Echoes of Prejudice: Diversity Management, Algorithmic Bias, and Inclusive Marketing in Italian Companies

This study examines whether the adoption of artificial intelligence (AI) in Italian companies is accompanied by organisational capacities to identify and mitigate bias in marketing and communication. An original 44-item questionnaire was administered between April 2025 and March 2026 using a non-probability convenience design with limited referral recruitment. Following source-based cleaning of mixed online and paper records, the analytic sample comprised 120 respondent-reported company profiles. Analyses combined descriptive statistics, Monte Carlo χ2 tests, Cramér’s V, multiple correspondence analysis (MCA), and a parsimonious exploratory logistic regression. AI use or intended use was reported by 54.2% of respondents, whereas written diversity and inclusion policies (17.5%), training (15.8%), AI-bias audits (19.0% of 84 AI-eligible profiles), and mitigation procedures (17.9% of eligible profiles) were uncommon. Company size was associated with written policy adoption (χ2(6) = 41.69, Monte Carlo p < 0.001, V = 0.417), and written policy was strongly associated with training (χ2(4) = 81.85, p < 0.001, V = 0.584). The first two MCA dimensions explained 43.9% of total inertia (98.5% of Benzécri-adjusted inertia) and distinguished a formalisation gradient, from weak to strong governance, on the first dimension, and a distinct organisational-uncertainty dimension on the second. In the exploratory regression, training was positively associated with reported AI auditing (odds ratio = 5.81, 95% confidence interval [1.14, 29.65]), although estimates were imprecise and non-causal. The findings identify a governance gap: technological adoption can precede the institutionalisation of inclusive communication controls and algorithmic accountability.