AI, Employee Productivity and Performance: An Integrated ISM–MICMAC and SEM Investigation of the Pathway from Organisational Investment to Individual Output
Abstract Organisations have committed substantial capital to artificial intelligence in workforce management without a settled account of why productivity returns arrive so unevenly across employees working in the same firm under the same systems. This study approaches that problem in two stages. Twelve enabling and constraining factors are first extracted from the AI-HRM literature and ordered through interpretive structural modelling (ISM), with the resulting structure classified by driving and dependence power using MICMAC analysis. The hierarchy that emerges places top management commitment and digital leadership at the base and employee productivity and performance at the eighth and final level, with employee trust in AI and AI-related job insecurity occupying a reciprocal middle position that separates organisational investment from employee outcome. The mid-hierarchy portion of that structure is then specified as a measurement model and estimated on 412 employee responses from Indian knowledge-intensive firms using confirmatory factor analysis and structural equation modelling. The measurement model fits acceptably and every construct clears conventional reliability and discriminant validity thresholds. Human-AI task complementarity is the strongest proximal predictor of productivity and performance, followed at some distance by trust in AI and by job insecurity, which carries a negative sign. AI governance and reskilling investment have indirect effects, albeit small but reliable, that are passed on through the process of trust. However, job insecurity undermines the complementarity-performance relationship, although the interaction does not contribute much (under one percent) to the explained variance and should be avoided. The two phases imply that investment in artificial intelligence to boost productivity primarily is reshaping jobs and the context in which they exist.
Authors
- Deepak Ranjan Sahoo (ORCID: https://orcid.org/0000-0003-4013-2422)
- Vaishali Deepak Sahoo (ORCID: https://orcid.org/0009-0001-8118-2509)
- Ashwini Deshpande Joshi
Institutions
- MIT Art, Design and Technology University (IN)
- Vishwakarma University (IN)
Publication Details
- Journal
- Research Square
- Published
- 2026-11-17
- DOI
- https://doi.org/10.21203/rs.3.rs-11371853/v1
- Primary Topic
- AI and HR Technologies
- Type
- preprint