Where pilots are open to AI, and where they draw the line
Involo is exploring how AI-assisted tools can support pilot wellbeing without eroding the trust pilots place in their own judgment.
(02) ApproachWPT 02
Situations pilots rarely rehearse out loud
Semi-structured interviews used scenario vignettes to surface reactions to AI support in situations pilots don’t usually get to rehearse out loud.
Explore each stage· click a stage, or ← → to walk· tap a stage
↺ What changedSelect a stage marked ↺ to see how findings fed back into earlier work.
(03) FindingsWPT 03
Three boundaries, one framework
Thematic analysis across the interviews converged on a small set of boundaries pilots consistently drew around AI assistance.
Finding 01Sourced
Trust follows where the AI lives.
Pilots' trust in AI support is influenced by the organizational context in which the system operates.
Evidence“[Participant quote showing trust shifting with who operates the tool]” [P# · role · region]
Implication RecommendationMake the system’s organizational home — who operates it and who can see what — explicit before a pilot discloses anything.
Conceptual sketch
Finding 02Sourced
Pilots wanted support, not authority.
Participants differentiated between AI providing supportive guidance and AI assuming authority over professional or personal decisions.
Evidence“[Participant quote drawing the line between guidance and decision-making]” [P# · role · region]
Implication RecommendationKeep the AI in a supportive role; decisions with professional or personal consequences stay with people the pilot chooses.
Conceptual sketch
Finding 03Sourced
Privacy meant professional traceability.
Privacy concerns extended beyond data collection to whether sensitive disclosures could be linked to a pilot's professional identity.
Evidence“[Participant quote about disclosures being linked back to them]” [P# · role · region]
Implication RecommendationSeparate what is disclosed from professional identity by default, and say plainly what can and cannot be linked.
Conceptual sketch
Where the AI lives
Interaction the pilot initiatesPerceived pathway (participant concern), not a documented data flow
(04) OutcomeWPT 04
From boundaries to product constraints
The Trust-Bounded AI Support framework now informs Involo’s product constraints going forward.
01Evidence
02Interpretation
03Design implication
04Product decision
Hover or select any node to trace everything upstream and downstream of it. Select again to release.
(Reflection)
What this study can, and can’t, tell us
Naming the edges of the evidence is part of the method. Summaries stay visible; detail opens on request.
Trade-offDepth over breadth Draft+
Thirteen in-depth interviews describe the range of ways pilots think about AI support, not how common each view is across the profession.
LimitationAnticipated, not observed Draft+
Scenario vignettes capture how pilots expect they would respond. They don’t show behaviour with a deployed system.
LimitationA sensitive topic, self-reported Draft+
The professional stakes the study examined may also shape what participants felt comfortable saying in an interview. [Add how the study design mitigated this]
Would do differently[A choice you would revisit]+
[What you would change and why]
Open question[A question the study raised but couldn’t answer]+
[Why it matters]
Future research[The next study]+
[e.g. testing the framework against a working prototype]