Novice risk work

Novice risk work is the work beginners do when they try to make an emerging technology safe, acceptable, and usable for more senior or accountable actors before they themselves have deep technical or organizational understanding. 2025-kellogg-novice-risk-work shows the failure mode in GenAI adoption: junior consultants expected to educate senior professionals, but their proposed mitigation tactics often under-addressed the risks seniors cared about because they leaned on human routine changes and project-level coordination rather than system design, deployer controls, or ecosystem-level interventions.

The construct matters because it separates willingness to help with AI adoption from capacity to diagnose AI risk. A junior professional may have more exposure, enthusiasm, or tactical fluency than a senior manager while still lacking the AI literacy, system-design vocabulary, and governance perspective needed to teach safe use. That makes novice risk work a boundary condition for bottom-up AI diffusion and reverse mentoring: the organization may get adoption energy without calibrated stewardship.

Connections

• Connects to ai-literacy because the key issue is not whether juniors are receptive to AI, but whether they understand enough about capabilities, failure modes, and mitigation levels to guide others.
• Connects to ai-adoption because adoption can be pulled forward by enthusiastic novices while risk mitigation remains immature or mislocalized.
• Connects to work-redesign because shifting from human routine changes to system design is a redesign problem, not merely a training problem.
• Connects to human-ai-collaboration because collaboration quality depends on whether users know when individual checking, interface design, deployer controls, or ecosystem governance is the right level of intervention.

Contradictions & open questions

• 2025-kellogg-novice-risk-work complicates reverse-mentoring narratives: juniors may be less status-threatening teachers, but they are not automatically better teachers when the technology is emerging, fast-changing, and under-governed.
• The evidence is early GenAI and one professional-services setting. The open question is when novice risk work becomes a productive bridge to organizational learning versus a source of false reassurance.
• The wiki needs evidence on interventions: whether structured AI literacy, risk taxonomies, prompt libraries, audit routines, or deployer-level controls help novice users teach seniors without overclaiming what project-level behavior changes can solve.