Work

Tekoälyn omaksuminen, vaikutukset ja rajat organisaation sisäisessä tiedonhallinnassa

Year: 2026
Type: dissertation
Abstract: This thesis examines how retrieval-augmented generation (RAG) tools are adopted and used in organizational knowledge work. It focuses on when employees find these tools useful, what supports recurring use, and how their use changes existing information and advisory processes. The study uses a mixed-methods case design in a Finnish industrial company. It combines 20 employee interviews with production logs and usage data from related information and advisory systems. The findings suggest that effective RAG-tool adoption is task-bounded. The tool created the clearest value in well-defined information-retrieval tasks. Recurring use depended on work habits, workflow integration, trust, and the quality of the organizational knowledge available to the system. Trust also developed in layers. Users separately evaluated the source material, retrieval, generated answers, and the need for human confirmation. Human advisors remained important when tasks required interpretation, problem diagnosis, or assurance. The study found no clear measurable reduction in advisory activity during the observed period. This shows that AI use and perceived value do not automatically lead to measurable process improvements. Finally, repeated or closely reformulated prompts were associated with lower user feedback. Prompt repetition may therefore provide a simple signal for identifying difficult human–LLM interactions. Overall, the findings suggest that organizational RAG adoption depends on the fit between the task, the user, the knowledge base, and the surrounding work process.
Institution: UPM (Finland)
Language: English
Open Access status: green
OpenAlex ID: W7225846525
Cites: 46
Cited by:
Related to: 10