Epistemic Trust as a Mechanism for Ethics Integration: Failure Modes and Design Principles from 70 Moral Imagination Workshops

Abstract: Bottom-up responsible innovation initiatives seek to empower technology development teams to engage in ethical reflection, yet such interventions frequently fail to achieve practitioner engagement. Why do some ethics interventions succeed while others are dismissed as irrelevant, adversarial, or disconnected from work? This paper proposes epistemic trust -- the degree to which practitioners regard an intervention, its facilitators, and its content as credible, relevant, and actionable -- as a conceptual model linking intervention design to engagement outcomes. Drawing on philosophical work on testimony and on practice-based qualitative analysis of over 70 moral imagination workshops with engineering teams between 2019 and 2025, we identify five dimensions of epistemic trust salient to ethics interventions (Relevance, Inclusivity, Agency, Authority, and Alignment) and present a typology of 23 failure modes that arise when these dimensions are inadequately addressed. We derive nine design principles for cultivating epistemic trust, grounded in our operationalisation of moral imagination through technomoral scenarios and structured deliberation. Our findings contribute to the literature on collaborative socio-technical integration by specifying conditions of uptake that existing frameworks leave undertheorised. We acknowledge limitations including selection effects from voluntary participation and the absence of formal outcome measures, and position our failure mode typology as practitioner hypotheses warranting further empirical validation.

• The paper identifies epistemic trust as a key mechanism linking the design of ethics interventions to practitioner engagement through analysis of 70 workshops.
• The study presents a detailed typology of 23 failure modes across five dimensions and offers nine actionable design principles to enhance ethical integration.
• The findings emphasize the importance of contextual tailoring and dual-expertise facilitation in making ethics interventions credible, relevant, and actionable within engineering teams.

Epistemic Trust as a Mechanism for Ethics Integration in Engineering Teams

Introduction: Framing the Challenge of Ethics Integration

The increasing institutionalization of responsible innovation processes within technology development teams is widely seen as necessary to fill regulatory gaps and address emergent social concerns arising from rapid technological advancement. However, bottom-up ethics interventions intended to empower technical practitioners frequently fail to generate sustained engagement. This paper ("Epistemic Trust as a Mechanism for Ethics Integration: Failure Modes and Design Principles from 70 Moral Imagination Workshops" (2604.11281)) presents a comprehensive practice-based inquiry into why such interventions succeed or fail, proposing epistemic trust as a decisive mechanism that links the design of interventions to their uptake and effectiveness.

The authors build their analysis on more than 70 "Moral Imagination" workshops conducted with engineering teams at Google. They identify five salient dimensions of epistemic trust—Relevance, Inclusivity, Agency, Authority, and Alignment—and present a typology of twenty-three failure modes, which are then mapped to nine design principles developed to enhance epistemic trust and, by extension, practitioner engagement. Their analysis integrates philosophical work on testimony, social epistemology, and empirical accounts from the facilitation of structured ethics reflection in technical environments.

Theoretical Foundations: Epistemic Trust as an Uptake-Driving Mechanism

The authors advance a mechanism model in which engagement with ethics interventions is contingent upon epistemic trust—the degree to which practitioners perceive the intervention, its facilitators, and its content as credible, relevant, and actionable. Drawing from philosophical treatments of testimony (e.g., Hardwig, Goldman, Zagzebski) and empirical findings on psychological safety in teams (e.g., Edmondson), the model holds that epistemic trust is multi-dimensional. The five proposed dimensions are:

• Relevance: Ethical reflection must be perceived as integral rather than extraneous to technical work.
• Inclusivity: Processes must legitimize diverse voices and valorize dissent to avoid superficial consensus.
• Agency: Practitioners must be enabled as moral agents, not passive recipients of ethical mandates.
• Authority: Facilitators must possess and demonstrate both technical and ethical competence to maintain credibility.
• Alignment: Interventions must be experienced as serving team/project objectives, not as adversarial or disconnected compliance rituals.

The paper situates these dimensions within prior collaborative integration approaches (e.g., STIR, Value Sensitive Design) and demonstrates that existing frameworks rarely theorize or operationalize the conditions of epistemic trust with comparable specificity.

Moral Imagination Methodology: Design and Dynamics

The "Moral Imagination" methodology builds upon pragmatist accounts of ethical reflection and is operationalized through structured, modular workshops. The approach is characterized by:

• The use of technomoral scenarios tailored through intensive intake processes to reflect current and potential future project realities.
• Structured role-play and perspective-taking, foregrounding participants' recognition of their own perspectival limitations and enabling generative envisioning of alternatives.
• Segmentation into "Reflect," "Articulate," and "Action" phases, culminating in the formulation of Responsibility Objectives integrated with extant project management artefacts.

Crucially, all participation was voluntary, and facilitation teams were assembled to balance ethical/philosophical and technical domain expertise.

Failure Modes: Typology and Mechanistic Analysis

Empirical analysis of workshops reveals 23 persistent failure modes, systematically organized by epistemic trust dimension:

• Relevance failures include perceptions of external imposition, abstraction disconnected from work, scenario irrelevance due to poor context adaptation, and conversational drift toward generalities or personal values.
• Inclusivity failures manifest as hierarchy reproduction (senior voices dominating), premature consensus, or resistance to role-play/scenario methods.
• Agency failures involve passive reception, responsibility deflection, inadequate calibration of dilemma complexity, and facilitation rigidity or drift.
• Authority failures occur when facilitators are perceived as technically naïve, when discussion is dominated by either technical or ethical jargon, or when challenge is experienced as adversarial.
• Alignment failures include adversarial framing of ethics, temporal discounting of ethical issues, unacknowledged divergence between facilitator/team goals, and isolation from practice (i.e., lack of integration with ongoing work).

The paper presents illustrative ethnographic vignettes for each and identifies cross-cutting modes (credibility-connection trap; participation paradox; co-optation risk; expertise hierarchy) that exacerbate or compound disengagement.

Design Principles: Operational Solutions for Cultivating Epistemic Trust

Based on the failure mode analysis, the authors propose nine design principles for maximizing epistemic trust:

1. Contextual Tailoring: Ensure scenarios and discussions are project-specific and technically anchored.
2. Non-Adversarial Framing: Present the workshop as collaborative inquiry, not critique or compliance exercise.
3. Dual-Expertise Facilitation: Staff workshops with facilitators fluent in both technical and ethical registers.
4. Structured Voice Distribution: Employ deliberation structures to equalize participation and invite dissent.
5. Deliberative Ownership: Invest participants with primary ethical agency.
6. Benevolent Challenge: Challenge assumptions constructively, but avoid adversarial stance.
7. Goal Transparency: Surface and align workshop and team goals explicitly.
8. Artefact Integration: Translate ethical insights into actionable commitments within project management systems.
9. Scenario Method Framing: Justify and scaffold role-play to minimize resistance.

The mappings between these principles and specific failure modes are detailed, offering a practitioner-oriented matrix for diagnostic and remedial purposes.

Discussion: Implications, Limitations, and Scope

The paper's primary contributions are (1) reframing "buy-in" for bottom-up ethics integration as a multi-dimensional epistemic trust problem, (2) providing an empirically grounded failure mode typology, and (3) offering actionable design principles for improving the likelihood of uptake and substantive ethical engagement in technical teams.

While the voluntary, self-selected nature of the workshop samples introduces selection effects and limits generalizability to contexts of mandated ethics integration, the authors explicitly position their typology and principles as hypotheses for further empirical validation rather than controlled causal claims. They further acknowledge structural limitations—namely, that epistemic trust at the team level is not sufficient to overcome organization- or domain-level drivers of unethical outcomes.

Of note, the paper directly addresses the risk of co-optation: alignment, while necessary for engagement, can degenerate into compliance rituals that shore up legitimacy without effecting meaningful social or technical change. The tension between trusted facilitation and critical challenge is treated as an irreducible design trade-off.

Conclusion

This research identifies and theorizes epistemic trust as a necessary condition for successful ethics integration in engineering contexts. It makes a robust contribution to the responsible innovation literature by specifying actionable dimensions of trust, typologizing common failure modes, and mapping these to modular design principles for intervention practice. The framework offers avenues for both process and outcome evaluation in future research, and it sets a new empirical and theoretical standard for the design and assessment of ethics interventions in technical domains.

The typology and principles are offered as both diagnostic tools and as a foundation for future comparative, experimental, and longitudinal inquiry into the efficacy of ethics interventions at scale. The analysis underscores that unless practitioners regard ethical reflection as credible, relevant, and actionable—unless epistemic trust is cultivated along all necessary dimensions—ethics interventions, regardless of resourcing or institutional endorsement, are unlikely to produce substantive change.