Examining the Barriers and Enablers of AI Adoption in a Multimedia Organization
Citation. Mkhize, T., Oosterwyk, G., & Kautondokwa, P. (2023). Examining the barriers and enablers of AI adoption in a multimedia organization. EPiC Series in Computing, 93, 117–128. https://doi.org/10.29007/mgdl
Summary. This qualitative single-case study examines AI-enabled decision making in a South African multimedia organization through ten semi-structured interviews. It identifies interlocking technical, organizational, and social conditions: efficiency, staff involvement, business support, governance, accuracy, contractual obligations, red tape, accountability, and system capability.
Key claims
- Participants treated efficiency as an adoption enabler when AI reduced turnaround time and increased throughput, but manual work could remain faster for short or poorly fitting tasks.
- Early staff involvement supported cooperation, local feedback, and ownership, while also making AI limitations visible rather than promising replacement-free automation.
- Accuracy, usable training data, integration with business systems, traceability, and formal feedback loops constrained whether AI outputs could be trusted in operational decisions.
- Contractual obligations, internal approval processes, and unclear operational ownership limited autonomous use and required human review before customer-facing content was released.
- Business sponsorship and investment supported pilots and production use, but the authors describe governance as still developing rather than as an established solution.
Evidence & limitations
- The study uses thematic analysis of interviews with ten participants working on AI implementations and projects in one South African multimedia organization; it supplies situated mechanism evidence, not prevalence estimates or causal effects.
- Its 2023 conference-proceedings status and sector-specific case limit generalization, especially to contemporary generative or agentic systems.
- The public EasyChair PDF exactly matches the ingested Drive binary by SHA-256; the DOI resolves to the published proceedings record.