Évaluation empirique des pratiques AGI dans l’industrie manufacturière française

Thesis Abstract

The rapid integration of artificial general intelligence (AGI) capabilities into manufacturing environments presents a strategic shift with potential productivity gains, quality improvements, and labor displacement risks, yet empirical evidence on actual practices, effectiveness, and organizational impact remains fragmented. This study addresses the gap by providing a rigorous empirical evaluation of AGI-driven practices in the French manufacturing sector, focusing on how AGI is adopted, operationalized, and linked to performance outcomes in real-world production settings. The aim is to determine the extent to which AGI interventions influence operational performance, workforce competencies, and process resilience, while unpacking contextual drivers and barriers across firms of varying sizes and maturities. Specific objectives include (1) mapping the prevalence and typologies of AGI applications deployed in French manufacturing plants over the past three years; (2) assessing the relationship between AGI deployment and key performance indicators such as Overall Equipment Effectiveness (OEE), defect rate, throughput, and energy efficiency; (3) examining the moderating roles of organizational factors (leadership support, data governance, and change management practices) on the AGI–performance linkage; (4) exploring workforce implications in terms of skill requirements, training needs, and job design; and (5) identifying contextual factors—industry subsector, plant size, and supply chain characteristics—that condition the benefits and risks of AGI adoption. The research adopts a mixed-methods design anchored in the Resource-Based View and Dynamic Capabilities Theory to explain how firms leverage AGI resources for competitive advantage while adapting capabilities to evolving environments. A sequential explanatory approach begins with a cross-sectional survey of 120 manufacturing plants across automotive, electronics, and consumer goods sectors in France, followed by in-depth case studies in 6 representative plants to triangulate survey findings. The population comprises French manufacturing firms actively employing AGI-enabled systems for production planning, quality control, predictive maintenance, and autonomous robotics. A stratified random sampling technique ensures representation by plant size (SMEs and large enterprises) and subsector. Data collection instruments include a structured questionnaire validated for construct reliability (Cronbach’s alpha > 0.70) and semi-structured interview guides with plant managers, operations directors, data scientists, and shop-floor engineers. Secondary data from plant dashboards provides objective metrics such as OEE, defect rates, cycle times, and energy consumption, supplemented by archival records on AGI project timelines and investment levels. Quantitative analysis employs descriptive statistics to profile AGI deployments and correlational methods to assess associations with performance metrics. Regression analyses (multivariate and hierarchical) test the direct and interaction effects of AGI intensity on operational outcomes, controlling for firm size, capital expenditure, and sector. Time-lagged analyses explore delayed effects. For the qualitative phase, thematic analysis of interview transcripts identifies emergent patterns related to organizational readiness, data governance, change management, and skills development, with coding guided by the theoretical constructs of the Dynamic Capabilities framework. A convergent mixed-methods integration synthesizes quantitative results with qualitative insights to provide a holistic understanding of how, why, and under which conditions AGI practices yield measurable benefits or pose risks. Expected findings anticipate a positive association between robust AGI implementation—characterized by integrated data pipelines, real-time analytics, and decision automation—and improvements in OEE, defect reduction, and throughput, moderated by strong data governance and leadership commitment. The study also expects to reveal nuanced workforce implications, including a shift toward higher-demand analytical and problem-solving competencies, along with targeted reskilling needs and potential transitional job design challenges. Variations are anticipated across subsectors and plant sizes, with larger firms more likely to realize substantial efficiency gains given scale economies and mature data platforms, while SMEs may face higher adoption costs and data integration hurdles. The contribution to knowledge lies in providing the first comprehensive, empirically grounded assessment of AGI practices in a European manufacturing context, clarifying the conditions under which AGI yields tangible performance gains and how organizational capabilities mediate those outcomes. The study informs managers and policymakers about optimal governance structures, investment prioritization, and human-capital strategies necessary to harness AGI’s potential while mitigating adverse effects. Practical recommendations include developing enterprise-wide data stewardship protocols, fostering cross-functional AGI governance boards, and implementing targeted training programs that align with anticipated process reengineering and quality improvements.