Will AI mean the end of call centres?

Ask ChatGPT if AI will replace humans in customer service and it offers a diplomatic answer, the summary of which is: “They will work side by side.”

However, people are not so optimistic.

Last year, K. Kritivasan, chief executive of Indian technology company Tata Consultancy Services, told the Financial Times that artificial intelligence could soon mean there is “minimum requirement” for call centers in Asia.

Meanwhile, the AI ​​will independently decide 80% of common customer service problems by 2029, predicts business and technology research firm Gartner.

There is currently a lot of hype around “AI agents”. This is a term for artificial intelligence systems that can operate more autonomously and make decisions.

They could improve upon current non-AI chatbots, known as “rules-based chatbots,” which can only answer a specific list of questions.

My own recent experience with parcel delivery company Evri's chatbot illustrates the current state of affairs outside of artificial intelligence.

My package didn't arrive and Ezra (the chatbot's name) suggested we “resolve this issue immediately.”

It asked for a tracking number and after I entered it, I was informed that my package had been delivered.

I was able to request proof of delivery, and when I did, I was shown a photo of the package…at the wrong front door. And there was no opportunity to continue the conversation after this “evidence” was presented.

In response, Every told the BBC it would invest £57 million to further improve the service.

“Our smart chat uses tracking data to offer the most helpful responses and ensure that a customer's package is delivered as quickly as possible if it does not arrive as scheduled,” it said.

“Our data confirms that the vast majority of people get the answers they need the first time in our chat within seconds. We always check feedback to ensure our services are as useful as possible and continue to make improvements as needed.”

On the other hand, rival parcel delivery firm DPD had to shut down its less restrictive AI chatbot after it criticized the company and berated users.

Finding the right balance between brand loyalty and genuine customer service is a challenge for businesses when adopting AI.

Some 85% of customer service executives Artificial intelligence chatbots are being studied, tested or deployed, according to Gartner. But it also turned out that only 20% of such projects fully meet expectations.

“With AI, you can have a much more natural conversation,” says Garner analyst Emily Potosky.

“But the downside is that the chatbot can hallucinate, it can give you outdated information or tell you something completely wrong. I would say that for package delivery, rules-based agents are good because there are so many permutations of questions about someone's package.”

Resources and money are some of the key reasons why companies might consider moving from human customer service to artificial intelligence. But Ms. Potoski notes that it is not certain that AI will be cheaper than humans.

“It's a very expensive technology,” she says.

The first thing any business looking to replace humans with AI must do is ensure they have extensive training data.

“There is a perception that knowledge management is becoming less important because generative AI can solve the problem of knowledge not being well organized, but in fact the opposite is true,” Ms. Potosky adds.

“Knowledge management is more important when deploying generative artificial intelligence.”

Joe Inzerillo, chief digital officer at software giant Salesforce, told the BBC that call centers provide a fertile training ground for AI, especially those that have been relocated to low-cost regions such as the Philippines and India.

This is due to the fact that there will be a large number of staff trainings, from which the AI ​​can also learn.

“You have a huge amount of documentation, and this is very useful for the AI ​​when it takes over the first line of defense,” he says.

Salesforce's AI-powered customer service platform AgentForce is currently used by a range of clients, from Formula 1 to insurance company Prudential, restaurant booking site Open Table and social media site Reddit.

Mr. Inzerillo says that when Salesforce first tried the platform, it learned some valuable lessons about how to make AI more human-like.

“Even though the person might have said, 'Sorry to hear that,' the agent simply opened the ticket,” Mr. Inzerillo says.

Therefore, the AI ​​has been trained to show more empathy, especially when a customer is having problems.

Salesforce also found that preventing an agent from talking about competitors proved problematic.

“This backfired when customers asked legitimate questions about the Microsoft Teams integration with Salesforce,” Mr. Inzerillo says. “The agent refused to help because Microsoft appeared on our list of competitors.”

The firm subsequently replaced this strict rule.

Salesforce has ambitious plans to further roll out its AI agents, and so far the company says they are proving popular with its customers. It also states that the vast majority of customers, 94%, prefer to interact with AI agents when given the opportunity.

“We are seeing customer satisfaction levels exceed what people get from people. Then AI can unlock a new level of customer service,” says Mr. Inzerillo.

It also meant the firm cut $100 million in customer service costs, but he sought to play down recent headlines that said it had led to 4,000 job cuts.

“A very large percentage of these people have been redeployed to other areas of customer service.”

Fiona Coleman runs QStory, a firm that uses artificial intelligence to offer call center workers greater flexibility in their shift patterns. His clients include eBay and NatWest.

While she sees the value of artificial intelligence in improving working conditions, she is not sure the technology can ever completely replace humans.

“There are times when I don’t want to engage in a digital interaction and want to talk to a person,” she says.

“Let's see what it will look like in five years – whether AI can apply for a mortgage or talk about a debt problem. Let's see if the AI ​​becomes sensitive enough.”

In fact, the use of AI in customer service may already be facing a backlash.

Legislation currently proposed in the US to move offshore call centers back to America would also require businesses to disclose their use of AI and transfer the caller to a human if asked to do so.

Meanwhile, Gartner predicts that by 2028 the EU may oblige what is called the “right to speak to the person” as part of consumer protection rules.