How to Get Your Customers to Trust AI

Transparency is supposed to build trust. But as companies rush to open the black box of artificial intelligence and explain how it works to customers, many are discovering a surprising truth: You can say too much and too little at the same time. The balance is hard to get right: Too little transparency breeds suspicion; too much overwhelms, blurring the very clarity it’s meant to provide. Customers contend with organizations’ attempts at transparency in the swirl of digital messages, disclosures, and prompts that they encounter in their interactions. A website touts its commitment to ethical AI, yet the explanation is buried behind a lengthy policy page that few may ever read. A product listing includes a disclaimer in fine print saying recommendations are “AI-informed,” without clarifying how the system works or what data it draws on. And then there are the exchanges with AI itself, when a chatbot declares that it’s powered by artificial intelligence but gives no insight into what that means for the customer or their information. Each of these gestures aims to be clear, yet many fall short. Some swamp customers with technical nuance or legalese, leaving them more confused than informed. Others offer only surface-level reassurance that feels incomplete. In both cases, customers start filling in the gaps on their own, and over time, that guesswork becomes its own form of mistrust. These moments capture the central paradox: Forms of transparency that are meant to build confidence can just as easily undermine it. Here is how your company can thread the needle in your interactions with customers. 1. Treat Transparency as Part of a Broader Trust System One reason transparency is so difficult to get right is that it doesn’t stand alone—it operates within a broader system of trust. Organizations earn trust when they communicate in plain, simple language, show they care, deliver quality experiences and products, and do so reliably. These four factors of trust—transparency (whether an organization communicates openly and clearly, humanity (whether it demonstrates empathy and fairness), capability (whether it delivers quality products and experiences), and reliability (whether it keeps its promises consistently)—form the basis of Deloitte’s TrustID score, a composite measure that meaningfully predicts behavior, including employees’ adoption of AI. Deloitte’s TrustID index, which includes survey responses from more than 500,000 U.S. customers annually, shows that these factors don’t move independently; they rise and fall together. Strengthening one can lift another, but it can also expose where others fall short. Transparency most strongly shapes perceptions of humanity and reliability. Its influence on capability tends to be weaker, evidence that openness alone doesn’t convince customers of capability. In practice, when companies explain in plain, human language how their AI makes recommendations, customers feel more informed and reassured. Perceptions of humanity and reliability rise sharply, while capability gains remain modest because transparency builds understanding but not proof of performance. Consider a financial services company that introduces an AI-powered planning tool designed to surface real-time investment guidance. A customer trying it out enters the same question on two separate occasions and is surprised to see the tool offer different recommendations each time even though nothing about his or her financial situation has changed. To the customer, it feels like the system is glitching. No one explained that the model continues to learn and adjust as new data comes in, quietly recalibrating itself with every new tick of market information. So what’s meant to feel trustworthy instead feels unreliable. And once reliability slips, customers begin to question whether the system works. That’s when humanity and capability scores begin to drop too. Autodesk, a multinational design-and-make software company, has been measuring the four factors of trust with customers since the methodology was first published in 2023. These insights don’t just sit in a report; they actively inform how the company designs products and communicates with users. One theme that kept emerging from those trust measurements was transparency. Customers said they wanted to better understand how Autodesk’s AI features worked and what data they relied on. So, the company decided to make transparency tangible—creating what it calls “AI Transparency Cards,” which are modeled after nutrition labels on food packaging. Each card breaks down an AI feature into its core components: what the model does, what data it uses, how that data is protected, and what safeguards are in place. The idea was to give customers a clear, concise “ingredients list” for AI—something they could digest at a glance. For less experienced users, Autodesk includes a companion page of plain-language explanations that define each term and concept—so even those without technical backgrounds can understand how the AI works and what data practices support it. For example, in Autodesk Construction Cloud—now part of Autodesk Forma, Autodesk’s unified industry cloud for planning, designing, building, and operating—the Photo Autotags feature uses AI to scan jobsite photos and automatically tag objects such as scaffolding, wiring, and safety gear so teams can organize and find images more easily. The transparency card shows what the model does (automated photo tagging using computer vision), where the data used to train the model comes from (customer content), and how it protects user data (it doesn’t use personal information and data is securely stored and transferred). AI Transparency Card for the Photo Autotags Feature of Autodesk’s Construction Cloud But Autodesk didn’t stop at creating the cards; it also wanted to know how they shaped trust. At its annual user conference, the company ran a live study to find out: Could a simple, plain-language “ingredients label” for AI meaningfully increase customers’ trust in it? The company gathered quantitative survey data from customers to establish a baseline TrustID score. Then, through intercept interviews, participants were surveyed both before and after viewing the Transparency Cards to capture changes in perception and to gather qualitative insights on why the cards built or eroded trust. The approach combined data and dialogue: measuring not only whether transparency moved the numbers, but how it felt to the people experiencing it. Autodesk saw the four factors evolve in real time. When customers engaged with the cards, transparency scores jumped 26%, and reliability rose 13%, showing that customers appreciated the effort to explain how the company’s AI worked and how it was developed. Those gains confirmed that transparency builds confidence, but they also revealed a tension beneath the surface. As perceptions of transparency and reliability increased, those of capability and humanity fell by 5% and 3%, respectively. Clarity, it turned out, wasn’t enough. The more customers understood how the AI operates, the more they wanted proof that it worked well—and that it was designed with their needs in mind. The interplay among these trust factors is reflected in Autodesk’s composite score: a 7% overall lift in customer trust. It’s a strong result, but more importantly, a breakthrough insight: Transparency didn’t just improve perceptions of openness; it raised expectations for everything else. 2. Tailor Transparency for Different Audiences The results of Autodesk’s work highlight another truth: Everyone doesn’t view transparency the same way. Some customers said the cards were too technical; others wanted even more detail. The conflicting reactions weren’t contradictions; they were clues. Deloitte’s broader trust research shows that what people want to know—and how they process information—depends on who they are, what role they play, and how they think. Who you are shapes not just how much transparency matters to you but also how you expect it to show up. For example, from ongoing trust research we know that older generations tend to be more cautious. They’re not indifferent to transparency; they’re simply harder to convince that brands are being forthright. Younger audiences, by contrast, believe brands are more transparent and place greater value on transparency as a driver of trust. But they also trust differently. Gen Z doesn’t take brands at their word; they verify through peers, creators, and communities. Their trust is social, skeptical, and participatory—built through shared experience rather than top-down communication. Findings from the TrustID index indicate that what role you play—and how you got there—also shapes what you need to know. Those in technical fields like IT, finance, and engineering seek depth and proof. They want to see the data and understand how it’s used. Those in customer-facing or operational roles care more about clarity and impact: how AI decisions affect their work and the people they serve. Education is another core driver: People with advanced degrees tend to expect greater transparency and reward the organizations that provide it. They’re used to parsing complex information and want precision over simplification. Leaders who get this right treat transparency as a communication design challenge, not a compliance exercise. They surface the information that helps people make sense of what they’re seeing while keeping the rest accessible but optional. The goal isn’t to reveal everything; it’s to create understanding. 3. Make Transparency Iterative Transparency isn’t a single moment—it’s a continuous loop. The shelf life of transparency is getting shorter. As systems change and customer understanding deepens, what once felt open can quickly feel outdated. The more advanced the technology, the faster expectations move. That’s why leading companies treat transparency as a feedback mechanism—a way to learn what builds confidence and where uncertainty remains. They don’t just share information; they listen to reactions, measure responses, and adapt. The act of closing the loop becomes part of how they earn credibility. Autodesk has seen this feedback loop play out directly with its customers. One customer shared that he had used the AI Transparency Cards with his own client, who had been hesitant to bring AI into active projects. The clear breakdown of how the model worked and what data it relied on eased the client’s concerns enough to move forward. It was a real-world proof point that transparency can shift not only perception but also behavior and that the impact of these efforts often extends far beyond the moment of disclosure. That kind of feedback only deepened the company’s commitment to iterate. After Autodesk’s first round of customer testing, the company didn’t declare victory; it went back to redesign. The next generation of Transparency Cards will go deeper, disclosing more about the underlying models behind each AI feature: their names, types, and dataset sources. Their Trusted AI team is also exploring how to communicate the quality and reliability of each feature’s output: how it can help customers understand not just how Autodesk builds AI but also how much they can trust its results. Through expandable menus, the reader will decide what level of detail they want. When done well, this iterative model turns transparency from a static statement into an active dialogue. It reinforces that understanding isn’t an end state; it’s a relationship. Over time, every act of clarification, correction, or follow-up adds up to greater credibility. The payoff for getting transparency right is tangible. According to the TrustID index, when customers rate an AI system as highly transparent, they’re 1.6 times more likely to share personal data, 8.5 times more likely to express high trust in the brand, and 1.6 times more likely to use AI tools daily. Those results reinforce a simple truth: Transparency is a performance driver. That’s why the companies that deliberately measure and manage trust are designing transparency as an ongoing relationship, not a one-time disclosure. They tailor it to different audiences, connect it across systems of trust, and continuously refine it based on feedback. Transparency doesn’t mean showing everything. It means showing what matters—clearly, consistently, and with purpose.

How to Get Your Customers to Trust AI

Summary.

Partner Center