Human-centric design for computational trust

Designing UX for autonomous multi-agent systems is a whole new game. These agents take initiative, make decisions, and collaborate, the old click and respond model no longer works. Users need control without micromanagement, clarity without overload, and trust in what’s happening behind the scenes. That’s why trust, transparency, and human-first design aren’t optional — they’re foundational. 1. Capability Discovery One of the first barriers to adoption is uncertainty. Users often don't know what an agent can do, especially when multiple agents collaborate across domains. Interfaces must provide dynamic affordances, contextual tooltips, and scenario-based walkthroughs that answer: “What can this agent do for me right now?” This ensures users onboard with confidence, reducing trial-and-error learning and surfacing hidden agent potential early. 2. Observability and Provenance In systems where agents learn, evolve, and interact autonomously, users must be able to trace not just what happened, but why. Observability goes beyond logs; it includes time-stamped decision trails, causal chains, and visualization of agent communication. Provenance gives users the power to challenge decisions, audit behaviors, and even retrain agents, which is critical in high-stakes domains like finance, healthcare, or DevOps. 3. Interruptibility Autonomy must not translate to irreversibility. Users should be able to pause, resume, or cancel agent actions with clear consequences. This empowers human oversight in dynamic contexts (e.g., pausing RCA during live production incidents), and reduces fear around automation. Temporal control over agent execution makes the system feel safe, adaptable, and co-operative. 4. Cost-Aware Delegation Many agent actions incur downstream costs, infrastructure, computation, or time. Interfaces must make the invisible cost visible before action. For example, spawning an AI model or triggering auto-remediation should expose an estimated impact window. Letting users define policies (e.g., “Only auto-remediate when risk score < 30 and impact < $100”) enables fine-grained trust calibration. 5. Persona-Aligned Feedback Loops Each user persona, from QA engineer to SRE will interact with agents differently. The system must offer feedback loops tailored to that persona’s context. For example, a test generator agent may ask a QA to verify coverage gaps, while an anomaly agent may provide confidence ranges and time-series correlations for SREs. This ensures the system evolves in alignment with real user goals, not just data. In multi-agent systems, agency without alignment is chaos. These principles help build systems that are not only intelligent but intelligible, reliable, and human-centered.

🌳 Design Patterns For Building Trust. With practical guidelines for designers on how to make products — AI and non-AI — more trustworthy, reliable and honest. In the noisy and polluted world today, trust doesn’t come for free. It doesn’t emerge by default. It must be earned and meticulously preserved — by being reliable, accountable and treating customers with respect. This holds true for people but it also for software. According to Anyi Sun, there are 5 psychological foundations of user trust: 1. Reliability 🔰 The degree to which the product consistently behaves as expected. It's a sense that that the product is dependable — based on a track record of past actions. Reliability comes from promising what you do, and doing what you promised. 2. Technical competence ��� Perceived intelligence, sophistication and capability of the product. It's user's belief that the product can successfully perform what they are being trusted to do. It's about trusting product's capability. 3. Understandability 🧠 The extent to which users feel they can understand how the system works or why it made a certain decision. The product must be able to articulate how a decision came along, with references to fragments that underpin a decision. 4. Faith and Care 🌱 Emotional, almost "blind trust" in the product, especially when users don't understand the underlying logic. It's a belief that the trusted party actually cares about the positive outcome for you, and intends to do good. 5. Personal attachment 🌳 A sense of rapport, connection or emotional engagement with the product. Typically it emerges when a user feels that they get meaningful value from the product, and from interactions with people supporting it. Personally, I would also add the value of repeated positive experiences that build confidence in the quality of the product, and hence its reliability. --- With AI products, hitting all these psychological foundations is extremely hard. Surely some people trust AI almost instinctively, others are more critical. But people's attitude often changes dramatically once they realized that they've made severe mistakes because of AI. Recovering from it is very hard. We can help with some design patterns: 1. Avoid "Ask me anything" → push for scoping and constraints 2. Slow down users in prompting → request specific details 3. Present multiple viewpoints, explain that experts disagree 4. Allow users to manage “memory”, profiles personalization 5. Highlight what is AI-generated and what isn't (AI disclosure) 6. Allow users to override AI-generated suggestions manually 7. Allow users to tweak AI output and refine it for their needs 8. Adapt AI's tone depending on the severity of user's task Trust is why people stay or leave. It builds long-term loyalty and helps users overcome hesitation. But it must be designed and retained — across all psychological foundations and with thoughtful UX work. I think designers will be quite busy for years to come. #ux #design

Humanizing AI Through the Kano Model In an era where generative AI has become a ubiquitous offering, true differentiation lies not in merely adopting the technology but in integrating human values into its core. Building on my earlier discussion about applying the Kano Model to Gen AI strategy, let’s explore how this framework can refocus development metrics to prioritize ethics and human-centricity. By aligning AI systems with human needs, organizations can shift from functional tools to trusted partners that inspire lasting loyalty. Traditional metrics such as speed, scalability, and model accuracy have evolved into basic expectations the “must-haves” of AI. What truly elevates a product today is its ability to embody values like safety, helpfulness, dignity, and harmlessness. These qualities, categorized as “delighters” in the Kano Model, transform AI from a transactional tool into a meaningful collaborator. Key Human-Centric Differentiators Safety: Proactive safeguards must ensure AI systems protect users from risks, whether physical, emotional, or societal. Safety is non-negotiable in building trust. Helpfulness: Personalized, context-aware interactions demonstrate empathy. AI should anticipate needs and adapt to individual preferences, turning routine tasks into meaningful experiences. Dignity: Ethical design principles—fairness, transparency, and privacy—must underpin AI development. Respecting user autonomy fosters long-term trust and engagement. Harmlessness: AI outputs and recommendations should prioritize user well-being, avoiding unintended consequences like bias, misinformation, or psychological harm. This human-centered approach represents a paradigm shift in technology development. While traditional KPIs remain important, they are no longer sufficient to stand out in a crowded market. Organizations that embed human values into their AI systems will not only meet user expectations but exceed them, creating emotional connections that drive loyalty. By applying the Kano Model, businesses can systematically align innovation with ethics, ensuring technology serves humanity rather than the other way around. The future of AI isn’t just about efficiency it’s about elevating human potential through thoughtful, responsible design. How is your organization balancing technical excellence with human values?

💎 The future of healthcare won’t be shaped by AI performance. It’ll be shaped by human perception. AI that isn’t aligned with clinical workflows will always be ignored, no matter how powerful. A recent systematic review in JMIR took on a deceptively complex question: What makes health care workers trust (or not trust) AI-based clinical decision support systems (AI-CDSSs)? 📖 Across 27 studies from 2020–2024, a clear narrative emerged, not about performance metrics or model architectures, but about relationships. Between humans and machines. Between clinicians and their own judgment. Between what’s explainable, and what’s opaque. Eight interconnected factors shape trust: 1️⃣ Transparency – not just visibility into models, but meaningful interpretability 2️⃣ Training & Familiarity – confidence is earned through hands-on experience 3️⃣ Usability – tools must fit the messy, fast-paced realities of clinical workflows 4️⃣ Clinical Reliability – trust grows with proven, consistent performance 5️⃣ Credibility – of the developers, the data, and the science behind the system 6️⃣ Ethical Alignment – legal clarity, fairness, and accountability still matter deeply 7️⃣ Human-Centered Design – because clinicians don't want to be replaced—they want to be supported 8️⃣ Customization & Control – AI must adapt to clinicians, not the other way around The review doesn’t pretend there’s a universal blueprint for trust—but it offers a human lens on what’s too often treated as a purely technical problem. Let’s design AI not just to work, but to be welcomed. #HumanAIInteraction #TrustInAI #ClinicalAI #AIEthics #ExplainableAI #SociotechnicalSystems #HealthcareAI #ResponsibleAI #HumanCenteredAI #DigitalHealth

#AI didn’t just change how work gets done. It changed who we trust. For decades, trust was earned through people. Credentials. Experience. Reputation. Now trust is quietly shifting to systems. If the dashboard says it’s fine, we relax. If the model recommends it, we comply. If the workflow moves forward, we assume someone checked. Medicine has lived through this transition. Clinical judgment slowly gave way to protocol. Protocol gave way to automation. And automation changed behavior long before outcomes were measured. AI accelerates that shift. The most dangerous moment is not when AI is wrong. It’s when AI becomes the default authority. Because authority without accountability feels efficient. And efficiency is seductive. Leaders need to ask a hard question. Who does my organization trust more, humans or models? If the answer is unclear, the system is already deciding for you. Best practices for preserving human authority in AI systems: Make recommendations explainable, not just accurate Force deliberate pauses before irreversible actions Design interfaces that invite questioning, not compliance Train people to challenge AI, not defer to it Tie accountability to humans, not tools AI should inform judgment, not replace it. Trust should be earned, not automated. The future belongs to organizations that understand this early. #AI #ArtificialIntelligence #AIGovernance #ResponsibleAI #Leadership #TrustInAI #HumanCenteredAI #DigitalTransformation #RiskManagement #DrGPT

Tired of AI projects that don't deliver? Try this human-centred approach. From my research over the past couple of years, I’ve noticed a recurring pattern. We often treat AI as a technology experiment rather than an upgrade to how people actually work. That mindset can quietly limit a project’s success. To support better decisions, I’ve developed a human-centred AI readiness checklist based on that research. I hope it’s useful for your next initiative. 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝘆 𝗮𝗻𝗱 𝗢𝘂𝘁𝗰𝗼𝗺𝗲 𝗖𝗵𝗲𝗰𝗸 (𝗖𝗥𝗜𝗦𝗣-𝗗𝗠 𝗺𝗶𝗻𝗱𝘀𝗲𝘁) →Are we clear on the operational outcome and metric we are improving? ↳If we cannot say “this reduces X by Y%”, we are chasing tools, not performance. 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗠𝗮𝗽𝗽𝗶𝗻𝗴 𝗖𝗵𝗲𝗰𝗸 (𝗟𝗲𝗮𝗻 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴) →Which real human decisions are we supporting? ↳AI should strengthen judgment points like prioritisation or scheduling, not automate activity without purpose. 𝗣𝗿𝗼𝗰𝗲𝘀𝘀 𝗦𝘁𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗖𝗵𝗲𝗰𝗸 (𝗟𝗲𝗮𝗻 𝗽𝗿𝗶𝗻𝗰𝗶𝗽𝗹𝗲) → Is the workflow stable enough to augment? ↳Automating instability scales, defects and frustrates the people doing the work. 𝗩𝗮𝗹𝘂��� 𝘃𝘀 𝗗𝗶𝘀𝗿𝘂𝗽𝘁𝗶𝗼𝗻 𝗖𝗵𝗲𝗰𝗸 (𝗣𝗼𝗿𝘁𝗳𝗼𝗹𝗶𝗼 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴) →Does the benefit outweigh frontline disruption? ↳Operational AI should improve flow, not create friction for teams. 𝗗𝗮𝘁𝗮 𝗥𝗲𝗮𝗹𝗶𝘁𝘆 𝗖𝗵𝗲𝗰𝗸 (𝗖𝗥𝗜𝗦𝗣-𝗗𝗠 𝗱𝗮𝘁𝗮 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴) →Does our data reflect lived operational reality? ↳Human trust collapses when AI runs on distorted inputs. 𝗛𝘂𝗺𝗮𝗻 𝗖𝗼𝗻𝘁𝗿𝗼𝗹 𝗖𝗵𝗲𝗰𝗸 (𝗛𝘂𝗺𝗮𝗻-𝗰𝗲𝗻𝘁𝗲𝗿𝗲𝗱 𝗔𝗜 𝗱𝗲𝘀𝗶𝗴𝗻) →Where does AI advise, where do humans review, and where does automation act? ↳Clear boundaries protect autonomy and accountability. 𝗥𝗶𝘀𝗸 𝗮𝗻𝗱 𝗥𝗲𝘀𝗶𝗹𝗶𝗲𝗻𝗰𝗲 𝗖𝗵𝗲𝗰𝗸 (𝗡𝗜𝗦𝗧 𝗔𝗜 𝗿𝗶𝘀𝗸 𝗺𝗼𝗱𝗲𝗹) →Have we planned for failure, overrides, and fallback workflows? ↳Operations must remain safe and continuous when systems misfire. 𝗢𝘄𝗻𝗲𝗿𝘀𝗵𝗶𝗽 𝗖𝗵𝗲𝗰𝗸 (𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝗺𝗼𝗱𝗲𝗹 𝗰𝗹𝗮𝗿𝗶𝘁𝘆) →Who owns outcomes, model behaviour, and data quality? ↳Human accountability must remain visible after launch. 𝗜𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻 𝗥𝗲𝗮𝗹𝗶𝘁𝘆 𝗖𝗵𝗲𝗰𝗸 (𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝘁𝗵𝗶𝗻𝗸𝗶𝗻𝗴) →Will this support how people actually work? ↳Tools that slow teams are quietly abandoned. 𝗔𝗱𝗼𝗽𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗧𝗿𝘂𝘀𝘁 𝗖𝗵𝗲𝗰𝗸 (𝗖𝗵𝗮𝗻𝗴𝗲 𝗱𝗶𝘀𝗰𝗶𝗽𝗹𝗶𝗻𝗲) →Are we designing for understanding, transparency, and behavioural adoption? ↳Trust grows when teams see AI improving their work, not replacing it. AI is an amplifier. It scales what we already have: good or bad ↳𝐆𝐚𝐫𝐛𝐚𝐠𝐞 𝐢𝐧. 𝐀𝐦𝐩𝐥𝐢𝐟𝐢𝐞𝐝 𝐠𝐚𝐫𝐛𝐚𝐠𝐞 𝐨𝐮𝐭.⁣ ⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣⁣ ⁣⁣⁣⁣⁣⁣⁣⁣The strongest AI initiatives aren’t just technology deployments. They are human-centred operating upgrades that happen to use AI. ♻️ Share if you found this useful. #AIinBusiness #HumanCenteredAI #Operations #Leadership #AIStrategy

For human AI design, we often focus on whether a system is usable, efficient, or trusted, but those questions only get us so far. They describe the surface of the interaction, not the deeper cognitive processes shaping it. What matters more to me is this: what kind of thinking is the system encouraging? Is it helping people reflect more carefully, make better judgments, and build stronger skills over time? Or is it making reliance faster, easier, and less thoughtful? Is it adapting to the user’s actual level of understanding, or just giving everyone the same kind of output and calling that personalization? Is it helping people stay meaningfully involved in the decision, or pushing them into the role of passive reviewer? I have been thinking a lot about the kinds of models we need if we want to design human AI systems more rigorously. Some of the most useful ones focus on adaptation. If an AI system can estimate what a user likely knows, believes, or needs, it can calibrate its explanations and support instead of overexplaining to experts or underserving beginners. Others focus on proactive intelligence, showing how systems can reduce uncertainty by predicting, updating, and acting in context rather than simply waiting for commands. Trust is another area where I think we need more precision. Acceptance of AI advice is not always a sign of good trust. Sometimes it reflects careful judgment. Sometimes it reflects shallow, fast reliance. That distinction matters, especially when designers are evaluating whether a system is genuinely helping users think better or just making them comply more quickly. I am also especially interested in models that preserve human agency. In high stakes settings, good design is not always about removing friction. Sometimes it is about introducing the right kind of friction so users pause, verify, and reflect before accepting an AI output. A confident sounding suggestion can still be wrong, and when that happens, the interface should support human judgment. Memory matters too. Many systems still feel transient. They do not maintain a stable, structured understanding of the user, the task, or the broader context over time. If we want more capable human AI collaboration, we need systems that can support longer horizon interaction without constantly resetting.

What does it take to keep humans thinking when AI offers easy answers? Meet Zana Buçinca, whose research shows that the design of AI systems can make or break our ability to think critically. As AI transforms work, her research optimizes it not just for efficiency but also for human growth and meaning. Zana is currently a Postdoctoral Researcher at Microsoft and will join MIT as an Assistant Professor next fall. We all tend to over-rely on AI recommendations, even when they’re wrong. Over time, this autopilot reliance can erode our ability to judge AI outputs independently. Zana's research emphasizes strategies to disrupt this autopilot thinking, keeping human judgment front and center. She found that adding cognitive forcing functions – nudges that make people pause and think – significantly reduced overreliance compared to standard explainable AI approaches, although the most effective designs were also rated the least usable, highlighting a tension between performance and preference (https://lnkd.in/gXrM9R-3). She also showed that popular shortcuts for evaluating human-AI collaboration (like proxy tasks and trust ratings) can be misleading because they don’t predict real-world success, prompting a shift toward more rigorous, outcome-based evaluation (https://lnkd.in/gCN64mJa). As AI becomes embedded in everyday workflows, the question we need to ask is not just, “Is the AI accurate?” We must also ask, “How does it change us?” Zana’s recent research uses offline reinforcement learning to personalize AI support for human-centric goals like skill development and human-AI complementarity, rather than maximizing correctness (https://lnkd.in/gxkrpuFe). In one study, she found that contrastive explanations that anticipate human misconceptions significantly improved people’s decision-making skills without sacrificing accuracy (https://lnkd.in/gRND_n7P). These contributions chart a path toward adaptive AI systems that preserve human judgment and strengthen our expertise. If you’re not yet following Zana’s research, I highly recommend checking it out! #AIInnovators #AppliedResearch #OAR #LeadingLikeAScientist

Over the past year, I’ve been exploring a question that keeps coming up: What makes AI truly human-centred and how do we build better AI? A milestone in this journey is the publication of my paper, “Integrating Psychological Theories into AI Design”, now out in Psychology (SCIRP). The paper builds on my MSc dissertation and argues that to build better AI people can trust and adopt, we need to look beyond algorithms. We must draw on psychology: how people think, feel and make decisions. In the paper, I highlight: - How Theory of Mind helps AI anticipate human intentions - Why emotional intelligence in AI builds trust but must be used with care - The role of cognitive load theory in reducing mental strain - Why human-centred design and ethics are not optional, but essential For me, this is more than research. It is a call to design technology that respects human psychology, builds trust and creates lasting positive impact. Full paper here: https://lnkd.in/gZpKfyfA I’d love to hear your thoughts on how can we ensure AI development leads to better AI for humans? #AI #HumanCentredAI #Psychology #EthicsInAI #Innovation #BuildingbetterAI