Customer support in the age of AI: what can go wrong

Customer support is the first candidate for automation and the replacement of live people with robots. At least, this is the opinion of the vast majority of businesses. This is also confirmed by the number of orders.The motivation is clear: a large number of staff can potentially be laid off. And the economic effect can be easily calculated, unlike other implementations.What can go wrong?Read also Artificial Intelligence vs. Humans: Why Businesses Choose to Reduce StaffThe American financial company Klarna replaced hundreds of customer service employees with artificial intelligence, considering it a cost optimization and a step into the future. However, the experiment failed: customers began to complain massively about the quality of service, and the service lost its flexibility and empathy. As a result, the company brought people back, recognizing that full automation is not able to provide a proper customer experience.A similar scenario occurred at Commonwealth Bank of Australia, which replaced 45 call center employees with a voice bot. They expected to reduce calls, but got the opposite effect – the number of calls increased, and the system failed. The bank was forced to apologize, restore jobs, and publicly admit the mistake.Why it happens?Indeed, the lion's share of support calls are typical and easily automated. "The lion's share" does not mean "all". Most complex requests arise due to internal bugs in IT systems, lack of clear documentation or its obsolescence, complexity of systems, long history of interaction between the client and the company.And also because of the unwillingness of businesses to play transparently. Try refusing services, closing your account, switching to a lower tariff, etc., and understand all the nuances of insurance or warranty contracts. Neither artificial intelligence nor non-artificial intelligence can quickly figure it out.And add to this completely outdated systems where knowledge bases are maintained (who has them in markdown format?) and a bunch of IT systems that need to be connected in parallel. All this has a bad effect on the result.In my opinion, we need to rethink the meaning of customer support.First of all, recognize that customer support communicates with people who are dissatisfied, frustrated, and close to abandoning services. A customer who has had his or her problem solved quickly and efficiently is a loyalty bonus.People generally hate customer support. It's important to understand why. And turn a negative into a positive.Customer support should be exemplary. It should be given all the possibilities – generating auto-responses/documents, access to information systems and the necessary CONTEXT.Read also Trump's Negotiation Fever: What to Expect NextBut not to think in terms of firing people or replacing them.And the point is not that humans are empathetic and AI is not. It's not. The point is that neither humans nor AI often have the right context. But AI won't have all the company's CONTEXT available in its memory anytime soon. Unlike humans, who can store this context informally.Therefore, we need to start considering customer support as an asset, not a liability. We need to work to reduce the flow of requests, not to solve them heroically. We need to use all possible technologies to make processing fast, high-quality, contextual.This is the main task of business.Original

Customer support in the age of AI: what can go wrong

Comments (0)

AI instead of the army. Pentagon's "digital brain" broke Maduro's regime – will Greenland stand it

How to create a warm atmosphere at home during blackouts: tips from an interior designer

Business Week. Rapid de-shadowing, Venezuelan blitzkrieg, and the dollar's roller coaster

Contract from a Turkish prison: How Poltava merchants circumvented Russian obstacles to trade

A future without mines: How Ukraine is building a unique humanitarian demining system