AI has the potential to streamline customer service. But are companies, and customers, ready for it?Getty Images
Artificial intelligence (AI) tools are improving, and companies of all sizes are beginning to incorporate them into their operations. Customer service is no exception.
AI can reduce service costs and increase response rates for companies. Virtual agents can field client queries sent through text or by voice, thanks to models trained on transcripts of service requests. Other applications can streamline the support process by determining customer intent, correctly directing inquiries, and improving document search and retrieval processes.
AI and connected-worker tech make it possible to provide client support in new ways, enabling near-immediate assistance in non-office settings such as factory floors or wind farms, says Kevin Miller, CTO of industrial AI developer IFS, which works with businesses in sectors that include manufacturing, utilities and engineering.
“To us, customer service means helping our enterprise customers do their best work,” Mr. Miller says. “With AI and connected-worker technology, IFS can support them much more quickly and on a greater scale than ever before.”
But employees report feeling increasingly concerned about the spread of AI-based automation – about 60 per cent of those who regularly use the technology at work worry about its impact on their jobs, according to a survey from CNBC and SurveyMonkey. Even as businesses and their customers become more comfortable encountering AI-related technology at work and at home, many are still deciding how they want those human-bot interactions to unfold – and what information they’re willing to provide.
“AI-based chatbots are increasingly replacing human service employees, especially with the availability of generative AI,” says Dr. Elizabeth Han, an assistant professor at McGill University whose research focuses on human-AI interaction.
More than 72 per cent of global enterprises incorporated AI into their operations by early 2024, according to McKinsey, and three-quarters said they believe the tech will significantly change their industries over the next few years, or disrupt them entirely. Generative AI is the source of much of this growth. Nearly two-thirds of McKinsey respondents – and nearly double those who said the same 10 months earlier – said their companies are using generative AI regularly.
“We’ve reached an exciting inflection point in AI,” says David Parry Jones, chief revenue officer at DeepL, an AI-powered translation company based in Germany. “In the past year, AI has gone from experimentation to mainstream adoption, with businesses scaling up from testing to widespread deployment.”
Language AI is increasingly important given the global nature of modern business, Mr. Parry Jones says. Tools like those provided by DeepL, whose machine translation technology includes multiple languages, aim to help enterprises provide personalized and multilingual customer service.
In retail, for example, cross-border e-commerce represents nearly one third of global online sales, according to DeepL’s new industry report. The company also works within a country’s borders.
“For example, in countries like Canada or the U.S., which have large multilingual populations, our translation tools can help businesses communicate locally – either internally between colleagues, or externally with customers, regardless of language,” Mr. Parry Jones says.
AI’s expanding language capabilities also come in handy for enterprise clients, Mr. Miller says. IFS serves clients in manufacturing and farm labour, industries where migrant worker forces are essential parts of their operations. Employees of IFS clients can provide and receive visual information – a photo of a part or a video for troubleshooting – that provides real-time analysis and instructions, Mr. Miller explains.
“Large language models are now so sophisticated that they can translate instructions in real-time. For example, instead of having to wait for an engineer to come help them troubleshoot an issue, migrant workers can get solutions delivered to them in the language they are most familiar with.”
These hybrid customer service approaches – ones that mix tech-based tools like automated chatbots with traditional customer-service staffers – offer both benefits and pitfalls.
For example, nailing the correct tone for automated support assistants can be tricky. McGill’s Dr. Han has studied how consumers react to AI’s emerging ‘emotional’ behaviours – its expressions, autonomous decision-making, and empathetic capabilities among them.
Positive emotions, such as the use of an exclamation mark or cheerful adjectives, increase customer satisfaction when they come from a human support agent. But Dr. Han and other researchers found those same emotions can be a turn-off if the ‘agent’ is an AI chatbot.
“We found that customers who prefer to have a friendly relationship with a customer service agent – communal-oriented customers – in fact like chatbots to express positive emotion,” she explains. “However, customers who just want a customer service agent to get the job done – exchange-oriented customers – have a negative reaction when a chatbot expresses positive emotion.”
The findings point to a potential trip-wire for applications using emotional AI, a market expected to grow to US$13.8-billion by 2032. For its part, DeepL tries to mitigate potential issues with tone through its algorithms and personalization options, Mr. Parry Jones says. This helps build client trust in its AI technology as the broader industry and regulatory landscape evolves, he adds.
Some research indicates consumers are increasingly comfortable with AI’s growing presence in their everyday lives. Though slightly more than one-fifth of consumers said they understood the technology well, 77 per cent felt neutral or comfortable using AI if it made their buying experience better, according to EY.
But developing powerful and accurate tools that rely on AI and machine learning models requires an influx of data to train them. Consumer trust in AI is relatively high, but mistrust of broader data collection efforts – the EY report notes only 18 per cent of people are willing to share personal data for ad personalization, for example – points to a significant potential hurdle for the sector’s continued red-hot growth.
AI’s rapid pace of change offers opportunities to learn more about how customers are interacting with these tools, Dr. Han says.
“At first, we might feel strange and threatened by these capabilities. We need to ultimately make sure how to utilize these capabilities in a safe and efficient way and maximize the benefits.”