Building customer trust in the age of AI

It’s hard to escape AI, from anthropomorphic cats completing deliveries on social media to using it for work, we’ve truly entered a new era. While there is still a lot to learn about AI and how it’s going to shape society and work, there is one thing we’re sure about.

People aren’t very optimistic about this new era. Pew Research found that 59 percent of the public have “not too much” or “no confidence” in U.S. companies to develop and use AI responsibly.

That skepticism can color how a customer looks at your business for using AI, so how are you supposed to build a relationship with them? In this post, we’ll look at how building trust is a cornerstone of this new age and how to do it.

People don’t really trust AI

Technology journalist Sam Biddle posted a pic he took of a menu at his local deli on X. The menu featured AI representations of the food they served, including shrimp that looped around on themselves. Biddle called the menu “Lovecraftian food horror” and the post went viral, racking up 3.2 million views and 31,000 likes.

This caused other people to share the bad AI food photos their local spots featured, creating a sort of viral avalanche of people complaining about local small businesses using AI to advertise their food.

A Gallup poll found that only 19 percent of people viewed businesses using AI in ads positively. Nearly 49 percent of people viewed businesses using AI in ads negatively.

Specifically, 62 percent of people found it unacceptable to create people or voices in ads using AI. Around 46 percent of people found using AI-created text, image or video in ads was also unacceptable. Basically, people do not like the idea of AI taking away potential work or opportunities from people.

Horror director Eli Roth found himself with AI backlash with his new film Ice Cream Man. Initially, he had talked about how he hand-drew animated sequences in the movie. Eagle eyed viewers later spotted the credits mentioning AI visual effects company Dark Half. When confronted, Roth issued a statement to Polygon confirming he used AI and that he “misspoke” earlier about its use in the movie.

While all these cases are somewhat different, they all point to the same effect: AI breaks the trust of people, whether they be audiences or customers. Why does this happen? It’s primarily because people feel tricked.

When people see a visual menu of your restaurant, they expect to see what their food will look like. If you give them AI images instead of actually taking photos of the food you serve, they feel duped. It’s the same for people expecting a piece of art to be made by a human, especially when someone says they handmade it.

Why would a customer or client trust your business enough to purchase your goods or services if they can’t trust your advertising, and essentially, your use of AI?

How to build trust

Creating trust starts with understanding perspective. According to a Stanford survey, 52 percent of people say AI makes them nervous. That’s up from 50 percent last year and 39 percent in 2022.

This technology makes people uneasy and, more importantly, it scares them. Keep that in mind as you approach the technology and you’ll have a good mentality for any AI decisions you need to make.

Second, we’re going to incorporate some lessons from the National Institute of Standards and Technology (NIST). It developed an AI risk management framework that incorporates trustworthiness. The framework is designed to help individuals, organizations and society to manage the risk associated with AI.

The framework identifies various traits that a trustworthy AI system includes. For example, it says that valid and reliable information is the base of trustworthiness. Above that base is safety, security and resiliency, explainability, privacy and fairness (with harmful bias managed). Connecting the two layers is accountability and transparency. All of this, together, creates a trustworthy AI system according to NIST.

While this framework is intended for those creating AI systems, there are some things we can learn from them.

Validity and reliability

This is the foundation of creating trust. Everything you do with AI has to include valid, accurate information and be done consistently over time. For example, if you decide to use AI to visually showcase your menu items, that’s not valid and reliable.

It’s not valid because that’s not what your food looks like, and it’s not reliable because if you ask AI to do it again it might draw up something very different. Neither is helpful to you or your customer.

The same goes for inputting business information into your AI systems. If you’re using it to analyze sales data you must make sure your data is valid and accurate, and then make sure what you’re asking for is repeatable to a certain degree. For example, creating a prompt that gives you a chart of information for the week and compares it to sales from a month, quarter or year ago.

Accountability and transparency

This is where a lot of AI backlash comes from, like Roth with his film. If you’re going to use AI for advertising or something else, you must be transparent about its use.

If you use AI on your website, or for an image or video, even if it’s a small part of it, you need to let people know. If you have a physical location or a restaurant and you have AI art you made, or AI menus, you must disclose it somewhere. It can be in small text on the menu itself or hung up on the wall where customers can easily see it.

Remember, the breakdown of trust in the AI era largely comes from people feeling tricked, and the way to avoid this is an almost radical level of transparency if you must use AI. Similarly, you have to be accountable for its use.

That accountability includes what could happen if you use AI in certain ways that people aren’t happy with, like creating people for an ad. It also includes its potential mistakes.

One of the best ways to create both accountability and transparency is to create an AI statement.

What is an AI statement?

Simply, it’s the best way to build trust in the AI era. An AI statement is your business’s public declaration of how you use AI safely, ethically and transparently.

Your AI statement should include:

• Purpose: Why your business uses AI in the first place.
• Where it’s used: Explicitly define where in your business AI is used. Is it for operations, or brainstorming recipes and art, or do you use it to help with accounting?
• Principles: You need to state your take on how you handle accuracy, privacy, bias reduction and even sustainability. These are issues people are concerned with, so stating how you protect customer data, ensure accurate information, reduce potential bias and even offset the environmental impacts of AI is important.
• Accountability: How do you handle AI errors? Is everything validated by a human being?
• Audit: A commitment to updating policy at least once a year.

Keep a human in the loop

UNESCO administered recommendations for the ethics of AI use, and one of its core principles is “Human oversight and determination”. Essentially, “member states should ensure that AI systems do not displace ultimate human responsibility and accountability”.

That means making sure a human checks anything that AI does. It needs to validate all outputs from AI to make sure it’s as accurate and reliable as possible, and also to make sure that someone is accountable for those outputs.

The other thing about AI work is that it can make your business look lazy. If a customer sees those circular shrimp, they may be wondering who in your business looked at that and thought it was okay to put out there. They may question your judgment, which will make them question why they would want your services in the first place.

It’s the same thing for a website, or copy. If they look or read something and catch an error that AI made, or it feels too much like AI, they may think your business is lazy. Some generations, like Gen Z, may then refer to your business as “AI slop”. While people can only identify AI text and images about 50 percent of the time, there’s enough possibility there for you to take the output seriously.

Build trust proactively

Creating trust in this wild west of AI has to be done proactively, and you can do it with the above AI tips.

You need to learn and understand that people don’t trust AI because it often makes them feel tricked and they don’t trust people in power to use or craft it responsibly. Then you need to practice radical transparency, create an AI statement and keep a human in the loop to ensure validity, reliability and accountability.

While you can’t rid yourself of all the risk of using AI, as so much of it depends on the companies building the systems we use, these steps will get you on the path to building trust with the people that matter the most to your business.