Use of artificial intelligence in customer service

Artificial Intelligence has proven to be a great ally of companies that provide services and/or products to the end consumer, optimizing service and facilitating the relationship with their customers.

Especially considering large-scale service, Artificial Intelligence tools not only make it possible to cut costs, but also generate an increase in efficiency and scalability in service, reducing, as a rule, the waiting time for service.

The evolution of AI tools led to the update of the Decree that regulates Customer Service (SAC). Decree No. 11,034/2022 relaxed the requirement to provide, already in the first electronic menu, personal and human service, without the need to provide the option of contact with the attendant, an obligation that was contained in the previous Decree. This is a relevant normative change that enabled the use of these tools at a broad level. 

In Complementary Health, for example, considering the maximum time for contact with the attendant, provided for in Ordinance MJ No. 2014/2008, which varies from 45 to 60 seconds, the extinction of the obligation to provide the option of contact with a human attendant in the first electronic menu generates a profound impact on costs related to Customer Service.

On the other hand, the evolution and mass use of Artificial Intelligence tools, including chatbots, in Customer Service, also brings challenges: despite the high accuracy rate in simple and more recurrent cases, as a rule, the tools are not able to properly handle complex or atypical cases. In the latter case, the need for subsequent human service may end up increasing the time of customer service.

The consumer's feeling of frustration at not having personalized service can impact on the company's image and reputation. According to a survey released in July 2024 by the consulting firm Gartner :

64% of consumers prefer that companies do not use artificial intelligence tools in their customer service centers.

The same survey indicates that:

53% of consumers would consider switching from a provider that uses AI in customer service to a competitor that does not.

These indices demonstrate a consumer's disbelief in the resolution of demands or at least in the effectiveness of the use of AI tools in Customer Service, a service that is often sought after when self-service solutions have not proven to be efficient. However, human service, although much more expensive, is not more effective in resolving demands than most chatbots currently used.

According to Faccio, consumer resistance is directly linked to the degree of humanization of interaction: 

The lack of resolution of demands for service provided by AI tools can have as a direct consequence the increase in consumer complaints, whether extrajudicial - through PROCON or platforms such as Reclame Aqui or Consumidor.Gov, or judicial.

However, considering the strong protective guidance of the Judiciary, this consumer liability is usually already accounted for by the company from the beginning as a rule.

The main point of attention directly related to the use of chatbots is the so-called "service loop", which occurs when the consumer is stuck in the electronic menu, without resolution of their demand and without being given the opportunity to contact a human attendant.

In these cases, jurisprudence has understood that there is a failure in the provision of the service, with the application of the theory of productive deviation, according to which the unnecessary loss of useful time imposed by the supplier for the recognition of the consumer's right is considered abusive and generates moral damage.

The solution, in this case, is to provide the consumer with at least one direct service channel by a human attendant, in addition to including, in the electronic menu, options that indicate sensitive or more complex situations, enabling the early identification of demand that are not adequately solvable by automated tools.

In summary, the use of Artificial Intelligence tools cannot create disproportionate barriers to human service.

In Faccio's experience, the use of artificial intelligence has been scaled mainly at the first level of care. As the complexity of the demands increases, there is the entry of human service. At the same time, AI itself is used to analyze and filter the most common types of problems and interactions, allowing you to direct continuous improvement actions in service.

In this context, the adoption of Artificial Intelligence tools in Customer Service must be accompanied by a clear implementation and monitoring strategy. The simple replacement of human service by automated systems, without adequate parameterization and supervision, can generate a reduction in the quality of service and an increase in the number of complaints.

For this reason, companies that use chatbots and virtual assistants in the SAC should pay attention to some relevant issues:

In addition, the efficiency of these tools depends directly on the quality of the knowledge bases used for training the systems and the constant updating of service flows. As new demands arise or products and services are modified, it becomes necessary to adjust the commands and parameters that guide automated responses. This process of continuous improvement allows you to gradually increase the resolution rate of automated service and reduce situations of consumer frustration.

Another relevant aspect is the adoption of mechanisms for intelligent scheduling of demands. More advanced systems are able to identify, from the analysis of the content of the request or the history of consumer interactions, situations that require individualized treatment or a higher degree of analysis, automatically directing these cases to human service. This type of automated triage allows Artificial Intelligence to act as an initial filter of demands, reserving human service for more complex or sensitive situations.

Another relevant point that should be considered by companies when adopting chatbots is their responsibility to the consumer. In light of the Consumer Protection Code (CDC), the supplier is objectively responsible for the quality and reliability of the service offered, regardless of fault. Thus, any errors, incorrect answers or the so-called "hallucinations" of the AI - situations in which the system generates plausible, but false or inaccurate information - can characterize failure in the provision of the service, especially when they induce the consumer to error or cause damage.

In this context, the chatbot must be understood as a tool used by the supplier in the consumer relationship and never transfers the technological risk to the user. Consequently, the company remains responsible for ensuring that the information provided is clear, adequate and correct, as required by the duty to information provided in the CDC. 

In summary, the evolution of technology and, consequently, the accuracy of the tools will certainly play a relevant role in changing consumers' perspectives on the use of chatbots in customer service.

In addition, the strategic definition of prompts and the constant updating of these commands, according to the demands effectively presented by consumers, can significantly increase the resolution rate of AI tools, reducing friction in service and mitigating regulatory and reputational risks for companies. 

In this scenario, the adoption of hybrid models, which combine intelligent automation with easy scaling for human service, tends to consolidate itself as the practice most compatible with CDC guidelines and market expectations.