Between mirrors and walls: symbolic disputes and algorithmic reconfiguration in accounting

Abstract

This theoretical essay examines how the diffusion of artificial intelligence (AI) may reconfigure the epistemic and institutional boundaries of accounting under specific conditions of data, governance, and academic incentives. Drawing on a critical review, thematic analysis, and integrative synthesis, it connects classic and recent literature across four axes: the sociology of scientific fields, philosophy of science, the psychodynamics of work, and AI applications in auditing, financial reporting, and tax classification. Four propositions are advanced: (1) AI expansion tends to reward hybrid research that integrates substantive theory, interdisciplinary design, and out-of-sample validation; (2) programs combining accounting and data science may expand institutional impact while facing legitimation barriers; (3) models that are robust to automated validation tend to gain recognition when supported by transparency, substantive interpretation, and governance; and (4) algorithmic ubiquity may strain paradigm fragmentation when institutional incentives foster interdisciplinary co-authorship. Early adoption experiences suggest context-specific operational gains, while regulators and recent research emphasize transparency, bias auditing, and data documentation. The essay concludes that accounting is called to shift from a logic of walls to a logic of bridges, aligning symbolic capital with practical relevance, methodological pluralism, and social responsibility.

Keywords:
artificial intelligence; accounting; boundary-work; narcissism of minor differences; symbolic capital

Resumo

O ensaio teórico avalia como a difusão da Inteligência Artificial (IA) pode reconfigurar fronteiras epistemológicas e institucionais da ciência contábil sob condições específicas de dados, governança e incentivos acadêmicos. Com base em revisão crítica, análise temática e síntese integrativa, articula literatura clássica e recente em quatro eixos: sociologia dos campos científicos; filosofia da ciência; psicodinâmica do trabalho; e aplicações de IA em auditoria, relato financeiro e classificação fiscal. Sugerem-se quatro proposições: (1) a expansão da IA tende a recompensar pesquisas híbridas que integrem teoria substantiva, desenho interdisciplinar e validação fora da amostra; (2) programas que combinam contabilidade e ciência de dados podem ampliar impacto institucional, mas enfrentam barreiras de legitimação; (3) modelos robustos a validações automatizadas tendem a ganhar reconhecimento quando acompanhados de transparência, interpretação substantiva e governança; e (4) a ubiquidade algorítmica pode tensionar a fragmentação paradigmática quando houver incentivos à coautoria interdisciplinar. Experiências iniciais sugerem ganhos operacionais contextuais, enquanto reguladores e pesquisas recentes reforçam transparência, auditoria de viés e documentação de dados. Conclui-se que a Contabilidade é chamada a deslocar-se da lógica de muros à lógica de pontes, alinhando capital simbólico a relevância prática, pluralismo metodológico e responsabilidade social.

Palavras-chave:
inteligência artificial; contabilidade; boundary-work ; narcisismo das pequenas diferenças; capital simbólico

Resumen

Este ensayo teórico examina cómo la difusión de la inteligencia artificial (IA) puede reconfigurar las fronteras epistemológicas e institucionales de la ciencia contable bajo condiciones específicas de datos, gobernanza e incentivos académicos. A partir de una revisión crítica, análisis temático y síntesis integrativa, articula literatura clásica y reciente en cuatro ejes: sociología de los campos científicos, filosofía de la ciencia, psicodinámica del trabajo y aplicaciones de IA en auditoría, información financiera y clasificación fiscal. Se formulan cuatro proposiciones: (1) la expansión de la IA tiende a recompensar investigaciones híbridas que integren teoría sustantiva, diseño interdisciplinar y validación fuera de la muestra; (2) los programas que articulan contabilidad y ciencia de datos pueden ampliar el impacto institucional, pero enfrentan barreras de legitimación; (3) los modelos robustos frente a validaciones automatizadas tienden a ganar reconocimiento cuando se acompañan de transparencia, interpretación sustantiva y gobernanza; y (4) la ubicuidad algorítmica puede tensionar la fragmentación paradigmática cuando existen incentivos a la coautoría interdisciplinaria. Las experiencias iniciales sugieren ganancias operativas contextuales, mientras reguladores e investigaciones recientes refuerzan la transparencia, la auditoría de sesgos y la documentación de datos. Se concluye que la contabilidad está llamada a pasar de una lógica de muros a una lógica de puentes, alineando capital simbólico con relevancia práctica, pluralismo metodológico y responsabilidad social.

Palabras clave:
inteligencia artificial; contabilidad; boundary-work ; narcisismo de las pequeñas diferencias; capital simbólico

INTRODUCTION

Accounting science is undergoing a transition in which machine-learning algorithms and robotic process automation (RPA) are already being incorporated, at uneven rates, into tasks such as reconciliations, tax classifications, and report preparation (Eisikovits et al., 2025). This technical advance engages with a disciplinary structure still marked by the methodological divisions in the positivist, interpretive, and critical traditions described by Chua (1986); studies of academic evaluation also show that prestigious journals continue to favor established approaches, leaving less room for heterodox agendas and collaboration across paradigms (Fraser & Sheehy, 2020). This practice reflects boundary-work, as mapped by Gieryn (1983), through which scientific communities reaffirm boundaries to protect resources and symbolic authority (Bourdieu, 1984a).

In professional practice, industry surveys indicate that many accountants expect their roles to be reconfigured in the coming years (Bowling, 2024), although the actual adoption of artificial intelligence (AI) varies with regulatory maturity, technological infrastructure, data availability, and governance policies. This unevenness contributes to the recurring perception of a gap between academic research and the needs of practice (Bennis & O’Toole, 2005; Clor-Proell et al., 2025). Under these conditions, fintechs and data-science consultancies may move into areas historically associated with the accounting profession, redirecting economic value and social legitimacy when the field fails to respond at the pace of innovation.

From a psychosocial standpoint, this dynamic resonates with Freud’s (1996) “narcissism of minor differences”: methodological subgroups, despite sharing a common object and vocation, accentuate distinctions to affirm their identities and thereby reduce opportunities for convergence. Research on institutional health suggests that such fragmentation may foster defensive strategies that preserve the status quo while delaying the adoption of emerging practices (Dejours et al., 1994). In this setting, algorithmic tools serve as an epistemological stress test. They reveal the strengths of adaptive research programs and the limitations of lines of inquiry that remain closed to methodological triangulation or large-scale validation.

The essay therefore asks: how do internal symbolic disputes affect accounting’s capacity to integrate artificial intelligence in ways that are relevant and socially legitimate? Its general objective is to understand how algorithmic pressure redistributes symbolic capital and reconfigures disciplinary boundaries. Four specific objectives follow: (1) to map current forms of boundary-work; (2) to analyze them through psychoanalytic and sociological mechanisms; (3) to examine how AI applications expose or mitigate these divisions while avoiding the cargo-cult science denounced by Feynman (1974); and (4) to propose a framework that places Popper (2002), Kuhn (2012), Lakatos (1999), and Feyerabend (2010) in productive tension as lenses for future research.

METHODOLOGICAL NOTE

This essay is neither a systematic review nor a bibliometric study. The literature was selected purposively in light of the theoretical question, combining classic works on science, scientific fields, and work with recent research on AI, accounting, auditing, algorithmic governance, and the professions. Priority was given to studies that: (i) discuss disciplinary boundaries and disputes over authority; (ii) examine symbolic capital, academic recognition, and professional hybridization; (iii) address ethics, transparency, and model validation; and (iv) bring the philosophy of science into dialogue with contemporary accounting phenomena such as continuous auditing, misstatement detection, tax classification, and financial reporting.

The thematic analysis grouped the literature according to convergences, tensions, and gaps and distinguished direct evidence from accounting and auditing from examples used by analogy from adjacent fields. The integrative synthesis recast these themes as theoretical and analytical propositions (P1-P4), not as causal hypotheses that have already been demonstrated. Empirical examples and industry reports are therefore treated as context-specific indications or heuristic illustrations whenever they do not derive from direct accounting research. This distinction preserves the essay’s character and makes its limits of generalization explicit.

The argument proceeds in four stages. The second section revisits work on disciplinary identity, paradigmatic disputes, and institutional defenses. The third argues that AI places pressure on disciplinary boundaries and develops testable propositions about epistemic integration. The fourth discusses implications for research, professional practice, and regulatory governance and suggests ways to mitigate technological risks. The fifth and final section brings the contributions together and outlines an interdisciplinary agenda for integrating algorithms into responsible innovation processes without amplifying latent vulnerabilities.

By positioning accounting “between mirrors and walls,” the essay argues that its future relevance will depend less on accelerating existing processes than on critically revising the symbolic boundaries that still limit internal dialogue and external engagement. Dismantling conceptual walls remains a prerequisite for digital mirrors to reflect not only improved techniques but also purposes aligned with demands for transparency, sustainability, and equity in data-driven environments.

THEORETICAL FRAMEWORK

Disciplinary identity: between technique, scientific ambition, and symbolic capital

The history of accounting reveals a metamorphosis that combines utilitarian practice, professional regulation, and academic aspiration. Recent studies of accounting’s disciplinary boundaries revisit its historical development as a record-keeping practice, a regulated profession, and an academic discipline, highlighting the coexistence of technical, institutional, and scientific functions (Volkova, 2024). In dialogue with Hopwood (1989), the coexistence of these registers can be described as an “amphibious identity”: accounting serves the market by fulfilling regulatory obligations while also serving science by producing critical knowledge.

This ambivalence reappears in vocational and academic debates. Demski (2007) provocatively criticized the strong professional influence on accounting curricula. In this vein, curricula centered on lists of standards, such as International Financial Reporting Standards (IFRS), pronouncements issued by the Comitê de Pronunciamentos Contábeis (CPC), and tax laws, may produce short-term specialists but not wide-ranging thinkers. By contrast, interpretive traditions maintain that accounting practices must be understood through sociological, historical, and political lenses (Hopwood & Miller, 1994). The tension goes beyond curricular rhetoric and shapes how departments set promotion criteria. In many universities, publishing in prestigious finance journals counts for more than influencing local standards, an indication that the symbolic capital associated with “hard” science continues to govern internal prestige.

Bourdieu’s (1984a, 1984b) concept of symbolic capital clarifies this dynamic. To advance within the scientific field, researchers compete for prizes, funding, and, above all, citations. Malsch et al. (2011) examine Bourdieu’s influence on the accounting literature, while Willmott (2011) warns that journal lists can narrow thematic and methodological diversity. In accounting, Vogt et al. (2021) underscore the importance of epistemological vigilance in applying a Bourdieusian approach to the field. These pressures can widen the distance between academia and professional practice when research does not directly address challenges in independent auditing, data governance, or the incorporation of environmental, social, and governance (ESG) standards. Academia, in turn, argues that abandoning methodological rigor would jeopardize its scientific legitimacy. The result is a reputational dilemma: whether to be recognized as an autonomous science or remain a tool that supports regulatory demands.

The struggle for prestige also shapes accounting’s relationship with neighboring disciplines. Studies indicate that accounting doctoral programs adopted agency theories and asset-pricing models to “borrow” statistical rigor from economics (Kaplan, 2011; Lee & Williams, 1999). Although this strategy raised accounting’s standing in interdisciplinary rankings, it also reinforced perceptions of epistemological dependence.

Boundary-work, narcissism of minor differences, and epistemological fragmentation

Boundary-work refers to the discursive and institutional practices that determine who has the authority to speak on behalf of science (Gieryn, 1983). In accounting, it operates at three levels:

1. External boundary: distinguishing accounting science from “mere” business practice. Professional councils emphasize exclusive certifications, while academic departments invoke socioeconomic theories to justify university status.

2. Boundary between subfields: distinguishing financial accounting, management accounting, auditing, and taxation. Each subfield claims distinctive competencies: financial accounting emphasizes command of “hard numbers”; management accounting claims expertise in decision-making; auditing highlights professional judgment and assurance; and taxation draws on technical and regulatory knowledge.

3. Paradigmatic boundary: separating quantitative positivist research, qualitative interpretive research, and critical approaches. The divide is methodological, with regressions and controlled experiments on one side and ethnographies, historical analyses, and critical studies on the other.

Golyagina and Valuckas (2020) show that official representatives of the Institute of Management Accountants (IMA) rely primarily on monopolization strategies, whereas members of the association in Russia mainly pursue expansion strategies. This configuration of power is dynamic and changes with regulation, technology, and market conditions, including financial crises and fraud scandals.

A similar dynamic accompanied the spread of enterprise resource planning (ERP) systems and, later, RPA. Rather than immediately fostering an integrated agenda on information, control, and professional judgment, these technologies were often absorbed as operational tools within particular subfields. Auditing, management accounting, taxation, and financial accounting incorporated them according to their own standards of competence and preserved established symbolic boundaries. The pattern shows how relatively small differences among subfields can obstruct broader sociotechnical agendas even when all of them confront shared problems of data, control, traceability, and accountability.

The narcissism of minor differences intensifies these disputes. Freud (1996) argues that nearly identical groups exaggerate their differences to preserve internal cohesion. In accounting, positivists and interpretivists share the goal of explaining practices of measurement, control, and accountability, yet often clash over “rigor” and “relevance.” Hussain et al. (2020) found that positivist articles draw on a narrower body of references, whereas critical-interpretive and interdisciplinary research uses more diverse sources, revealing bibliographic segmentation within the field. This segmentation further fragments accounting knowledge.

This setting encourages rituals of scientificity without genuine ontological questioning, the phenomenon Feynman (1974) called cargo-cult science. Several authors argue that reproducing earnings management or value relevance models with minor variations generates a substantial volume of incremental research output while leaving digital-era dilemmas unresolved, including who is responsible for algorithmic bias in tax-credit estimates. Accounting thus risks reproducing methodological sophistication without a commensurate increase in its capacity to explain emerging problems.

Extending the analysis of power, Burri (2008) shows, by comparison with radiology, that new imaging technologies can reorder symbolic capital because visualization practices and artifacts become sources of prestige in particular settings. By analogy, the spread of machine-learning tools in accounting creates opportunities for emerging subfields such as data analytics; challenges traditional hierarchies among auditors, controllers, analysts, and tax professionals; and sharpens struggles over boundaries.

Institutional suffering, collective defenses, and organizational cynicism

From the perspective of the psychodynamics of work, Dejours (1999) argues that intense meritocratic pressure causes psychological suffering. Accounting academia exhibits these pressures as publication targets, short review deadlines, and international rankings become instruments of control. Scientific communities respond by activating collective defenses.

1. Methodological formalism: strict adherence to statistical checklists for heteroscedasticity, endogeneity, and robustness may serve as “armor” against the anxiety of producing work with social significance. Critical scholars warn that when method becomes an end in itself, the field may sacrifice creativity and originality (Willmott, 2011).

2. Institutional cynicism: disbelief in the intrinsic value of scholarly production, masked by outward conformity. Dejours (1998) described an analogous phenomenon in French factories; in academia, it appears in corridor remarks such as “one publishes for the sake of publishing” and in minimal submissions made only to meet faculty scoring requirements. Sloterdijk (1983) calls this attitude cynical consciousness, an “enlightened false consciousness” in which agents recognize the system’s contradictions but persist because they believe others will do the same if they do not.

Empirical indicators underscore the seriousness of the problem, although they are not specific to accounting. Evans et al. (2018) reported moderate to severe symptoms of anxiety in 41% and depression in 39% of the graduate students surveyed, rates far above those observed in the general population. A study reported in Nature likewise documented insecurity and distress among postdoctoral researchers regarding career prospects, workload, and workplace culture (Woolston, 2020). Cynicism is more than a personal posture: management studies find a negative association between organizational cynicism and innovative behavior. Firoozi et al. (2016), for example, identified this association among staff in sports and youth agencies. These findings are used here as contextual and analogical evidence for understanding academic pressures, not as a direct measure of the accounting field.

In increasingly digitized workplaces, these defenses may reappear when performance-measurement systems promise neutrality but intensify comparison, control, and pressure for productivity. Automation does not eliminate institutional suffering by itself. It may instead shift that suffering from repetitive tasks to new forms of surveillance, individual accountability, and competition for recognition.

Algorithmic technology, interdisciplinarity, and the future of accounting

Algorithmic systems, including RPA, blockchain, generative AI, and machine-learning models, are redesigning recording, reconciliation, classification, and auditing tasks where data and analytical infrastructure are more highly digitized. Reports on the future of work indicate growing pressure for reskilling, the automation of routine tasks, and changing professional competencies, although these projections should be read as broad trends rather than linear forecasts for every accounting activity (World Economic Forum, 2020). Accounting must therefore reconsider its core expertise: should it remain centered on manual calculation, or move toward critical, contextual analysis of algorithmic outputs?

If internal boundaries remain rigid, each subfield may treat AI as an accessory to its own agenda: financial accounting will use algorithms to forecast earnings; auditing, to sample or examine transactions; management accounting, to optimize budgets; and taxation, to classify transactions. This strategic and fragmented use of AI by different subfields may deepen divisions between methodological paradigms rather than foster interdisciplinary integration. However, AI also exposes the limitations of each isolated approach. Detecting bias in tax-risk classifications, for example, requires collaboration among statisticians, legal scholars, data-ethics specialists, and accounting professionals.

Pioneering initiatives point to possible paths, particularly in financial supervision. The Cambridge SupTech Lab (2024), affiliated with the Cambridge Centre for Alternative Finance, has developed initiatives to support the digital transformation of financial supervision, emphasizing governance, responsible technology use, and accountability in regulatory ecosystems. Because financial supervision is an adjacent field rather than direct evidence about the accounting profession, the example should be transferred with caution. Even so, it offers relevant lessons for model auditing, algorithmic transparency, and accountability in data-intensive environments. These initiatives suggest that accounting can contribute to debates on algorithmic responsibility by turning its tradition of accountability into a comparative advantage.

To seize this opportunity, institutions must revise their incentives by valuing interdisciplinary publications, incorporating digital competencies into curricula, and rethinking performance evaluations that reward volume alone. In Bourdieu’s (1984a) terms, this means redefining what counts as symbolic capital: not only impact factors, but also the capacity to integrate ethical, social, regulatory, and technological dimensions. Under these conditions, accounting can strengthen its position as an applied and critical social science in the digital age, balancing numerical accuracy, professional judgment, and institutional reflexivity.

THEORETICAL PROPOSITION AND ARTICULATION

AI enters accounting as a general-purpose technology, an artifact applicable across multiple organizational domains, and may shift disciplinary boundaries previously seen as stable in subfields more exposed to digital data, analytical infrastructure, and regulatory pressure. Recent reviews map applications in auditing, tax provisions, cash-flow forecasting, risk assessment, and accounting and finance research (Hasan, 2022; Liaras et al., 2024). Preliminary evidence on the performance of language models in professional accounting and auditing examinations also suggests that a substantial share of declarative accounting knowledge has become codifiable. This does not, however, eliminate the need for contextual judgment, professional responsibility, and substantive interpretation (Eulerich et al., 2024).

This finding exposes internal asymmetries. Rule-based curricular content, such as the “manuals” criticized by Demski (2007), is precisely the kind of knowledge that algorithmic systems can most readily reproduce in standardized assessments. Critical competence consequently shifts toward framing problems, interpreting institutional contexts, negotiating ethical ambiguity, and designing hybrid human-algorithm systems. Kaplan (2011) had already identified underinvestment in emerging topics. Competition from low-cost algorithmic outputs widens that gap and presses accounting research to produce insights that machines cannot readily imitate.

In this context, AI places selective pressure on parts of the accounting field. Topics or methods with little incremental value may lose prominence, while transdisciplinary approaches oriented toward concrete problems and interprofessional collaboration gain relevance when they combine infrastructure, data access, and validation. This dynamic encourages networks that bring together classical econometrics, machine learning, text mining, qualitative analysis, and regulatory knowledge. One example is the use of machine learning to detect anomalies in general-ledger data, which may help auditors select higher-risk entries (Bakumenko & Elragal, 2022). Rather than automatically dissolving boundaries, AI makes them more visible and presses subfields to justify their criteria of relevance, validation, and public value.

Research programs aligned with data science may broaden accounting’s practical impact, yet university evaluation systems continue to reward individual metrics and discourage collective effort. This misalignment creates tension between institutional recognition and practical value. Interdisciplinary projects require time for data collection, computational infrastructure, access to proprietary data, and multidisciplinary teams, none of which promotion committees or funding agencies consistently value. Research on professions and AI accordingly suggests that algorithmic systems do not eliminate jurisdictional disputes. They reconfigure those disputes through new forms of boundary-work (Faulconbridge et al., 2025).

As professional boundaries become more permeable, specializations may emerge in algorithm auditing, corporate AI governance, and the verification of explainability in accounting models (Dwivedi et al., 2023). In organizations with digitized data, robotic automation is likely to absorb routine processes, shifting part of accounting work toward curating algorithmic outputs, exercising professional skepticism and contextual judgment, and communicating risk. Curricula must therefore incorporate computational ethics, applied statistics, bias assessment, and data governance so that professionals can validate systems that process large volumes of information continuously.

Philosophical reflection matters less as a survey of canonical authors than as a lens for concrete accounting phenomena mediated by AI. Popper (2002) helps frame severe tests in fraud detection, tax classification, and cash-flow forecasting; Kuhn (2012) helps interpret anomalies revealed by continuous auditing or by applying natural language processing to reports; Lakatos (1999) distinguishes programs that generate new knowledge from merely cosmetic uses of models; and Feyerabend (2010) supports the methodological pluralism needed to combine statistical models, case studies, and institutional critique without dispensing with transparency and responsibility. This approach is consistent with work that already draws on Lakatos to evaluate research programs in management accounting (Modell, 2022).

In bringing these frameworks together, the essay does not attempt to reconcile their philosophies of science fully. Popper (2002), Kuhn (2012), Lakatos (1999), and Feyerabend (2010) do not form a homogeneous epistemological matrix; their tensions are used as analytical devices. On this reading, AI may broaden opportunities for empirical validation, expose anomalies in traditional routines, place pressure on the protective belts of established programs, and require methodological plurality under explicit standards of transparency, responsibility, and public scrutiny. Without these conditions, algorithmic sophistication may simply relocate methodological formalism to new computational artifacts.

These arguments can be summarized as four epistemological effects:

1. Popper (2002): the ability to generate predictions at scale and reproduce them automatically expands the scope for testability in areas such as fraud, accounting misstatements, and tax classification, provided that hypotheses, data, validation criteria, and out-of-sample tests across new firms, periods, and jurisdictions are made explicit.

2. Kuhn (2012): the proliferation of anomalies detected by AI in continuous auditing, notes to the financial statements, sustainability reports, or tax records may intensify crises in paradigms based solely on manual sampling or limited validation, without automatically producing a revolution.

3. Lakatos (1999): programs that integrate machine learning may add progressive auxiliary hypotheses when they increase explanatory or predictive power over substantive accounting problems; lines of inquiry that merely adjust parameters ex post without theoretical gains tend to become degenerative.

4. Feyerabend (2010): the critique of methodological monopolies reinforces the need for openness to combinations of classical statistics, neural networks, qualitative analysis, critical studies, and dialogue with regulators, provided that such plurality is supported by theoretical justification, transparency, and responsibility.

These arguments yield four interrelated propositions:

P1: The more widely AI is used throughout the accounting information lifecycle, the greater the reputational premium is likely to be for research that combines applied empirical evidence, substantive theory, and interdisciplinary dialogue.

P2: Programs that align accounting expertise with data science are likely to have greater institutional impact, but they face barriers to legitimation in traditional systems of academic evaluation.

P3: Accounting models that withstand automated out-of-sample testing are likely to gain scientific legitimacy more quickly when supported by transparent documentation, independent validation, and substantive interpretation.

P4: Widespread AI adoption may reduce paradigmatic segmentation when institutions reward interdisciplinary co-authorship, thereby placing pressure on the narcissism of minor differences.

These propositions describe a setting in which AI neither eliminates accounting nor resolves its internal disputes on its own. Instead, it may redefine excellence by making the connections among empirical validity, practical relevance, data governance, and public responsibility more visible. Academic and professional success is therefore likely to depend on the ability to turn complex problems into collaborative projects grounded in verifiable evidence, attentive to institutional contexts, and subject to human scrutiny within hybrid human-algorithm environments.

DISCUSSION AND IMPLICATIONS

This section revisits the four propositions developed above and discusses their implications for research, professional practice, and accounting’s institutional framework. The analysis connects the redistribution of symbolic capital in Bourdieu’s terms (1984a, 1984b), institutional defenses associated with the psychodynamics of work (Dejours, 1999; Dejours et al., 1994), the boundary-work described by Gieryn (1983), and the risk of empty formalism associated with cargo-cult science (Feynman, 1974). The aim is not to present AI as an automatic solution to disciplinary impasses, but to examine the conditions under which it may reconfigure standards of relevance, validation, and public responsibility in accounting science.

P1: Methodological hybridity, validation, and scientific reputation

AI tools expand the possibilities for empirical testing when versioned datasets, explicit validation criteria, and out-of-sample comparisons are available. In continuous auditing, for example, automated models can examine large volumes of transactions and flag residual patterns that escape traditional sampling procedures, providing empirical material for more precise conjectures (Issa et al., 2016). This process may strengthen progressive research programs in the Lakatosian sense, provided that the models add explanatory or predictive power rather than merely technical sophistication.

Recent research shows that combining machine learning with established accounting theory can illuminate fraud, accounting misstatements, information quality, and user behavior in complex data environments (Bao et al., 2020; Bertomeu et al., 2021; Liaras et al., 2024). Projects driven solely by data, with no engagement with conceptual foundations, can produce spurious relationships that undermine theoretical relevance. Conversely, strictly deductive approaches may fit the evidence poorly when they disregard digital datasets, texts, transaction records, and new forms of evidence.

Methodological hybridity becomes especially valuable in this setting. Triangulating structured data, texts, contextual measures, and substantive interpretation strengthens independent replication and public scrutiny. AI contributes not only by accelerating tests, but also by raising the standard for aligning hypotheses, empirical evidence, validation, and accounting interpretation.

P2: Professional hybridization, data, and academic legitimation

Natural language processing algorithms applied to notes to the financial statements, conference-call transcripts, sustainability reports, and regulatory documents bring quantitative methods closer to interpretive approaches concerned with organizational narratives (Luthfiani, 2024). This technological “common language” may temper the narcissism of minor differences described by Freud (1996), as researchers from different schools begin to negotiate choices about data curation, bias mitigation, explainability criteria, and the interpretation of results.

Research on professions and AI nevertheless indicates that algorithmic systems do not eliminate jurisdictional disputes. They reconfigure them through new forms of boundary-work as professionals and researchers compete for authority over data, models, metrics, and interpretation (Faulconbridge et al., 2025). AI therefore changes what boundaries contain rather than dissolving them. Prestige may increasingly depend on computational infrastructure, data access, documentation, governance, and interprofessional collaboration, while still drawing on established theoretical and methodological repertoires.

These collaborations nevertheless encounter evaluation systems whose bibliometric indicators reward narrow specialization. Data-science projects require lengthy data collection, computational infrastructure, negotiated access to proprietary data, and multidisciplinary teams, none of which promotion committees or funding agencies consistently value (Kaplan, 2011). Impact metrics should therefore account for social usefulness, reproducibility, code transparency, controlled data access, and engagement with regulators and users of accounting information.

P3: Automated validation, documentation, and substantive interpretation

Automated replication across multiple contexts may accelerate the consolidation of more generalizable accounting models when data are comparable, testing criteria are declared in advance, and results can be contested. Studies of fraud detection and accounting misstatements indicate that machine-learning models can improve predictive capacity in complex datasets, particularly when evaluated out of sample and accompanied by substantive interpretation (Bao et al., 2020; Bertomeu et al., 2021). When such evidence helps resolve persistent anomalies, puzzle-solving capacity shifts in a Kuhnian sense, shortening the cycle of theoretical and curricular renewal (Kuhn, 2012).

This recognition should not be confused with automatic validation. Rapid testing can conceal weaknesses when datasets contain uncorrected cultural, sectoral, historical, or regulatory biases. Sustainable legitimation requires model-audit protocols, code versioning, documented analytical pipelines, independent validation, and controlled data access where possible. Although these practices are more established in computer science than in accounting, they are essential if AI-generated evidence is to be evaluated, challenged, and reused by other researchers.

Statistical robustness does not replace accounting interpretation. A model may perform well predictively while reproducing flawed assumptions, institutional asymmetries, or historical patterns of exclusion. The scientific legitimacy of AI-mediated accounting models therefore rests on the combination of empirical performance, substantive explanation, transparent documentation, and responsibility for the consequences of their use.

P4: Interdisciplinary co-authorship, symbolic capital, and governance of work

Research and innovation programs have increasingly incorporated criteria related to impact, collaboration, and the social use of results. In Europe, Horizon Europe documents organize work programs around calls, expected outcomes, impacts, and the scope of research and innovation, reflecting a funding model oriented toward applicability and collaboration (European Commission, 2024). Although this orientation is not specific to accounting, it signals an institutional environment in which collaboration, dissemination, and the potential use of results carry greater weight in project evaluation.

For accounting science, this shift points to a broader understanding of symbolic capital. Academic prestige may remain associated with articles in highly ranked journals, but it is likely to encompass data repositories, auditable code, sectoral observatories, evidence useful to regulators, and collaboration with users of accounting information. Broadening the concept will not end disciplinary disputes. It may redirect them toward criteria such as reproducibility, explainability, information security, social impact, and contestability.

Organizations and industry groups report operational benefits in closing, reconciliation, tax classification, report summarization, and inconsistency detection. These accounts point to potential efficiency gains, but they often come from vendors, consulting firms, or isolated organizational experiences without independent validation. They should therefore be treated as context-specific indications that AI is already pressing accounting to redefine tasks, expertise, and quality standards, not as definitive proof of algorithmic superiority or a uniform trend across the profession.

For the profession, the most important effect may be not only time savings but also the reallocation of accounting judgment. Algorithmic systems can automate repetitive work, yet they increase the need to validate assumptions, audit models, document data, assess bias, and communicate uncertainty. In organizations that use AI more intensively, professional expertise is therefore likely to shift from processing information directly to curating and interpreting results and assuming responsibility for them in hybrid human-algorithm environments.

From the perspective of the psychodynamics of work, automation is not simply a source of operational relief. AI may alleviate suffering associated with repetitive tasks, rework, manual reconciliations, and tight deadlines. When deployed without governance, however, it can create new forms of institutional suffering through more intensive targets, real-time performance surveillance, algorithmic comparisons among professionals, and reduced decision-making autonomy. The question is therefore not whether AI eliminates or deepens suffering, but how particular organizational arrangements redistribute cognitive load, control, and professional recognition.

To capitalize on gains and mitigate risks, the following measures are recommended:

1. Systematic reskilling in data science, AI governance, model auditing, and algorithmic ethics across undergraduate curricula, graduate programs, and continuing education.

2. Transparency and explainability protocols covering data documentation, validation criteria, model limitations, and human responsibility for interpreting results.

3. Collaborative laboratories involving universities, companies, regulators, and accounting firms to enable applied research on real problems in controlled environments.

4. Review of auditing standards and practices so that AI-generated evidence is recognized only when accompanied by safeguards for independence, integrity, security, and contestability.

5. Organizational policies that protect workers by preventing algorithmic productivity metrics from becoming opaque mechanisms of surveillance, comparison, and work intensification.

Limits and risks deserve attention. Reviews of AI-based decision-making in accounting and auditing identify recurring challenges of objectivity, privacy, transparency, accountability, and reliability (Lehner et al., 2022). The uncritical use of algorithms may reproduce the empty formalism identified by Feynman (1974), a situation in which scientific form is preserved without analytical substance. Black-box models make it difficult to contest estimates and may compromise professional skepticism, especially when their outputs are treated as unquestionable technical authority. In addition, historical databases may carry biases that, if not identified and mitigated, perpetuate distortions in credit assessment, loss estimation, risk classification, or compliance decisions.

The psychodynamics of work clarifies another risk: innovations that threaten professional identities, established competencies, and traditional sources of recognition can trigger defensive reactions. Dejours et al. (1994) help explain why some members of a community may reject, weaken, or only superficially absorb technologies that shift boundaries of authority. Without gradual learning, participation, and governance, AI is likely to be received either as a threat to the profession or as a fetish of modernization. Both responses constrain critical adoption.

Mitigating these risks requires model-governance policies, independent peer audits, and critical perspectives that question assumptions embedded in algorithms. Bias audits, dataset documentation through datasheets for datasets (Gebru et al., 2021), and stakeholder participation in system design can reduce opacity and increase contestability. Along similar lines, Saltelli et al. (2020) argue that socially relevant models should be transparent, candid about their limitations, and open to scrutiny. Collaboration between domain experts and AI systems, as Jarrahi (2018) suggests, preserves human interpretive capacity and lends greater robustness to conclusions.

The discussion indicates that responsible AI adoption can help realign accounting research, professional practice, and regulation when it is accompanied by governance, transparency, and changes to institutional incentives. Progress depends on rethinking how legitimacy is measured, supporting open-data ecosystems, and cultivating critical skills among future accountants. Transformation is possible but not automatic. It requires a collective commitment to constructive skepticism, worker protection, public responsibility, and methodological plurality under explicit validation standards.

FROM NARCISSISM TO ALGORITHMIC INTEGRATION

The trajectory examined in this essay indicates that artificial intelligence is more than a technical addition to accounting routines. It exposes accumulated vulnerabilities, places pressure on symbolic boundaries, and prompts accounting to reconsider its standards of scientific relevance, professional legitimacy, and public responsibility. Internally, the discipline remains divided by paradigmatic segmentations that, since Chua (1986), have structured disputes among positivist, interpretive, and critical approaches. Externally, the diffusion of algorithmic systems may shift comparative advantage toward actors able to combine data science, accounting judgment, governance, and ethical sensitivity.

The essay’s central argument is that AI does not automatically dissolve these boundaries. Its effects depend on institutional incentives, governance arrangements, and the field’s capacity to turn symbolic disputes into sociotechnical collaboration. The metaphor of “mirrors and walls” captures this tension: mirrors expose the field’s vulnerabilities, whereas walls represent the disciplinary, methodological, and institutional boundaries that still impede integration among research, practice, and regulation.

Theoretical contributions

The essay makes four main theoretical contributions. First, it brings together boundary-work, symbolic capital, and the narcissism of minor differences to explain why the incorporation of AI into accounting is not merely a technical issue. Resistance to or acceptance of AI involves struggles over authority, prestige, methodological territory, and institutional recognition. The analysis therefore moves beyond an instrumental view of AI toward a sociological reading of the boundaries of the accounting field.

Second, the essay connects the sociology of scientific fields, psychoanalysis, and the psychodynamics of work to interpret accounting’s internal fragmentation. The narcissism of minor differences helps explain why subfields with similar objects of study may exaggerate disagreements to preserve their identities. The psychodynamics of work, in turn, shows that resistance to innovation may reflect suffering, threats to competence, and loss of recognition rather than intellectual conservatism alone.

Third, the epistemological discussion treats Popper, Kuhn, Lakatos, and Feyerabend as lenses held in tension rather than as a homogeneous matrix. Popper frames testability in fraud detection, misstatements, and tax classification; Kuhn helps explain anomalies emerging from continuous auditing and digital reporting; Lakatos distinguishes progressive from degenerative programs in AI-based accounting research; and Feyerabend challenges methodological monopolies that prevent researchers from combining modeling, interpretation, and institutional critique. Together, these perspectives cast AI as a force that broadens the possibilities for validation, reveals anomalies, and questions overly restrictive methodological conventions without turning it into an automatic solution to accounting science’s impasses.

Fourth, propositions P1-P4 establish a testable conceptual agenda. They suggest that the spread of AI may increase the reputational value of hybrid research, favor programs that combine accounting and data science, accelerate the legitimation of models that withstand automated validation, and place pressure on paradigmatic segmentation when institutions reward interdisciplinary co-authorship. The claim is conditional: AI can promote integration only when supported by institutional design, transparent validation, and genuine collaboration.

Practical and professional contributions

In practical terms, the essay suggests that the accounting profession will shift part of its core expertise from processing information directly to curating, interpreting, and assuming responsibility for results produced in hybrid human-algorithm environments. AI systems can automate reconciliations, classifications, summaries, and inconsistency detection, but they cannot replace the professional judgment required to interpret assumptions, assess risk, and communicate uncertainty.

The principal practical implication is that responsible AI adoption requires new professional capabilities. These include applying data science to accounting, auditing models, documenting datasets, assessing bias, ensuring explainability and information security, and governing algorithms. Accounting education should retain its technical and conceptual foundations while integrating them with the sociotechnical capabilities needed to support decisions in data-intensive environments.

The essay also cautions against equating operational benefits with legitimacy. Industry reports point to gains in closing, reconciliation, tax classification, and inconsistency detection, but evidence from vendors, consulting firms, or isolated organizational experiences must be treated carefully. For such gains to make a lasting professional contribution, they require independent validation, traceability, auditing, and mechanisms through which results can be challenged.

AI can also alleviate or exacerbate institutional suffering. It may reduce repetitive work, rework, and pressure from tight deadlines. Without governance, however, it can produce new forms of surveillance, more intensive targets, algorithmic comparisons among professionals, and reduced decision-making autonomy. The essay does not present AI as a redemptive solution. Its practical contribution is to set criteria for responsible adoption that preserve human judgment, professional recognition, and worker protection.

Institutional and curricular implications

At the institutional level, universities, journals, funding agencies, regulators, and professional bodies need to reconsider their criteria for recognition. A traditional model of symbolic capital centered almost exclusively on publications and bibliometric indicators is ill suited to evaluating research that relies on proprietary data, multidisciplinary teams, computational infrastructure, repositories, auditable code, and regulatory impact.

Graduate programs can respond through curricula that connect accounting theory, statistics, data science, digital ethics, and AI governance. They can also encourage preregistration, controlled data access, code versioning, documentation of analytical pipelines, and independent replication. These practices strengthen public scrutiny and reduce the risk of empty methodological formalism.

For journals and reviewers, the proposed agenda calls for recognizing contributions beyond the traditional article. Documented datasets, auditable models, measurement instruments, technical reports, evidence useful to regulators, and projects co-produced with organizations can support a broader understanding of scientific impact. This does not lower academic standards; it expands the criteria through which the relevance of accounting research is recognized.

For regulators and professional bodies, the principal implication is the need for standards of algorithmic governance. AI-generated evidence should qualify as auditable documentation only when it meets requirements for independence, integrity, security, explainability, and contestability. Accounting’s tradition of accountability can become a comparative advantage here because the profession already has an institutional repertoire for dealing with trust, evidence, responsibility, and control.

Future research agenda

The four propositions can inform a range of empirical designs, but they should be read as a theoretical and analytical agenda rather than as conclusions already supported by longitudinal evidence. Bibliometric studies and co-authorship network analyses could test whether the diffusion of AI increases interdisciplinary collaboration and reduces paradigmatic endogeneity. Longitudinal research could examine whether programs that combine accounting and data science achieve greater institutional impact through citations, regulatory adoption, corporate use, or the production of replicable artifacts.

Another promising avenue is to test accounting models across jurisdictions, sectors, and regulatory regimes. Automated out-of-sample validation can show which models are externally robust and which merely reproduce local adjustments. Such studies should combine predictive metrics with substantive interpretation and should not conflate statistical performance with theoretical explanation.

Studies on governance, ethics, and algorithmic responsibility in accounting environments are also needed. Research may examine how auditors, controllers, analysts, regulators, and users of accounting information deal with AI-generated recommendations, especially when opacity, bias, or conflict arises between algorithmic output and professional judgment. Experiments with users of accounting information may assess the effects of AI on trust, comprehensibility, risk perception, and decision-making.

Finally, the psychodynamic dimension opens a largely unexplored research agenda. Qualitative studies, in-depth interviews, and mixed methods could examine how AI changes recognition, autonomy, suffering, and collective defenses in academic departments, audit firms, tax departments, and shared-service centers. This agenda matters because algorithmic transformation changes more than tasks; it reconfigures professional identities, symbolic hierarchies, and forms of belonging.

CONCLUSION

The accounting field faces an institutional choice. It can preserve narrow disciplinary walls, using AI only to accelerate existing routines and reinforce old divisions. Along that path, digital mirrors will reflect and perhaps amplify existing impasses: methodological formalism, distance between research and practice, internal struggles for prestige, and limited engagement with social needs.

The alternative is to turn these mirrors into instruments of reflexivity. Walls should be dismantled when they obstruct collaboration, while standards of rigor should be preserved when they protect the quality of evidence. Accounting can move to the center of debates on transparency, sustainability, governance, and economic justice in data-driven environments if it aligns symbolic capital, public responsibility, and methodological plurality under explicit validation standards.

Between mirrors and walls, the decisive question is not simply whether to adopt or reject AI. The challenge is to use it without surrendering accounting judgment, institutional critique, or social responsibility. Meaningful algorithmic integration will not replace accounting science; it will expand the field’s capacity to explain, audit, and transform practices for the benefit of society.

References

• Bakumenko, A., & Elragal, A. (2022). Detecting anomalies in financial data using machine learning algorithms. Systems, 10(5), 130. https://doi.org/10.3390/systems10050130
  » https://doi.org/10.3390/systems10050130
• Bao, Y., Ke, B., Li, B., Yu, Y. J., & Zhang, J. (2020). Detecting accounting fraud in publicly traded U.S. firms using a machine learning approach. Journal of Accounting Research, 58(1), 199-235. https://doi.org/10.1111/1475-679X.12292
  » https://doi.org/10.1111/1475-679X.12292
• Bennis, W. G., & O’Toole, J. (2005). How business schools lost their way. Harvard Business Review https://hbr.org/2005/05/how-business-schools-lost-their-way
  » https://hbr.org/2005/05/how-business-schools-lost-their-way
• Bertomeu, J., Cheynel, E., Floyd, E., & Pan, W. (2021). Using machine learning to detect misstatements. Review of Accounting Studies, 26, 468-519. https://doi.org/10.1007/s11142-020-09563-8
  » https://doi.org/10.1007/s11142-020-09563-8
• Bourdieu, P. (1984a). Homo academicus Les Éditions de Minuit.
• Bourdieu, P. (1984b). Distinction: a social critique of the judgement of taste Harvard University Press.
• Bowling, S. (2024, July 23). How AI is transforming accounting. Accounting Today https://www.accountingtoday.com/list/how-ai-is-transforming-the-accounting-profession
  » https://www.accountingtoday.com/list/how-ai-is-transforming-the-accounting-profession
• Burri, R. V. (2008). Doing distinctions: boundary work and symbolic capital in radiology. Social Studies of Science, 38(1), 35-62. https://doi.org/10.1177/0306312707082021
  » https://doi.org/10.1177/0306312707082021
• Cambridge SupTech Lab. (2024). SupTech Week 2024: Insights and innovations for digital financial supervision https://cambridgesuptechlab.org/event/suptech-week-2024-insights-and-innovations-for-digital-financial/
  » https://cambridgesuptechlab.org/event/suptech-week-2024-insights-and-innovations-for-digital-financial/
• Chua, W. F. (1986). Radical developments in accounting thought. The Accounting Review, 61(4), 601-632. https://doi.org/10.2308/TAR-4479026
  » https://doi.org/10.2308/TAR-4479026
• Clor-Proell, S. M., Even-Tov, O., Lee, C. M. C., & Rajgopal, S. (2025). Bridging the gap between academia and practice in accounting. Accounting Horizons, 39(1), 1-14. https://doi.org/10.2308/HORIZONS-2024-071
  » https://doi.org/10.2308/HORIZONS-2024-071
• Dejours, C. (1998). Souffrance en France: la banalisation de l’injustice sociale Éditions du Seuil.
• Dejours, C. (1999). A banalização da injustiça social (2ª ed.). FGV.
• Dejours, C., Abdoucheli, E., & Jayet, C. (1994). Psicodinâmica do trabalho: contribuições da escola dejouriana à análise da relação prazer, sofrimento e trabalho Atlas.
• Demski, J. S. (2007). Is accounting an academic discipline? Accounting Horizons, 21(2), 153-157. https://doi.org/10.2308/acch.2007.21.2.153
  » https://doi.org/10.2308/acch.2007.21.2.153
• Dwivedi, Y. K., Kshetri, N., Hughes, L., Slade, E. L., Jeyaraj, A., Kar, A. K., Baabdullah, A. M., Koohang, A., Raghavan, V., Ahuja, M., Albanna, H., Albashrawi, M. A., Al-Busaidi, A. S., Balakrishnan, J., Barlette, Y., Basu, S., Bose, I., Brooks, L., Buhalis, D., ... Wright, R. (2023). Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. International Journal of Information Management, 71, 102642. https://doi.org/10.1016/j.ijinfomgt.2023.102642
  » https://doi.org/10.1016/j.ijinfomgt.2023.102642
• Eisikovits, N., Johnson, W. C., & Markelevich, A. (2025). Should accountants be afraid of AI? Risks and opportunities of incorporating artificial intelligence into accounting and auditing. Accounting Horizons, 39(2), 117-123. https://doi.org/10.2308/HORIZONS-2023-042
  » https://doi.org/10.2308/HORIZONS-2023-042
• Eulerich, M., Sanatizadeh, A., Vakilzadeh, H., & Wood, D. A. (2024). Is it all hype? ChatGPT’s performance and disruptive potential in the accounting and auditing industries. Review of Accounting Studies, 29, 2318-2349. https://doi.org/10.1007/s11142-024-09833-9
  » https://doi.org/10.1007/s11142-024-09833-9
• European Commission. (2024). Horizon Europe work programme 2023-2025: 1. General introduction (European Commission Decision C(2024) 2371 of 17 April 2024). European Union. https://ec.europa.eu/info/funding-tenders/opportunities/docs/2021-2027/horizon/wp-call/2023-2024/wp-1-general-introduction_horizon-2023-2024_en.pdf
  » https://ec.europa.eu/info/funding-tenders/opportunities/docs/2021-2027/horizon/wp-call/2023-2024/wp-1-general-introduction_horizon-2023-2024_en.pdf
• Evans, T. M., Bira, L., Gastelum, J. B., Weiss, L. T., & Vanderford, N. L. (2018). Evidence for a mental health crisis in graduate education. Nature Biotechnology, 36, 282-284. https://doi.org/10.1038/nbt.4089
  » https://doi.org/10.1038/nbt.4089
• Faulconbridge, J., Sarwar, A., & Spring, M. (2025). How professionals adapt to artificial intelligence: the role of intertwined boundary work. Journal of Management Studies, 62(5), 1991-2024. https://doi.org/10.1111/joms.12936
  » https://doi.org/10.1111/joms.12936
• Feyerabend, P. (2010). Against method (4th ed.). Verso.
• Feynman, R. P. (1974). Cargo cult science. Engineering and Science, 37(7), 10-13. https://resolver.caltech.edu/CaltechES:37.7.0
  » https://resolver.caltech.edu/CaltechES:37.7.0
• Firoozi, M., Mokhtari, A., & Mokhtari, G. (2016). The relationship between organizational cynicism and innovative behaviors of sport and youth offices in Hamedan Province. International Journal of Science Culture and Sport, 4(3), 295-303. https://izlik.org/JA47RC99SX
  » https://izlik.org/JA47RC99SX
• Fraser, K., & Sheehy, B. (2020). Abundant publications but minuscule impact: the irrelevance of academic accounting research on practice and the profession. Publications, 8(4). https://doi.org/10.3390/publications8040046
  » https://doi.org/10.3390/publications8040046
• Freud, S. (1996). O mal-estar na civilização Imago.
• Gebru, T., Morgenstern, J., Vecchione, B., Vaughan, J. W., Wallach, H., Daumé, H., III., & Crawford, K. (2021). Datasheets for datasets. Communications of the ACM, 64(12), 86-92. https://doi.org/10.1145/3458723
  » https://doi.org/10.1145/3458723
• Gieryn, T. F. (1983). Boundary-work and the demarcation of science from non-science: strains and interests in professional ideologies of scientists. American Sociological Review, 48(6), 781-795. https://doi.org/10.2307/2095325
  » https://doi.org/10.2307/2095325
• Golyagina, A., & Valuckas, D. (2020). Boundary-work in management accounting: the case of hybrid professionalism. The British Accounting Review, 52(2), 100841. https://doi.org/10.1016/j.bar.2019.100841
  » https://doi.org/10.1016/j.bar.2019.100841
• Hasan, A. R. (2022). Artificial intelligence (AI) in accounting & auditing: a literature review. Open Journal of Business and Management, 10(1), 440-465. https://doi.org/10.4236/ojbm.2022.101026
  » https://doi.org/10.4236/ojbm.2022.101026
• Hopwood, A. G. (1989). Accounting and the pursuit of social interests. In W. F. Chua, T. Lowe, & T. Puxty (Eds.), Critical perspectives in management control (pp. 141-157). Palgrave Macmillan. https://doi.org/10.1007/978-1-349-07658-1_8
  » https://doi.org/10.1007/978-1-349-07658-1_8
• Hopwood, A. G., & Miller, P. (Eds.). (1994). Accounting as social and institutional practice Cambridge University Press.
• Hussain, S., Liu, L. Y. J., & Miller, A. D. (2020). Accounting as a dichotomised discipline: an analysis of the source materials used in the construction of accounting articles. Critical Perspectives on Accounting, 66, 102086. https://doi.org/10.1016/j.cpa.2019.04.007
  » https://doi.org/10.1016/j.cpa.2019.04.007
• Issa, H., Sun, T., & Vasarhelyi, M. A. (2016). Research ideas for artificial intelligence in auditing: The formalization of audit and workforce supplementation. Journal of Emerging Technologies in Accounting, 13(2), 1-20. https://doi.org/10.2308/jeta-10511
  » https://doi.org/10.2308/jeta-10511
• Jarrahi, M. H. (2018). Artificial intelligence and the future of work: human-AI symbiosis. Business Horizons, 61(4), 577-586. https://doi.org/10.1016/j.bushor.2018.03.007
  » https://doi.org/10.1016/j.bushor.2018.03.007
• Kaplan, R. S. (2011). Accounting scholarship that advances professional knowledge and practice. The Accounting Review, 86(2), 367-383. https://doi.org/10.2308/accr.00000031
  » https://doi.org/10.2308/accr.00000031
• Kuhn, T. S. (2012). The structure of scientific revolutions (4th ed.). University of Chicago Press.
• Lakatos, I. (1999). The methodology of scientific research programmes Cambridge University Press.
• Lee, T. A., & Williams, P. F. (1999). Accounting from the inside: legitimizing the accounting academic elite. Critical Perspectives on Accounting, 10(6), 867-895. https://doi.org/10.1006/cpac.1998.0281
  » https://doi.org/10.1006/cpac.1998.0281
• Lehner, O. M., Ittonen, K., Silvola, H., Ström, E., & Wührleitner, A. (2022). Artificial intelligence based decision-making in accounting and auditing: ethical challenges and normative thinking. Accounting, Auditing & Accountability Journal, 35(9), 109-135. https://doi.org/10.1108/AAAJ-09-2020-4934
  » https://doi.org/10.1108/AAAJ-09-2020-4934
• Liaras, E., Nerantzidis, M., & Alexandridis, A. (2024). Machine learning in accounting and finance research: a literature review. Review of Quantitative Finance and Accounting, 63, 1431-1471. https://doi.org/10.1007/s11156-024-01306-z
  » https://doi.org/10.1007/s11156-024-01306-z
• Luthfiani, A. D. (2024). The artificial intelligence revolution in accounting and auditing: opportunities, challenges, and future research directions. Journal of Applied Business, Taxation and Economics Research, 3(5), 516-530. https://doi.org/10.54408/jabter.v3i5.290
  » https://doi.org/10.54408/jabter.v3i5.290
• Malsch, B., Gendron, Y., & Grazzini, F. (2011). Investigating interdisciplinary translations: the influence of Pierre Bourdieu on accounting literature. Accounting, Auditing & Accountability Journal, 24(2), 194-228. https://doi.org/10.1108/09513571111100681
  » https://doi.org/10.1108/09513571111100681
• Modell, S. (2022). Is institutional research on management accounting degenerating or progressing? A Lakatosian analysis. Contemporary Accounting Research, 39(4), 2560-2595. https://doi.org/10.1111/1911-3846.12792
  » https://doi.org/10.1111/1911-3846.12792
• Popper, K. R. (2002). The logic of scientific discovery Routledge.
• Saltelli, A., Bammer, G., Bruno, I., Charters, E., Di Fiore, M., Didier, E., Espeland, W. N., Kay, J., Lo Piano, S., Mayo, D., Pielke, R., Jr., Portaluri, T., Porter, T. M., Puy, A., Rafols, I., Ravetz, J. R., Reinert, E., Sarewitz, D., Stark, P. B., ... Vineis, P. (2020). Five ways to ensure that models serve society: a manifesto. Nature, 582, 482-484. https://doi.org/10.1038/d41586-020-01812-9
  » https://doi.org/10.1038/d41586-020-01812-9
• Sloterdijk, P. (1983). Crítica da razão Estação Liberdade.
• Vogt, M., Silva, M. Z., & Valle, I. R. (2021). “Comendo pelas beiradas”: vigilância epistemológica e a abordagem bourdieusiana no campo contábil. Cadernos EBAPE.BR, 19(1), 58-69. https://doi.org/10.1590/1679-395120190117x
  » https://doi.org/10.1590/1679-395120190117x
• Volkova, O. N. (2024). Disciplinary boundaries of accounting: structure and history of research programs and institutions SSRN. https://doi.org/10.2139/ssrn.5057182
  » https://doi.org/10.2139/ssrn.5057182
• Willmott, H. (2011). Journal list fetishism and the perversion of scholarship: reactivity and the ABS list. Organization, 18(4), 429-442. https://doi.org/10.1177/1350508411403532
  » https://doi.org/10.1177/1350508411403532
• Woolston, C. (2020). Postdoc survey reveals disenchantment with working life. Nature, 587, 505-508. https://doi.org/10.1038/d41586-020-03191-7
  » https://doi.org/10.1038/d41586-020-03191-7
• World Economic Forum. (2020). The future of jobs report 2020 https://www3.weforum.org/docs/WEF_Future_of_Jobs_2020.pdf
  » https://www3.weforum.org/docs/WEF_Future_of_Jobs_2020.pdf

ACKNOWLEDGMENTS

The author thanks the editorial team and the reviewers for their contributions to improving the manuscript.

• RESEARCH DATA AVAILABILITY
  This theoretical essay did not generate a research dataset.
• ARTIFICIAL INTELLIGENCE USAGE
  The generative artificial intelligence tool ChatGPT (OpenAI) was used to support the organization, drafting, and language editing of the manuscript. The author critically reviewed and validated the final version and assumes full responsibility for the content.
• REVIEWERS
  The reviewers did not authorize the disclosure of their identities.
• PEER REVIEW REPORT
  The peer review report is available at https://periodicos.fgv.br/rap/article/view/97420
• 14
  [Translated version with AI assistance]

Edited by

• EDITOR-IN-CHIEF
  Hélio Arthur Reis Irigaray, Fundação Getulio Vargas, Rio de Janeiro, RJ, Brazil. ORCID: https://orcid.org/0000-0001-9580-7859

• ASSOCIATE EDITOR
  Fabricio Stocker, Fundação Getulio Vargas, Rio de Janeiro, RJ, Brazil. ORCID: https://orcid.org/0000-0001-6340-9127

Data availability

This theoretical essay did not generate a research dataset.

Publication Dates

• Publication in this collection
  21 Sept 2026
• Date of issue
  2026

History

• Received
  16 June 2025
• Accepted
  19 June 2026