Conference Agenda

Generative AI is increasingly permeating academic work and enabling the rapid production of “plausible” texts, code, and solutions with limited observable effort. In this European policy context, the European-Union is actively addressing the implications of AI for higher-education across teaching, learning, and research, by promoting labour-market-relevant digital skills among university students and researchers and by issuing ERA-level guidelines for the responsible use of generative AI in research (European-Commission, 2020; Directorate General, 2024).

We conceptualize generative AI as an institutional disruption in Christensen’s sense (Christensen, 1997) that unsettles the equilibrium linking higher-education to labor markets by eroding the credibility of degrees and academic outputs as signals under conditions of information asymmetry (Spence, 1973).

We interpret this evolving disruption through a Bourdieusian lens (Bourdieu, 1986, 1996; Bourdieu & Passeron, 1990), emphasizing AI’s challenge to mechanisms of distinction and the reproduction of the social order, maintained through various forms of capital (cultural, social, economic). Credentials, as a form of state-recognized institutionalized cultural capital, symbolize competence and social status. Academic titles consolidate social hierarchies through symbolic power. Thus, educational credentials help produce the “state nobility”, elites whose authority rests partly on state-sanctioned institutionalized cultural capital (Bourdieu, 1996).

The aim of this paper is to develop an integrated framework for understanding how generative AI reshapes higher-education’s signaling and distinction mechanisms.

Research questions: How does the integration of generative AI in higher-education reshape the credibility of higher-education degrees as a signal of competence to the labor market? How do higher-education institutions respond to these changes while seeking to maintain institutional prestige?

From a Bourdieusian perspective, higher-education functions as a field that monopolizes “legitimate culture” and credentialing, converting institutional recognition into cultural and symbolic capital (Bourdieu, 1986, 1996). Academic success is formally narrated as talent and effort, yet universities systematically reward prior cultural capital and reproduce social hierarchies through forms of symbolic violence (Bourdieu & Passeron, 1990; Nairz-Wirth & Wurzer, 2015). When exclusivity is threatened-through mass expansion or alternative pathways, institutions tend to generate new distinctions and boundary mechanisms to preserve stratification (Bourdieu, 1996). Generative AI disrupts these mechanisms of distinction by allowing high-status performances (e.g., competent academic voice, plausible argumentation) to be simulated without fully internalizing academic modes of inquiry and writing, thereby creating new channels of positioning (e.g., AI tool proficiency, prompt engineering, verification literacy) that may not overlap with traditional cultural capital.

Signaling theory clarifies why this matters for the labor market: education serves as a signal of potential productivity only if the effective costs of acquiring credentials are sufficiently differentiated by ability, and if false signals are deterred by verification and penalties (Spence, 1973; Connelly et al., 2025). Generative AI compresses these differential costs by lowering the cost of producing key academic outputs (seminar papers, final assignments, presentations), increasing the risk of false signaling unless institutions develop credible verification or penalty mechanisms (Connelly et al., 2025). When grades and degrees rely on outputs that can be generated at low cost, their signaling power may erode, prompting employers to shift attention toward alternative indicators such as work experience, tests, interviews, or work tasks (Van Belle et al., 2020), and increasing the salience of new credential forms (Tholen, 2023; Kässi & Lehdonvirta, 2024).

Inspired by disruptive innovation theory, we further interpret AI as challenging the value model of higher-education and the organizational “DNA” that makes internal change difficult (Christensen, 1997; Christensen & Eyring, 2011). Rather than immediate or linear reform, adaptation is therefore expected to proceed gradually through drift, layering, conversion, and displacement (Streeck & Thelen, 2005), as institutions contend with pressures to re-establish credibility while maintaining hierarchical differentiation within the field (Bourdieu, 1996).

The emergence of generative artificial intelligence in 2022 has been described by some scholars as the fourth industrial revolution (Serrano & Sánchez-Vera, 2024). This technology has become firmly embedded in personal, professional, and educational contexts, and universities are no exception. Evidence of this can be found in the guidance issued by several international organisations (CRUE, 2024; EUA, 2023; UNESCO, 2023) on the appropriate use of artificial intelligence (AI) within higher education.

The implementation, ongoing development, and use of AI in universities pose significant challenges for the entire academic community. These challenges include ensuring pedagogical coherence and aligning institutional strategies and technological policies with the achievement of educational objectives (García-Peñalvo et al., 2024). Furthermore, as noted by CRUE (2024), the adoption of AI in higher education must address four key dimensions: ethics and accessibility; data security and privacy; training and digital literacy across the academic community; and the safeguarding of academic quality.

According to Bruner (2011), organisations must adapt their management systems to changes in their environment if they are to maintain their position and achieve their strategic goals. In this regard, higher education institutions need to examine the impact that AI is having on their day-to-day management practices and to propose changes that enable them to adapt to the use of this emerging technology by members of the academic community.

Given that university managers play a central role in driving such changes, this contribution seeks to address the following research question: What concerns do university managers have regarding the implementation of AI in university teaching?

Based on the results of the EdU-InA project (Ref. PID2023-149069OA-I00), this paper aims to analyse the concerns expressed by university managers in relation to the need to adapt teaching management models in higher education to the use of artificial intelligence.

In Italian Universities dropout remains a critical issue with relevant social, economic, and institutional implications. Despite recent improvements across all degree types, the phenomenon continues to be significant (ANVUR, 2023), making the understanding of university dropout rates a major concern for higher education institutions.

Given the multifaceted problems associated with dropout, a key issue lies in identifying effective strategies to understand, monitor and predict this phenomenon at an early stage.

Within this framework, particular attention is devoted to predictive models aimed at detecting and preventing academic failure. Previous research has demonstrated the effectiveness of supervised machine learning approaches for this purpose, particularly when applied to administrative and academic records (Berens et al., 2019; Yağcı, 2022). Moreover, machine learning algorithms have been shown to be powerful predictors of student outcomes, supporting their use as early warning indicators (Delogu et al., 2024), thanks to their ability to capture complex, non-linear relationships among academic, socio-demographic, and other contextual variables. However, predictive accuracy alone is not sufficient when models are intended to support early intervention policies. In this context, ensuring fairness in machine learning models applied to educational data has become relevant (Raftopoulos et al., 2025), especially in settings characterized by heterogeneous populations. Therefore, fairness in early warning systems should focus on ensuring that predictive models do not systematically disadvantage specific groups while preserving data by respecting observed differences across groups. In such settings, calibration metrics are often considered appropriate (Kleinberg et al., 2016).

Given the rapid evolution of these models, other contributions proposed innovative models by applying it to Italian university case (Cannistrà et al., 2022; Ragni et al., 2024; Zanellati et al., 2024). Several studies have also highlighted the role of prior educational pathways and early academic performance (Priulla et al., 2024).

The availability of a nationwide dataset on higher education, including school information and INVALSI large-scale assessments data, represents a valuable resource for better understanding the key factors associated with university path and dropout. After comparing several machine learning models, that have been shown to perform particularly well with this type of data and for classification problems, the aim of this study is to analyze which model could represent a robust baseline for future research on the risk of academic failure. In particular, the study focuses on examining how information available at enrolment and during the first academic year affects model performance, by combining traditional feature importance measures with SHAP (SHapley Additive exPlanations), and on the use of fairness metrics as a diagnostic tool to assess whether boosting models provide equally reliable and actionable risk estimates across groups with different rates of the outcome of interest.