Member checking was conducted by sharing a summary of key themes with three participants, who confirmed their resonance with the findings.
• Transferability: We provided a thick description (Geertz, 1973) through detailed participant demographics (Table 1), a rich contextual backdrop of South African HR, and extensive verbatim quotations in the results. This allows readers to assess the potential transferability of findings to similar contexts.
• Dependability: A clear audit trail was maintained, including records of all coding decisions, meeting notes from analytical sessions and iterative versions of the thematic map.
• Confirmability and reflexivity: A dedicated reflexivity statement was crucial. Prior to analysis, we documented our preconceptions (like hypothesising that AI would create anxiety). During analysis, we maintained reflective journals, held regular meetings to bracket these assumptions and actively searched for disconfirming evidence. For instance, we specifically looked for narratives where AI boosted confidence, ensuring our analysis was data-driven, not confirmation-biased.

Ethical considerations

Ethical clearance to conduct this study was obtained from the University of the Western Cape, Humanities and Social Science Research Ethics Committee (Ref. No. HS23/6/85) and the SABPP. Informed consent was secured from all participants. Anonymity and confidentiality were maintained by using participants’ codes and storing data on password-protected servers.

Results

The thematic analysis generated five interconnected themes that, taken together, illustrate how participants make sense of AI adoption in relation to their experiences of the IP. These themes form what participants described as a reinforcing dynamic, detailed next. An overview is provided in Table 2.

Theme 1: External attribution of success and the diminishing self

A dominant and pervasive pattern was the tendency to attribute positive work outcomes externally, predominantly to the AI system itself, rather than to one’s own expertise in deploying it. This was not a casual acknowledgement of a tool’s utility but a fundamental discounting of personal agency. For instance, P3 (Manufacturing) reflexively diminished her role:

‘I’m not that amazing hey. I just use ChatGPT stuff. It does the clever part, I just ask the questions.’ (P3, female, 8 years experience, manufacturing)

This sentiment was echoed by P8 (Consulting), who observed it systemically:

‘Practitioners are finding it difficult to accept that their abilities are the reason for success. They are simply attributing their performance to the AI, as if they are just the button-pusher.’ (P8, female, 14 years experience, consulting)

Even in cases of successful strategic decisions, the narrative shifted. P10 (Energy), with 20 years’ experience, reflected:

‘When our predictive model flagged a retention risk we’d missed, and the intervention worked, the board praised the “smart system.” I sat there thinking, “But I’m the one who designed the intervention protocol.” The credit just flows to the tech.’ (P10, male, 20 years experience, energy and utilities)

This consistent external attribution directly enacts a core cognitive component of IP, preventing the internalisation of accomplishment.

Theme 2: Erosion of self-efficacy and professional identity

Closely linked to and often stemming from external attribution was a palpable erosion of professional self-efficacy. Participants described a devaluation of human judgement in the face of algorithmic output. P12 (Logistics) stated:

‘If the AI can analyse 10 000 CVs in minutes and rank them, what’s my “professional intuition” really worth? It starts to feel like a guess against a calculation.’ (P12, male, 16 years experience, logistics and transport)

This challenged the foundation of their professional identity. P5 (Mining) articulated this identity struggle:

‘I built my career on being a good judge of people and situations. Now I’m told the algorithm’s judgement is more objective. It makes you question your entire career’s worth. Am I just an outdated interface?’ (P5, female, 18 years experience, mining)

The data suggests that when AI performs the ‘analytical heavy lifting’, the human contribution can be perceived and felt as secondary, eroding the confidence necessary for strategic efficacy.

Theme 3: Fear of exposure and the perceived competency gap

The data revealed acute anxiety about being ‘found out’ as lacking the requisite knowledge for an AI-augmented role. This fear was heightened by the ‘smart’ reputation associated with data-driven work. P4 (IT) shared:

‘My manager introduced me as our “AI champion” to a client. I think she oversold me … I panicked. I know how to use our tools, but I don’t understand the machine learning behind them. What if they ask a deep technical question?’ (P4, male, 10 years experience, information technology)

This perceived gap between expected and actual knowledge was widespread. P7 (Public Sector) observed:

‘There’s an assumption that if you work with these systems, you’re also a data scientist. The fear isn’t just of failing, but of being exposed as not being the tech-whiz everyone assumes you are.’ (P7, male, 25 years experience, public sector)

This fear of exposure is a hallmark of IP and was directly tied to the specialised, often opaque, nature of AI.

Theme 4: Performance perfectionism and negative recall bias

Possessing powerful AI tools paradoxically raised the internal performance benchmark to unattainable levels, fostering a cycle of perfectionism. The belief emerged that with such advanced support, failure was unacceptable. P6 (Healthcare) explained:

‘If you have a tool that’s supposed to eliminate bias from hiring and you still get a discrimination complaint, the feeling isn’t “the tool has limits,” it’s “I failed to use it perfectly.”’ (P6, female, 12 years experience, healthcare)

This led to a negative recall bias, where minor setbacks loomed large. P14 (FMCG) recounted:

‘We used an engagement survey tool with sentiment analysis. One team’s score dipped. Even though 9 improved, all I could think was, “The AI gave me this insight and I still couldn’t fix that one team.”’ (P14, female, 19 years experience, FMCG)

The tool, intended to aid, became the measure against which their personal adequacy was harshly judged.

Theme 5: Social comparison and perceived inferiority

Artificial intelligence proficiency emerged as a new, potent axis for social comparison among peers, intensifying feelings of inferiority. Participants described a competitive, sometimes judgemental, environment. P10 observed:

‘There is fear in the profession that those who remain poorly skilled in AI may be outperformed or seen as obsolete. You’re constantly checking what tool your colleague is using.’ (P10, male, 20 years experience, energy and utilities)

P2 (Retail) described the personal impact:

‘My junior colleague attended a workshop on prompt engineering and now gets “better” answers from the same chatbot. I feel like I’m being lapped by someone half my age and experience. It’s incredibly demoralising.’ (P2, male, 15 years experience, retail)

This comparison was exacerbated by the rapid pace of change, fuelling insecurity about future relevance.

Interconnections and the thematic map: The interpretive coherence of the thematic map (Figure 1) lies in these perceived interconnections as described by participants. For example, social comparison (Theme 5) was often narrated as feeding fear of exposure (Theme 3) and intensifying performance perfectionism (Theme 4). It is this perceived synergistic interaction of themes, vividly present in participants’ accounts, that forms the basis of our thematic map, illustrating how AI is experienced by HR professionals as a multifaceted context within which IP-related feelings may be activated or amplified.

Discussion

This study set out to investigate the underexplored nexus between AI adoption and the IP among HR professionals in South Africa. Moving beyond speculative commentary, our findings offer interpretive insights into how AI integration is perceived by HR professionals as a significant organisational trigger for IP, activating and amplifying feelings of self-doubt through a network of interrelated psychosocial mechanisms. The thematic map presented in Figure 1 synthesises these participants’ accounts, illustrating how AI adoption is experienced in relation to five interrelated themes: external attribution of success, erosion of self-efficacy, fear of exposure and inadequacy, performance perfectionism, and social comparison and inferiority. Drawing on participant narratives, these themes appear to interact in ways that cultivate persistent feelings of fraudulence, although the cross-sectional nature of the data means these relationships should be understood as interpretive rather than causal.

Theorising the pathways: Support and nuance

Our findings strongly support the three pathways theorised in the literature review, while adding crucial nuance. Firstly, the attribution problem was not merely hypothetical; it was a lived reality for participants. The tendency to credit success to AI (Theme 1) directly undermined self-efficacy (Theme 2), corroborating McDowell et al. (2015) and illustrating how AI’s black box nature (Kellogg et al., 2020) can systematically disrupt the internal attribution of success necessary for professional confidence. However, our data also revealed a nuance: this external attribution was not always passive. Some participants, such as P8, actively resisted this narrative, suggesting that organisational culture and leadership communication play a critical role in shaping attributional styles.

Secondly, the data vividly confirmed the pace of obsolescence as a key stressor. The pervasive anxiety captured in Theme 3 (Fear of Exposure) and Theme 5 (Social Comparison) was frequently linked to the relentless need for upskilling. This aligns with technostress literature (Tarafdar et al., 2019) and confirms that the moving goalpost of competence is a powerful trigger for IP-related inadequacy. Notably, this anxiety was equally present among seasoned professionals (such as P2, P10), challenging the assumption that experience inoculates against technological insecurity. This suggests AI-induced IP is a systemic, rather than individual, vulnerability linked to the velocity of change itself.

Thirdly, and most profoundly, the study validates the professional identity challenge. Themes 2 and 4 (Erosion of Self-Efficacy and Performance Perfectionism) point directly to an identity crisis. Participants struggled to articulate their value as AI subsumed analytical tasks, echoing the identity vacuum theorised by Petriglieri and Petriglieri (2020). This was not a simple skills gap but an existential renegotiation of the HR role. The finding that this crisis affects senior professionals (Table 1) is particularly significant, indicating it threatens the strategic core of the HR function, potentially weakening its influence at a time when ethical and human-centred AI governance is most needed.

Contextualising the findings: The South African lens

The South African context is not a neutral backdrop but an intensifying lens. The pressures of digital transformation intersect with deep-seated socio-economic challenges such as skills disparities, inequality and high unemployment (Modise, 2025; Mulalo & Morukhu, 2025). For our participants, adapting to AI was compounded by the responsibility to implement it fairly in an unequal society. This dual pressure likely exacerbates the performance perfectionism (Theme 4) and fear of exposure (Theme 3) observed, as professionals feel the weight of getting a high-stakes technological transition right amid complex social dynamics. The national scope of our sample, with consistent reports across provinces, indicates that this is a widespread national challenge, not an issue isolated to tech-centric hubs such as Gauteng.

Critical reflections and model implications

While our model posits AI as an antecedent to IP, we must critically engage with the assumption of unidirectional causality. The data suggest a recursive relationship: pre-existing impostor tendencies may also shape how an individual perceives and engages with AI, potentially leading to avoidance or over-dependence. Furthermore, our study, by design, captured perceptions of challenge and threat. Future research should investigate contexts where AI tools are successfully framed as empowering ‘co-pilots’, potentially mitigating IP feelings. Our model, therefore, represents a predominant pathway observed in our data but is not an inevitable deterministic law.

Practical implications: From diagnosis to intervention

While interpretive and context-bound, the findings suggest several considerations for practice that move beyond generic AI literacy:

• Reframing AI training as identity work: Organisations must design AI training that explicitly addresses the psychological and identity-level impacts uncovered here. Training should champion the AI-human partnership, clearly delineating and valuing the ‘uniquely human’ skills of ethical judgement, empathy and contextual intelligence (Huang & Rust, 2021), thereby directly countering the identity crisis (Theme 2 & Theme 3).
• Redesigning performance systems for attributional clarity: Performance management must evolve to evaluate and reward the ethical orchestration, critical interpretation and strategic application of AI insights, not just the outputs. This creates clear criteria for success that professionals can internalise, directly combating external attribution (Theme 1) and rebuilding self-efficacy (Theme 2).
• Cultivating psychologically safe cultures: To mitigate the stresses of the ‘pace of obsolescence’ (Theme 3 & Theme 5), organisations should foster cultures where discussing technological anxiety and skill gaps is normalised. Structured peer-mentoring circles and leadership role-modelling of vulnerability in learning can build collective resilience and reduce the stigma that fuels IP.

Limitations

This study has several limitations that should be acknowledged when interpreting its findings. Firstly, while the qualitative design facilitated an in-depth exploration of subjective experiences, it limits the statistical generalisability of the results. Secondly, the purposive sampling strategy focused on senior HR managers who were members of the SABPP and worked in large organisations. Consequently, the perspectives of junior HR practitioners, professionals in small and medium-sized enterprises or non-SABPP members are not represented, which may affect the transferability of the findings. Thirdly, the reliance on self-reported data through interviews introduces the potential for social desirability bias, where participants may have under-reported feelings of inadequacy or impostorism. Fourthly, the absence of baseline measures of IP prior to AI adoption means this study cannot establish causality or definitively isolate AI as a distinct antecedent. It remains possible that AI integration functions primarily as a contextual amplifier of pre-existing impostor tendencies rather than a novel causal trigger; accordingly, the findings should be interpreted as capturing participants’ perceptions of AI’s role in their experiences of self-doubt rather than as evidence of a deterministic causal relationship. Finally, the cross-sectional nature of the data provides a snapshot in time and cannot capture how these psychological dynamics evolve as AI adoption matures within organisations or as individuals gain greater familiarity with AI tools over time.

Future research

The findings and limitations of this study point to several productive avenues for future inquiry. To examine the prevalence and strength of the relationships suggested by our thematic map, large-scale quantitative surveys could be conducted across diverse HR populations in South Africa and other contexts, testing the extent to which the patterns identified here hold across broader samples. Longitudinal studies are essential to track the evolution of AI-induced impostor feelings over time, examining how they fluctuate with increased familiarity, training interventions and organisational support. Furthermore, comparative research across different industries, organisational sizes and national cultures would help to identify unique contextual stressors and protective factors. Finally, experimental or intervention-based studies could test the efficacy of specific strategies such as reframed AI training or redesigned performance metrics in mitigating the IP and fostering a more resilient, AI-confident HR profession.

Conclusion

This study offers a compelling, empirically grounded interpretive account of how the adoption of AI in HRM is perceived by professionals as a significant psychological and professional challenge, one that can trigger or amplify feelings associated with the IP. Through examining attributional ambiguity, identity threat, and obsolescence anxiety and how they interact in the South African context, this research broadens the understanding of AI’s workplace impact beyond a focus on technological benefits and skills gaps. It reveals a hidden layer of human cost in the digital transition. For AI integration to be sustainable and human-centric, a proactive, psychologically informed approach is not optional but essential. Organisations must foster a new, confident professional identity for HR: one that sees human judgement not as obsolete, but as the essential ethical complement to the power of AI.

Acknowledgements

This article is based on research originally conducted as part of Faraaz Omar’s doctoral thesis titled ‘A study to understand the feelings of impostor phenomenon among human resource professionals in selected South African organisations’, submitted to the Faculty of Economic and Management Sciences, University of the Western Cape in 2025. The thesis was supervised by Prof. Bright Mahembe. The supervisor was not involved in the preparation of this article and thus was not listed as co-author. Portions of the thesis have been revised, updated and adapted for publication as a journal article. The authors affirm that this article complies with ethical standards for secondary publication, and appropriate acknowledgement has been made of the original work.

The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.

Faraaz Omar: Conceptualisation, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Writing – original draft, Writing – review & editing. Tendency Beretu: Conceptualisation, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Validation, Writing – original draft, Writing – review & editing. All authors reviewed the article, contributed to the discussion of results, approved the final version for submission and publication, and take responsibility for the integrity of its findings.

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

The data that support the findings of this study are not openly available because of reasons of sensitivity and are available from the corresponding author, Faraaz Omar, upon reasonable request.

The views and opinions expressed in this article are those of the authors and are the product of professional research. They do not necessarily reflect the official policy or position of any affiliated institution, funder, agency or that of the publisher. The authors are responsible for this article’s results, findings and content.

References