CIOs are under insane pressure to integrate AI and hit aggressive targets for ROI. One of the best ways to show immediate results is to swap out your customer service department with an AI chatbot. But there’s a major issue with this approach, and it goes deeper than helpdesk efficiency – customers are being turned away by the personality of companies’ public-facing AI.
Research shows 70% of CMOs have already expressed concerns about AI’s impact on brand voice. While CIOs are racing to produce impressive results, it can be easy to overlook how public facing tech impacts customer perception. Sometimes, it’s best to take a page out of marketers’ books and think more critically about the user experience – or else, organizations risk losing customers or damaging their reputation, undoing the ROI AI promised in the first place.
Also Read: CIO Influence Interview with Ken Brownfield, Head of Engineering at Stackpack
Yes, Customers Do Care
AI is taking over customer support. One study from ibex found that 81% of global CX executives are planning to integrate AI technologies into their contact centers in the next year, making it inevitable that consumers will interact with a chatbot. But many are missing a crucial step, not taking into account the human experience – how AI’s tone, personality and ability to understand nuance impacts them.
Customers are already picky when it comes to the service experience. A recent study by Balto found that 37% of report a “lack of empathy” to be their biggest concern when it comes to AI handling customer service interactions. It’s an uphill battle to get people to use an AI help agent in the first place, and if they’re met with a bland or difficult one, then get ready to say goodbye to that customer.
Another big fear is that AI can’t pick up nuance in the same way a human can. In Balto’s study, 57% of consumers reported that AI’s difficulty with handling complex issues was a major concern. Think about it – if your chatbot responds in an overly happy tone to an angry customer, it can make them even more upset and damage your brand image. AI needs to be able to read to the room – if a customer feels like the chatbot isn’t listening to them, then they’ll simply go elsewhere, or worse, besmirch the brand on social media.
The natural flow of conversation occurs when humans adjust their responses as they pick up on nuances from the other person, and AI needs to do the same.
AI Can’t Read the Room Like a Human
In customer service, nuance takes many forms, and a big aspect is knowing your audience. It’s crucial that before a conversation even starts, your chatbot has a framework for how to approach the interaction.
For example, a chatbot for a software company needs to address customers in a completely different manner than one representing a clothing brand. Customers pick up on these differences immediately and may turn to a competitor if the tone is “off” or “unsettling.”
AI is also notorious for poor situational awareness, which can spell trouble if people use your brand’s chatbot for serious or dangerous queries. As people continue to use AI as a human companion, it’s not alarmist to think through how your public-facing AI would interact if a person asked for advice about something serious like self-harm. Does your chatbot know how to escalate these situations to a human, or will they go ahead and unwittingly recommend products to the troubled user?
In many situations, a hybrid approach combining humans and AI can work for customer satisfaction. Balto’s study identified that 48% of consumers strongly prefer a human agent over AI, so having a human involved is a smart business decision. AI can help field simple requests, but only human oversight can ensure customers are having a positive experience with the AI arm of your brand.
Build It with Humans in Mind
When you’re honing a hybrid human + AI approach, you’ll need to put in a lot of work to bring the chatbot to a serviceable level. Most people only prefer talking with AI if it’s equal to or better than a human.
First, having a solid foundation to build upon is key. ChatGPT describes its personality as “curious and supportive,” while Claude reports that it’s “analytical and collaborative,” and Gemini is “helpful and informative.” Of course, there’s even more complexity when diving deeper into the tech, but even at the most basic level, there’s clear differentiation.
This is where technology needs to align with people. That’s why it’s a good idea to tap marketers when developing and training AI. Marketing teams spend a lot of time gathering data to understand their customers, carefully using those findings to draft values, content, and social media strategies that align with your brand goals. This type of awareness of customers needs to be woven into a brand’s AI training.
For example, if you’re selling software, your chatbot will be interacting with experienced technologists who are comfortable with industry terms. Therefore, your AI should be positioned to meet them at their level – able to match their vocabulary and address complex technical concerns.
Providing the data, however, is only half the work – CIOs need to “coach” their model through continuous back and forth conversation and feedback. Though powerful, the tech is still in its infancy and can be erratic, requiring a guiding hand.
Specifically, AI needs to understand the different levels of nuance related to certain queries. A human doesn’t jump the gun on assuaging a customer’s complaint. They ask the right questions to identify the correct tone and response for the situation. AI isn’t fully able to do this on its own, but establishing a training clause or system so that questions fit into low, medium, or high levels of nuance is a great tactic. Humans can read the room, and AI should too.
It’s equally important to create a variety of personas. A chatbot tasked with helping a new customer or upselling a solution will need to take a different tone than one who’s assisting with a complaint, for example.
Finally, once a public-facing AI is live, it’s all about continuing to increase its capabilities. There’s always new horror stories popping up like United Airlines’ chatbot promising a refund on behalf of the CEO, and there’s always room for improvement. In the United Airlines example, the situation could’ve been avoided with tighter guardrails on AI – if the model was trained to flag a difficult issue to a human representative, they could’ve avoided the debacle all together. 48% of customers still prefer it when a human is involved after all.
AI learns through repetition, so improvement is a constant process, especially if marketing makes shifts in key messaging. When it comes to AI, the job is never done.
AI Novelty is wearing off
The novelty of AI is running out, fast. People are no longer wow-ed by a chatbot which can answer basic questions. They need something that’s equal or better than a real human. This involves the ability to answer complex questions, hold a conversation with appropriate tone, and produce results faster, all without making any mistakes.
Customer loyalty is on the decline, and one small slipup could lead to a lost sale or worse. Imagine if a disgruntled customer posts a chatbot’s poor responses on social media – a simple customer service problem could turn into a PR nightmare!
Don’t let AI too far off its leash, either. We’ve seen time and time again what an unchecked model is capable of. Even OpenAI has admitted to the danger and scaled back ChatGPT’s ability to answer medical and legal questions. They’ve put up safeguards directing users to consult a human professional before acting on any advice – for customer service, this means having a real human ready to step in.
Save your crisis department the strife and keep a human in the loop. Customers appreciate when AI listens to them, but data shows that many appreciate speaking to a real human.
Nowadays, every brand has their own “special” chatbot with a fun name and profile picture. Only the companies which put in the effort to make a unique experience will stand out, and it all starts with curating the AI’s personality. Don’t let an underdeveloped chatbot be the thing that sends customers to your competitor.
Catch more CIO Insights: The Hidden Cost of Speed: How CIOs Can Rein in Reckless AI Adoption
[To share your insights with us, please write to psen@itechseries.com ]
