Would you trust an AI with your wallet?

Trust is having a moment. And for good reason.

The backlash Facebook is facing boils down to breaking the trust of consumers in how they protect their data. To rebuild consumer confidence, Facebook is turning to AI to solve their problems.

As AI is used more by brands to improve operations and decision-making, and even more so to interact with consumers on a daily basis, it is now more critical than ever to build a strong foundation of trust. The potential for AI is huge; in the near future, brands and individuals will be represented by an AI on the internet, and the only way for brands to reach their consumers will be to go through their AI. As this new AI era approaches, the trust threshold will remain fragile.

In 2015, when Sony Pictures launched “Slappy,” a conversational AI in support of the Goosebumps Movie, they saw astronomical engagement on Facebook Messenger: an average of 10 minutes per conversation, with some conversations lasting up to 2 hours. But the AI was so good that some people thought they’d made an actual new friend. One user even invited Slappy to her chemotherapy appointment the next day.

This raises some serious ethical questions about how brands present their AI to their customers. Should your bot be upfront about being a bot or try to pass off as a human agent? Is it the customer’s responsibility to approach chat, voice or email interactions with caution? [For the record, I would recommend owning the fact that an AI is an AI].

In addition to how an AI is presented, trust in an AI is built (and broken) in many ways. As AI permeates more parts of the customer journey across marketing, sales and service, consumers will need to be able to trust your AI. 

How to get consumers to trust your AI

Trust an AI with your refined intent

Does the AI actually understand what the person is asking or trying to accomplish? If the AI is constantly getting confused, the person will not trust the AI to help accomplish the task at hand and stop engaging.  AIs need to be able to understand the majority of what a person is communicating, including today’s digital slang (emojis, stickers). AIs also need to be built on a deep learning platform so it is constantly expanding its knowledge base, increasing the frequency of classifying intent correctly.

Trust an AI with your identity (piecemeal or not)

A person should never have to reintroduce themselves to an AI and rather all of the information they have revealed to the AI over the course of the relationship should help to drive relevance in future conversations. The AI should also connect to a brand’s CRM in order to personalize the interactions based on cross-channel profiles. However, in order to retain trust, always enable a person to opt-in for data sharing and storing.

Trust an AI with your wallet

AI will soon be the single channel for all brand and consumer activity, including purchases. Being able to trust the AI to store payment information will help to create a completely seamless commerce experience, greatly enhancing the customer experience. If a person doesn’t trust the brand to keep payment information secure, though, the magic of a single channel touchpoint will be lost. Clearly communicate how payment information is secure or stored if held by the brand or connect to a third-party like PayPal or Apple Pay.

Trust an AI with your emotions

If a friend tells a joke when you’re discussing something serious, you would probably grow frustrated or upset. Similarly, an AI needs to be able to read and react to a person’s sentiment and tone.  Beyond the text or spoken word, how is the person feeling? An AI can tap into this by looking at the conversation in its entirety, assigning a sentiment grade to every interaction in order to say the proper thing or send appropriate content. If an AI is a repeat offender of not understanding a person’s emotions (particularly on the poles, really happy / excited or really sad / mad), they might get fed up with the relationship.

Trust an AI to transfer context across AIs

Brands may soon have various AIs at work simultaneously managing different parts of the customer experience. There might be customer support AIs answering questions, AIs supporting human agents to reply quicker and more accurately, AIs that focus on guided selling and personal shopping, etc. AIs will be specialists. If, for instance, someone needs to return something she has ordered (with the personal shopping AI), she might need to be “transferred” to the customer support AI, who can help to facilitate the return.  If this “hand-off” occurs, the necessary information (like items, loyalty status, etc.) needs to be passed along, so the conversation isn’t starting from square one.

Today’s consumers are more comfortable with, and sometimes prefer, interacting with AI. There is no better way for brands to reinforce that positivity than by building trust. If trust is broken, it will be very hard to convince a person to give it another chance.