Air Canada AC-T has been experimenting with incorporating artificial intelligence into its operations for years, but it wasn’t until recently that the company hit a turning point of sorts.
Starting this summer, travellers impacted by unexpected flight delays or cancellations will no longer need to have phone conversations with human call-centre agents. They’ll be able to deal with an AI-powered voice assistant instead.
“So, in the case of a snowstorm, if you have not been issued your new boarding pass yet, and you just want to confirm if you have a seat available on another flight, that’s the sort of thing we can easily handle with AI,” Mel Crocker, the company’s vice-president and chief information officer, said in a recent interview.
Over time, Mr. Crocker said, the technology will learn from the problems it deals with and gain the ability to resolve even more complex customer service issues. The airline expects this to drastically reduce the amount of time its passengers spend waiting on the phone.
But the technology could also have a drastic effect on a different group of people: Air Canada’s customer service agents.
In theory, they will be freed from dealing with simple customer issues, so they can focus instead on more complex ones. But if the technology gets smart enough, it could one day eliminate their jobs altogether.
“We’re not going into this with a view of killing jobs,” Mr. Crocker said, noting that the company’s staffing levels have remained relatively unchanged for years. “But if we can use a human to solve something that requires a human touch, and technology to solve something that can be automated, we will do that.”
Call-centre jobs are ripe for automation. The problems agents deal with on the phone tend to be repetitive – a customer sorting out a payment issue, checking on the status of a shipment, or arranging a service.
AI is developing at such a rapid pace that the people building it predict it will soon eliminate the need for many customer service agents, or change their jobs so dramatically that many of them will have to be retrained to work alongside the technology.
Whether a person thinks this is good or bad depends largely on who they are, where their paycheques come from, and what they think about the future of work.
AI developers and employers argue that automating call-centre jobs will improve the overall customer experience. And they say it will eliminate mundane jobs that most workers don’t enjoy anyway. The workers that remain in the field, they say, will become “super-agents,” tasked with solving problems robots can’t, and potentially earning better pay for doing so.
But labour advocates warn of the potential side effects of AI, particularly constant employee surveillance. They envision the technology being used to gather and process data on how quickly employees work to solve customer issues. They also say call-centre workers might not necessarily want their jobs to become more complex. Solving simple customer issues can be gratifying.
“We are already looking into how AI is going to change the jobs of our members,” said Kaylie Tiessen, an economist with Unifor, Canada’s largest private-sector union. “What’s critical for us is that companies consult the people who actually do the job to understand what skills they have been using to execute the task.”
CallMiner, a Massachusetts-based software company, has spent years developing AI-driven conversation analytics software geared specifically toward call centres. One of the company’s most popular tools analyzes conversations with customers and provides online assistance to agents.
The tool can guide an agent into selling new products to a customer. “Because our AI has analyzed tens of thousands of pieces of data from numerous conversations, it can actually predict what additional purchase a customer may want,” said Jeff Gallino, the company’s co-founder and chief technology officer. The tool can also track agents’ on-the-job performance.
CallMiner’s AI system can be incorporated into apps and websites, where it can interact directly with customers. In these cases, it serves the same purpose a human representative does: selling things to clients, or helping them modify services they have already paid for.
“If I need to add a channel to my cable provider, sometimes an agent attempts to upsell me and I get frustrated, because that’s not the right product I need. But what we’ve found is that the AI can do it better, because it already knows, based on the customer’s history, what additional service the customer may want,” Mr. Gallino said.
CallMiner’s software is used primarily in retail banking and by insurance companies, according to Mr. Gallino. The company has a few key Canadian clients, which he declined to name. He would say only that they consist of a handful of banks and a municipality.
Mr. Gallino said he has started to see a shift in the way companies are incorporating AI into customer service. “The mundane tasks are being moved to cheaper, automated channels. This will definitely mean fewer jobs of this sort in the future, but there are still workers needed to do the hard part of the job. It also means they tend to stick around longer because they are more challenged,” he said.
Information on exactly what proportion of Canada’s labour force works in call centres as customer service representatives is scarce.
“It’s hard to tell how automation is going to impact this group of workers when companies aren’t required to disclose how many workers of this sort they have, and whether they are abroad or work domestically,” said Sean O’Brady, an assistant professor at McMaster University’s DeGroote School of Business whose research is focused on automation and call-centre jobs.
According to Statistics Canada, in 2021 there were 223,825 workers providing “customer service information” to the retail, insurance, telecom and utility industries. But that number does not account for the entire customer service industry. For instance, it does not include customer service representatives in the travel or banking industries.
Through his research, Prof. O’Brady has found that unionized call centres have been the slowest to automate. And he has found that there have not yet been significant losses of call-centre jobs across the country because of AI.
What is happening instead, he said, is that companies are installing AI-driven apps (such as the software CallMiner sells), which provide advice to agents in real time as calls are taking place.
In other words, in the customer service realm, AI is still in the process of changing the nature of jobs, rather than replacing human labour altogether.
“It certainly can improve customer service. On the other hand, it could tell experienced agents to go in directions that do not follow their instincts,” Prof. O’Brady said.
The Massachusetts-based company Cogito is a leader in this segment of AI technology. It sells software called Dialog that coaches agents to “enhance” their emotional intelligence while on calls.
Virginia Doellgast, a professor of comparative employment relations at Cornell University who has extensively studied Cogito’s technology, describes it in one of her latest research papers. “As agents are talking with customers, a screen pops up and tells them that they should speak with more empathy or energy, described as an ‘empathy cue,’” she writes.
Prof. Doellgast has argued that these technologies increase management’s control over workers and limit their ability to complain or contest unfair decisions. She has also found that lack of control or perceived unfairness in a work setting are associated with a higher degree of worker stress and burnout.
One way AI technology in call centres could tighten management’s grip is by monitoring worker behaviour.
“It can provide information to employers if say, an employee has a lot of pauses in the course of a conversation,” Prof. O’Brady said. “It will ask why there are pauses, and whether there are negative productivity outcomes associated with it.”
Unions like Unifor have already begun expressing concerns about the worker-surveillance aspect of AI.
At Unifor, some bargaining units have set up committees that monitor how technology could fundamentally change workers’ job functions. The union is also looking at how collective agreements could be renegotiated to include language that ensures employers notify the union and workers well in advance of technological change being introduced.
Unions have been dealing with the threat of automation for decades, particularly in auto plants, where robotics have forever changed assembly lines. Some unions paying attention to this latest wave of technology-related labour displacement have managed to win significant concessions from employers.
In 2018, the Culinary Workers Union local in Las Vegas won the right to six months notice before the introduction of new technology that could result in worker displacement. It also got employers to agree to train workers to use the technology at no cost to the employees.
CallMiner’s Mr. Gallino said it will ultimately be difficult to stop large-scale AI adoption in customer service, considering the significant cost savings associated with using the technology instead of human labour.
“That same human is costing dollars and dollars per interaction. Even if a company can make their calls a minute shorter using AI directly, or AI assistance, that’s millions of dollars saved over time.”
Air Canada’s initial investment in customer service AI technology was much higher than the cost of continuing to pay workers to handle simple queries, Mr. Crocker said.
But over time, he said, the airline believes investing in automation and machine learning technology will lower its expenses.
“But that’s not our primary reason for doing this,” he added. “The biggest benefit of AI to us is that it fundamentally creates a better customer experience. And happier customers means they are travelling more with Air Canada.”