Will workers teach the AI that could replace them?
Knowledge transfer, sharing and management appear on 259 recent papers across the management and AI journals, but only 48 of them carry an AI or digital keyword, and knowledge hiding on 3. Firms are building assistants trained on employees' documents and chats. Do employees share less when sharing feeds a machine?
What prompted this
Since October 2025, 259 papers carry a knowledge-transfer, knowledge-sharing, knowledge-management or knowledge-hiding keyword (112 in management journals, 105 in AI): "knowledge transfer" on 114, "knowledge sharing" on 56, "knowledge management" on 43, "knowledge hiding" on 3. Forty-eight of them also carry an AI, generative-AI, language-model or digital keyword, most of those about tools rather than about people's willingness to share.
Questions
- When a firm announces that an internal assistant will be trained on employees' work, does documentation and answering of colleagues' questions rise or fall?
- Do experts whose know-how is most codifiable hide more, and do incentives for sharing offset that?
- Does the assistant reduce the value of expert employees in pay and promotions, and do they see it coming?
Data that could help
Internal collaboration logs (documents, chat answers, wiki edits) from firms that deployed assistants, with deployment dates, and employee surveys.
Next step
Compare documentation and question-answering activity in internal tools before and after an AI assistant announcement at a firm that shares its logs.