Challenges in trust performance evaluation

"Why does our top performer get the worst reviews?" the VP asked me. I was reviewing their annual performance data. "Show me," I said. She pulled up the ratings. Diana: 2.8 out of 5. Below average on "collaboration." Low marks for "team player." "What's her actual performance?" I asked. "Exceeded every target. Landed our biggest client. Trained three new hires." "So why the low scores?" "Her peer reviews are dragging her down." I scanned the comments. "Too direct." "Challenges ideas too much." "Not supportive enough." "Let me talk to Diana," I said. "I used to give honest feedback," Diana told me. "Said our pricing model was broken. Got dinged for 'negativity.'" "What happened with the pricing?" "They finally fixed it six months later. After we lost two major accounts." "What else?" "I questioned why we needed  eleven approvals for a simple contract change. Manager said I wasn't being collaborative." "Are you still giving feedback?" "No. I learned my lesson. Now I smile. Nod. Say everything's great. My reviews are improving." "But nothing's actually improving?" "We're making the same mistakes. Just with better vibes." She chuckled. I went back to the VP. "Your review system doesn't measure performance," I said. "It measures compliance." "That's not true." "When was the last time someone got promoted for challenging bad ideas?" Silence. "When did someone get rewarded for preventing a mistake?" More silence. "You've trained your best people to stay quiet. And your mediocre people to stay nice." A few months later, they redesigned the system. Added a category: "Constructive Challenge." Points for identifying problems early. Rewards for preventing costly mistakes. Diana got promoted. "What changed?" I asked the VP. "We stopped confusing agreement with alignment. Stopped mistaking silence for harmony." "And?" "Turns out our 'difficult' people were our most valuable. They actually cared enough to speak up." Here's the truth about performance reviews: Most companies don't reward performance. They reward performance theater. The person who says the meeting was great beats the person who says it wasted an hour. The person who agrees with bad ideas beats the person who prevents disasters. You think you're measuring contribution. You're measuring conformity. And your best people? They've already figured out the game. They're just deciding whether to play it or find somewhere that values truth over comfort. _____ Like my content? Give me a follow. Want to see more of it? Click the 🔔 on my profile.

46% → 29% - this is how fast trust in direct managers collapsed in just 2 years. And trust in senior leadership stuck at 32%. These numbers come from leadership research by Development Dimensions International and the latest Edelman Trust Barometer. Organisations are currently pouring attention, money and strategy into AI. But the most urgent leadership risk today is not technological. It is relational. WHY IT’S HAPPENING: 1️⃣ Leadership used to mean perspective. Now it means pressure. Many leaders no longer sit above the system, they are crushed by it. Instead of offering clarity, they transmit urgency. Over time, leadership stops feeling grounding. And trust dies when leadership feels as anxious as the team. 2️⃣ Employees don’t need more transparency. They need interpretability. Information is abundant. But meaning is not. Leaders share updates, dashboards and announcements but not interpretations of trade-offs, tensions or doubt. When people understand what is happening but not why, distrust takes root. 3️⃣ Performance is rewarded. Truth is not. Organisations claim to value speaking up. But they promote those who protect momentum, not those who disturb comfort. Over time, courage looks naïve. Silence looks professional. And trust cannot survive in systems that admire performance more than integrity. So in my view this is not an engagement problem. It is a psychological safety problem. HOW I HELP LEADERS REBUILD TRUST ▪️I make trust measurable. I help organisations measure psychological safety, so leaders can see where truth still flows and where it doesn’t. ▪️I make leadership teams safe first. I work with boards and executive teams to build psychological safety at the top, because trust never scales higher than senior leadership safety. ▪️I turn trust into daily leadership practice. I embed psychological safety into how decisions are made, how conflict is handled, and how feedback flows so trust becomes operational, not aspirational. From my practical experience, trust can recover faster than it collapsed - when leaders stop performing certainty and start creating safety paired with high performance standards. P.S. From your perspective, what has changed most in organizations that explains why trust in leaders is collapsing so fast?

Most performance problems are really trust problems. When performance slips, the instinct is often to tighten control by increasing oversight and metrics. On the surface, this looks responsible, but underneath, it often does the opposite of what’s intended. Being watched changes how people work: attention narrows, risk tolerance drops, and energy goes into managing impressions rather than solving problems. Survival mode takes over, and survival mode is efficient at avoiding mistakes, not at doing meaningful work. Trust creates a different internal state. When people feel trusted, they take ownership instead of seeking permission, and they think more creatively because the cost of being wrong no longer feels threatening. Research on psychological safety consistently shows this change: Trust expands cognitive capacity. Fear constricts it. If performance is lagging, it’s worth asking whether the issue is effort or whether the environment signals it’s safer to comply than to contribute. Where in your leadership might building trust address a performance issue that control alone will not? Are you thriving? 👉 https://lnkd.in/gmJVJ9Xj

“People don’t give their best when they’re watched. They give their best when they’re trusted.” This statement is simple, but in manufacturing environments, its impact is profound. From my experience, constant observation is not the biggest issue. Lack of trust is. Because when you feel that your management does not trust your decisions, it creates something much deeper than pressure. It creates doubt. You start questioning your judgement. You start second-guessing every action. You hesitate before making decisions you were once confident in. And over time, every decision begins to feel like it requires approval, validation, or acceptance not because it truly does, but because the environment has conditioned you to think that way. This is where performance begins to decline. Not due to lack of capability, but due to lack of confidence. In manufacturing, where timing, problem solving, and ownership are critical, this has real consequences: Slower decision making Reduced initiative Fear of accountability Loss of engagement And ultimately, reduced performance But beyond operational impact, there is also a human impact. When trust is consistently absent, it begins to affect: Mental wellbeing Professional identity Confidence in one’s own experience and knowledge And this is something organisations often underestimate. Because you cannot expect high performance from people who are not trusted to perform. Trust does not mean lack of control. It means clarity of standards, support when needed, and confidence in people to act within those standards. Strong leadership is not about monitoring every move. It is about creating an environment where: People understand expectations People are trained and supported People are trusted to take responsibility And people feel safe to make decisions and learn from them In my view, the most effective teams are not those constantly controlled. They are those trusted, empowered, and accountable. Because when trust is present, performance follows. And when it is missing, no level of supervision will compensate for it. Trust is not a soft value in manufacturing. It is a performance driver.

This study evaluates how interpretability and outcome feedback impact user trust and collaborative performance with AI. 1️⃣ Interpretability is defined as the ability to explain how a model arrives at its results in terms understandable to humans. 2️⃣ Outcome feedback is defined as providing the actual outcome after a prediction to confirm its accuracy. 3️⃣ Through experiments involving prediction tasks, the study finds that interpretability does not significantly enhance trust, while outcome feedback reliably increases it. 4️⃣ Neither interpretability nor outcome feedback greatly improves task performance, highlighting a trust-performance paradox. 5️⃣ Increased trust in AI does not always result in improved performance, revealing a trust-performance paradox. 6️⃣ Users tend to both overtrust and undertrust AI when given outcome feedback, which can undermine performance. 7️⃣ Time-dependent trends show that negative experiences with AI diminish future trust and performance. ✍🏻 Daehwan Ahn, Abdullah Almaatouq, Monisha Gulabani, Kartik Hosanagar. Impact of Model Interpretability and Outcome Feedback on Trust in AI. Conference on Human Factors in Computing Systems. 2024. DOI: 10.1145/3613904.3642780

One of the clearest lessons from Q4 is that customer trust in AI is not easy to gain. It’s also impossible to shortcut. Last week at Davos, our CEO Jason Droege spoke about what we’re seeing across the industry and Scale AI’s approach to research, enterprise embedding, and deployment. He also expanded on those themes in a blog post outlining our approach, why it drove momentum in 2025, and how it's shaping our 2026 strategy. I’ve also been reflecting on what I’m hearing in the industry and from our customers related to proving and improving AI reliability. Some are held up by imperfect data or worried about edge cases. Others face issues with policy constraints and fragmented accountability. Many of the conversations often sound something like this: “The demo looked great, but production felt unpredictable.” “We are not sure the agent followed policy.” “When we can’t measure behavior, everything gets blocked.” To me, the underlying theme here is both practical and philosophical. There’s a lack of trust. Trust breaks down not because AI systems fail outright, but because teams lack visibility into how they behave. When outputs can’t be measured, explained, or audited, even good results feel fragile. Here are a few ways to preemptively address deployment blockers and ways to proactively build trust internally and externally. 📏 Make reliability measurable Trust begins when behavior is observable. Enterprise tasks are complex and nuanced and rarely reflected in generic benchmarks. Reliability should be proven against real workflows, policies, and decision criteria before deployment instead of being discovered only after failures. 🛡️ Make evaluation defensible  LLM-as-a-judge only works when it is calibrated, auditable, and aligned to domain standards. Scores that cannot be explained or defended do not create confidence — they undermine it. 🏋♀️ Make results actionable When failures can’t be diagnosed, teams can’t improve reliability. Actionable evaluations help teams know why systems fail and improves reliability over time. 🤝 Make partnerships accountable Enterprise AI systems rarely live within a single team. Behavior is shaped across internal stakeholders, platforms, and external partners. When ownership is unclear, fragmented responsibility erodes trust over time. Clear accountability with shared standards, visibility, and escalation paths is essential to building reliable systems that teams can stand behind.

Chain-of-Trust: A Progressive Trust Evaluation Framework Enabled by Generative AI 👉 Why Traditional Trust Evaluation Falls Short Modern collaborative systems—from smart factories to distributed AI—rely on diverse devices working together. But how do we ensure these collaborators are trustworthy when: - Device capabilities update asynchronously - Network delays create incomplete data snapshots - Task requirements vary dramatically Traditional "all-at-once" trust assessments struggle with these dynamics, often leading to over-resourcing or security gaps. 👉 What Makes Chain-of-Trust Different Researchers from Western University, University of Glasgow, and University of Waterloo propose a staged evaluation framework: 1. Task decomposition: Break complex tasks into sequential requirements (e.g., "3D mapping" needs service availability → secure transmission → sufficient compute → reliable delivery) 2. Progressive filtering: Evaluate collaborators stage-by-stage using only relevant attributes 3. Generative AI integration: Use LLMs' contextual reasoning to:   - Interpret evolving task requirements   - Analyze partial attribute data   - Adapt evaluations dynamically through few-shot learning 👉 How It Works in Practice For a 3D mapping task: 1. Stage 1: Filter devices offering 3D mapping services 2. Stage 2: Verify communication security/bandwidth 3. Stage 3: Assess computing power/isolation 4. Stage 4: Confirm result delivery reliability At each stage, GPT-4 analyzes only the needed attributes, progressively narrowing trusted candidates. Key Results: - 92% accuracy vs. 64% in single-stage evaluations (GPT-4) - 40% resource reduction vs. full-attribute collection - No model retraining required for new tasks Implications: This approach addresses three critical gaps: 1. Handling asynchronous device updates 2. Preventing resource waste on irrelevant attributes 3. Maintaining context-aware evaluations Paper Authors: Botao Zhu, Xianbin Wang (Western University) Lei Zhang (University of Glasgow) Xuemin (Sherman) Shen (University of Waterloo) For those working on distributed systems or AI collaboration frameworks, this paper offers a practical blueprint for trustworthy resource allocation in dynamic environments.

In an industry focused on measuring everything from length of stay to readmission rates, we've overlooked our most fundamental metric: trust. This invisible foundation determines whether our sophisticated systems and advanced technologies actually improve health outcomes. When patients trust their providers, they share critical information, adhere to treatment plans, and return for necessary care. When providers trust their systems, they experience less burnout and make better clinical decisions. Trust isn't just a nice-to-have—it's the prerequisite that makes all other healthcare outcomes possible. The Trust Deficit Yet healthcare faces a profound trust crisis. Patients question whether financial interests outweigh clinical judgment. Providers wonder if systems support their work or just monitor productivity. Both navigate fragmented journeys where crucial information disappears between handoffs. We've designed systems that actively undermine trust: confusing billing, fragmented communication, and environments prioritizing efficiency over connection. Each frustrating interaction erodes the trust essential to healing. Trust as a Design Principle What if we designed for trust as intentionally as we design for efficiency? This means: +Creating transparency where there's typically obscurity: Making costs clear before services are rendered, explaining the why behind clinical decisions, and acknowledging uncertainty when it exists +Building consistency where there's typically variation: Ensuring care feels cohesive across touchpoints and providers share a complete picture of the patient's journey +Enabling human connection where there's typically transactional exchange: Designing environments and workflows that support meaningful conversation and relationship building +Demonstrating competence through thoughtful details: From clear wayfinding to seamless transitions between departments, showing that every aspect of the experience has been considered Measuring What Matters If trust is essential, we must measure it with the same rigor we apply to clinical metrics. This goes beyond satisfaction surveys to capturing specific moments where trust is built or broken: +Did you feel your concerns were taken seriously? +Was information shared in a way you could understand and act upon? +Were financial aspects of your care explained clearly and accurately? +Did your care team demonstrate they were communicating with each other? +Would you feel comfortable bringing up a sensitive health concern with your provider? Trust as Competitive Advantage The organizations that will thrive in healthcare's future aren't just those with the best technology or the most efficient processes—they're those that systematically build and protect trust at every touchpoint. In a world where patients have increasingly diverse options for care, trust becomes the differentiator that builds loyalty and word-of-mouth referrals.

In November 2025, during an Executive Committee meeting with a fintech client, a pivotal question was raised by a board member: “If a large enterprise evaluates us tomorrow for a buyout, can we prove we are trustworthy — not just secure?” This question highlighted a significant shift that many organizations are unprepared for. Security has evolved through three key phases: - Initially focused on defense - Then centered on compliance - Now, it is about digital trust as a growth function In regulated markets, trust influences: - Procurement speed - Sales cycle duration - Partnership approvals - Regulatory scrutiny management By 2026, the focus of boards will shift from asking CISOs, “Are we secure?” to “Can we prove trust — continuously, to anyone who matters?” This shift alters the landscape entirely. For CISOs, the role will expand from control ownership to trust orchestration. For CEOs, trust will transform into a revenue lever rather than merely a legal checkbox. In board and executive discussions, I utilize a three-stage lens: - Posture: What you claim to have (policies, controls, frameworks) - Proof: What you can demonstrate on demand (evidence, mappings, assurance) - Performance: What you can sustain over time (monitoring, metrics, outcomes) Most organizations find themselves stuck at the Posture stage; they may appear compliant on paper, but struggle when proof is requested. They often falter when continuous proof is necessary. The meeting concluded with a critical realization: “Security tools don’t create trust. Evidence, visibility, and consistency do.” Digital trust is no longer just an IT issue; it has become a board-level growth conversation, and it is approaching faster than many leaders anticipate. Follow for weekly insights on digital trust, security, and leadership. Comment “TRUST” if you would like to receive the Digital Trust Maturity Lens.