Factors That Affect Robot Trust

Why would your users distrust flawless systems? Recent data shows 40% of leaders identify explainability as a major GenAI adoption risk, yet only 17% are actually addressing it. This gap determines whether humans accept or override AI-driven insights. As founders building AI-powered solutions, we face a counterintuitive truth: technically superior models often deliver worse business outcomes because skeptical users simply ignore them. The most successful implementations reveal that interpretability isn't about exposing mathematical gradients—it's about delivering stakeholder-specific narratives that build confidence. Three practical strategies separate winning AI products from those gathering dust: 1️⃣ Progressive disclosure layers Different stakeholders need different explanations. Your dashboard should let users drill from plain-language assessments to increasingly technical evidence. 2️⃣ Simulatability tests Can your users predict what your system will do next in familiar scenarios? When users can anticipate AI behavior with >80% accuracy, trust metrics improve dramatically. Run regular "prediction exercises" with early users to identify where your system's logic feels alien. 3️⃣ Auditable memory systems Every autonomous step should log its chain-of-thought in domain language. These records serve multiple purposes: incident investigation, training data, and regulatory compliance. They become invaluable when problems occur, providing immediate visibility into decision paths. For early-stage companies, these trust-building mechanisms are more than luxuries. They accelerate adoption. When selling to enterprises or regulated industries, they're table stakes. The fastest-growing AI companies don't just build better algorithms - they build better trust interfaces. While resources may be constrained, embedding these principles early costs far less than retrofitting them after hitting an adoption ceiling. Small teams can implement "minimum viable trust" versions of these strategies with focused effort. Building AI products is fundamentally about creating trust interfaces, not just algorithmic performance. #startups #founders #growth #ai

Friends in sales! A new Harvard Business Review article reveals what I've been saying all along about LLMs: "the root issue isn't technological. It's psychological." Here are Six Principles (and why behavioral change is everything): This study focuses on customer service chatbots, but the insights inform AI adoption across organizations (and what leaders need to do differently). LLMs don't behave like software. They behave like humans. Which means we need a human behavioral approach. This is about change management. (BTW, this is what we do at AI Mindset at scale, but more on that below.) ++++++++++++++++++++ SIX PSYCHOLOGICAL PRINCIPLES FOR EFFECTIVE CUSTOMER SERVICE CHATBOTS: 1. Label Your AI as "Constantly Learning" Users are 17% more likely to follow an AI's suggestions when it's framed as continuously improving rather than static. People forgive small errors when they believe the system is getting smarter with each interaction, similar to working with an enthusiastic new hire. 2. Demonstrate Proof of Accuracy Trust comes from results, not technical explanations. Showing real-world success metrics can increase trust by up to 22%. Concrete evidence like "98% of users found this helpful" is more persuasive than explaining how the tech works. 3. Use Thoughtful Recognition (But Don't Overdo It) Subtle acknowledgment of user qualities makes AI recommendations 12.5% more persuasive. BUT! If the flattery feels too human or manipulative, it backfires. Keep recognition fact-based. 4. Add Human-Like Elements to Encourage Better Behavior Users are 35% more likely to behave unethically when dealing with AI versus humans. Adding friendly language, empathetic phrasing, and natural interjections can reduce unethical behavior by nearly 20% by creating a sense of social connection. 5. Keep It Direct When Users Are Stressed When users are angry or rushed, they want efficiency, not empathy. Angry users were 23% less satisfied with human-like AI compared to more straightforward responses. In high-stress situations, clear and direct communication works best. 6. Deliver Good News in a Human-Like Way Companies rated 8% higher when positive outcomes came from a human-like AI. People tend to attribute positive outcomes to themselves, and a warm, human-like delivery amplifies that emotional boost. Focusing on psychological principles rather than technical features in chatbots and adoption will create AI experiences that users actually want to adopt, driving both satisfaction and results. Huge thanks to the authors: Thomas McKinlay 🎓, Stefano Puntoni, and Serkan Saka, Ph.D. for this tremendous work! +++++++++++++++++ UPSKILL YOUR ORGANIZATION: When your organization is ready to create an AI-powered culture—not just add tools—AI Mindset would love to help. We drive behavioral transformation at scale through a powerful new digital course and enterprise partnership. DM me, or check out our website.

The 4Cs (Capability, Control, Comfort, and Comprehension) of AI adoption have always been the real decision-makers. Neural networks were more capable than logistic regression, decision trees or random forests, yet they took longer to be adopted. Why? Because we lacked control and comprehension. Explainability helped with understanding, but without comfort and a sense of control, trust remains elusive. That’s why most GenAI products today still carry a disclaimer: “Results are AI-generated. Verify before use.” GenAI might be capable but it’s not meeting the other 3Cs. Now with the rise of AI agents, control is loosening even further and impacting comfort more than ever. It’s not just about model outputs; it’s about understanding and guiding how systems flow. In hierarchical or complex agentic architectures, it’s difficult to determine which agents will be triggered, when, and why. It’s thrilling to imagine AI dynamically making decisions, but for the teams accountable for outcomes, it can be a nightmare. We saw comfort slowly build with driverless cars, but are we ready for technologies that we can’t fully control? We have already seen mixed reactions from different segments of users. We once relied on the Turing Test to benchmark human-machine parity. The logic of the Turing test is one of indistinguishability. If interrogators are not able to reliably distinguish between a human and a machine, then the machine is said to have passed. Today, LLMs arguably pass that bar. Yet we move the goalposts, not ready to declare machines as good as humans. Now we started asking questions: which “humans” are we comparing against with. Because it’s not just about what AI can do; it’s about what we can understand, trust, and manage. Until we get there, capability alone won’t be enough. Control and comfort will take time. Until then, let’s enjoy the ride and the fancy tools that come with it. #ExperienceFromTheField #WrittenByHuman

If your AI is technically flawless but socially tone-deaf, you’ve built a very expensive problem. AI isn’t just about perfecting the math. It’s about understanding people. Some of the biggest AI failures don’t come from bad code but from a lack of perspective. I once worked with a team that built an AI risk assessment tool. It was fast, efficient, and technically sound. But when tested in the real world, it disproportionately flagged certain demographics. The issue wasn’t the intent—it was the data. The team had worked in isolation, without input from legal, ethics, or the people the tool would impact. The fix? Not more code. More conversations. Once we brought in diverse perspectives, we didn’t just correct bias—we built a better, more trusted product. What this means for AI leaders: Bring legal, ethics, and diverse voices in early. If you’re not, you’re already behind. Turn compliance into an innovation edge. Ethical AI isn’t just safer—it’s more competitive. Reframe legal as a creator, not a blocker. The best lawyers don’t just say no; they help find the right yes. Design for transparency, not just accuracy. If an AI can’t explain itself, it won’t survive long-term. I break this down further in my latest newsletter—check it out! What’s the biggest challenge you’ve seen in AI governance? How can legal and engineering work better together? Let’s discuss. -------- 🚀 Olga V. Mack 🔹 Building trust in commerce, contracts & products 🔹 Sales acceleration advocate 🔹 Keynote Speaker | AI & Business Strategist 📩 Let’s connect & collaborate 📰 Subscribe to Notes to My (Legal) Self

🎯 𝗔 𝗺𝗮𝘀𝘀𝗶𝘃𝗲 𝗻𝗲𝘄 𝘀𝘁𝘂𝗱𝘆 𝗼𝗻 𝗽𝗮𝘁𝗶𝗲𝗻𝘁 𝗮𝘁𝘁𝗶𝘁𝘂𝗱𝗲𝘀 𝘁𝗼𝘄𝗮𝗿𝗱 𝗔𝗜 𝗶𝗻 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲! Patients see potential, but support hinges on factors like explainability. Here are some key findings: 𝗙𝗶𝗻𝗱𝗶𝗻𝗴𝘀 “This cross-sectional study surveying 13,806 patients using a nonprobability sample from 74 hospitals in 43 countries found that: while patients were generally supportive of AI-enabled health care facilities and recognized the potential of AI, they preferred explainable AI systems and physician-led decision-making. In addition, attitudes varied significantly by sociodemographic characteristics.” 𝗙𝗼𝗿 𝗲𝘅𝗮𝗺𝗽𝗹𝗲: “Attitudes exhibited notable variation based on demographic characteristics, health status, and technological literacy. Female respondents (3511 of 6318 [55.6%]) exhibited fewer positive attitudes toward AI use in medicine than male respondents (4057 of 6864 [59.1%]), and participants with poorer health status exhibited fewer positive attitudes toward AI use in medicine (eg, 58 of 199 [29.2%] with rather negative views) than patients with very good health (eg, 134 of 2538 [5.3%] with rather negative views). Conversely, higher levels of AI knowledge and frequent use of technology devices were associated with more positive attitudes. Notably, fewer than half of the participants expressed positive attitudes regarding all items pertaining to trust in AI. The lowest level of trust was observed for the accuracy of AI in providing information regarding treatment responses (5637 of 13 480 respondents [41.8%] trusted AI). Patients preferred explainable AI (8816 of 12 563 [70.2%]) and physician-led decision-making (9222 of 12 652 [72.9%]), even if it meant slightly compromised accuracy.” 𝗠𝗲𝗮𝗻𝗶𝗻𝗴 “These findings highlight the global imperative for health care AI stakeholders to tailor AI implementation to the unique characteristics of individual patients and local populations and provide guidance on how to optimize patient-centered AI adoption.” 𝗜𝗻 𝗮 𝗡𝘂𝘁𝘀𝗵𝗲𝗹𝗹 AI or not, what truly matters in healthcare remains the same: trust, empathy, clear communication, and a personalized approach. That hasn’t changed, and it won’t. ___________________________________ (Source in the comments.)