Regulators critical of AI chatbots in customer service

The lack of human interaction in customer service is considered one of the biggest annoyances, according to a recent study by the Netherlands Authority for Consumers and Markets (ACM). Chatbots sometimes provide "poor or even incorrect answers." This often leads to people being talked to without actually being helped.

Street language and jokes

Moreover, it's not always clear whether consumers are interacting with a chatbot or a real employee. This is because the computer sometimes uses jokes, mistakes, slang, and emojis to appear more human. Chatbots are also sometimes referred to as customer service representatives, and responses are deliberately delayed, giving the false impression that a real person is typing.

Once consumers have become involved in a conversation with a chatbot, it is not always possible to speak to a human afterwards.

People remain important

"It remains very important that customer service is organized in such a way that a person is always available," says Edwin van Houten, director of consumer affairs at the Netherlands Authority for Consumers and Markets (ACM), to RTL Z. He mentions dozens of complaints in the 'recent period'.

"Using AI is fine, chatbots can work very well for accessibility and answering easy questions, but as an organization, always make sure that you are represented by a human being," he continues.

The regulators say that companies are obligated under consumer law to communicate directly, effectively, and accurately with their customers. "This is part of the quality standards you, as a consumer, can expect from customer service," Van Houten adds.

Digital Services Act

However, his organization and the Dutch Data Protection Authority (AP) observe that these regulations are not yet sufficiently clear for every company. "This clarity already exists, thanks to the Digital Services Regulation (DSA) (Editor's note), for so-called 'intermediary services' such as social media, marketplaces, and online platforms," ​​according to the ACM and the AP.

This regulation explicitly states that there must be the option "to choose direct and efficient means of communication that do not rely exclusively on automated tools." Such an addition is lacking in consumer law and, according to the regulators, should also be legally enshrined here.

Stricter rules

It is also noted that European regulations will become stricter next year. Companies will then have to make it clear from the very first interaction that they are communicating via an AI chatbot.

However, the ACM and the AP could find the rules even stricter. There are also concerns about privacy. "Chatbots are a form of generative AI, which is trained with large amounts of information and data. This may also include confidential data and documents."

"This could allow—for example, malicious actors—the chatbot to provide more information than necessary to answer 'regular' customer questions, and to steal this confidential information. This compromises data security and could even lead to data leaks ," the AP writes.

People are not always better

The Netherlands Authority for Consumers and Markets (ACM) and the Dutch Data Protection Authority (AP) are in discussions with companies. The Customer Service Federation (KSF), which has several large companies as members, endorses the call and "is happy to cooperate with the regulators." Work is underway to develop guidelines for the ethically responsible use of AI, the federation writes.

Not being able to reach a customer service representative can lead to some pretty drastic scenarios, as Mart noted:

The KSF doesn't agree with the regulators on all points. According to the federation, more regulations don't automatically lead to better customer service. "Regulations tend to become an end in themselves," says KSF director Roel Masselink. "What really helps is transparency and smart collaboration between people and technology. It should be about whether the customer is being served properly."

Better served by computer?

He further says that the use of AI technology in customer contact is inevitable due to the shrinking working population, increasing customer demand and increasing cost pressure.

Van Houten isn't worried that the AP and "his" ACM's call will be overtaken by the times. After all, it could be argued that AI is becoming increasingly smarter and better, meaning that computers can also better assist customers. He says he's very interested in the development of chatbots, but that currently, it's been concluded that they're not yet good enough. "And even now, customer service needs to be in order."