The Gist

  • AI-Human collaboration. Humans and AI combine strengths for improved customer experience.
  • AI in customer service. AI augments tasks but human involvement is still crucial.
  • Organizational importance. Success with AI relies on organizational changes, not just technology.

The rapid evolution of artificial intelligence in recent months has many organizations looking to incorporate the technologies into a number of business processes, including customer service. 

The answer is generally an AI-human collaboration — known as collaborative intelligence — where the organization taps what each is best at. That means the social skills, creativity and leadership abilities of human agents, with the speed, quantitative capabilities and scalability of AI. By combining efforts, humans and AI applications can enhance the strengths of the other for a hoped-for improved customer experience.

“Customer service operations — as in other enterprise areas — seek to improve efficiencies, lower costs, and improve results,” explained Bern Elliot, distinguished vice president and analyst at research firm Gartner Inc. “The types of benefits from such collaborations seek to improve user experiences, insights, accuracy, timeliness, costs, and revenues.”

This Is Not Your Father’s Artificial Intelligence

While artificial intelligence has been around for decades, the technology is becoming much more robust, Elliot stressed. As a result, he expects it to be a critical part of advanced customer service operations at many organizations if they aren’t already.

“There is general awareness of the value of combining advanced AI methods into customer service operations with human activities, sometimes called AI augmentation of agent actions and tasks,” Elliot explained. 

As AI becomes more capable, it will enable the automation of a growing number of tasks, Elliot said. It can currently assist humans in many aspects of customer service and engagement, especially in the area of decision-making or by finding and providing information quickly to a customer seeking assistance. 

But because customer service is such a delicate and critical area to get right, there are many situations where a person’s direct involvement is still needed, and AI serves in a largely supporting role. 

“There will be many opportunities for AI solutions to focus on assisting, or augmenting, or collaborating with individuals to assist them in their tasks,” Elliot explained. 

Still, overall, Elliot said adoption varies greatly at this time.

“The maturity of AI usage overall in an enterprise is often a good predictor for the adoption of AI augmented agent capabilities,” Elliot said. 

Related Article: How Human Insight Enhances AI-Driven Marketing Personalization

How Organizations Are Using AI and LLMs in Customer Service

Elliot cites the following as some common examples of how organizations are incorporating large language models into customer service functions:

  • Using LLMs to generate alternative phrase suggestions for people writing responses.
  • Using LLMs to summarize conversation with a customer, and then the agent can review and accept the summary.
  • Using LLMs to collect and summarize information about previous interactions with a customer and prepare a summary for an agent before the agent gets on the call.

Organizations that have made significant strides in these areas have no doubt learned that the most significant factors to success are organizational, not technological, Elliot stressed.

“Yes, the technology has to be there. But that can be done in a POC or MVP, and then scaled as needed,” Elliot said. “However, in order to implement the AI-human collaboration, agent actions and processes must often change. Sometimes business process must also change, to properly leverage the new functions. Those changes can be the greater challenge.”

Related Article: Most Consumers: Don't Use AI to Personalize My Experience

Learning Opportunities

The Drawbacks and Shortcomings With AI Contributions

As impressive as AI can be with some tasks, it also has a number of shortcomings in the areas of accountability and trustworthiness. This is where the human part of the AI-human collaboration equation comes firmly into play, and it is important for organizations to understand this.

For example, professionals working with AI programs often want to know why AI provided a certain answer or recommendation to a task inquiry. Since AI cannot discern accurate data from inaccurate data, explainable AI methods can often give false or misleading explanations.

Such falsehoods or misleading outputs can quickly erode the trust of a human counterpart in the benefits of an AI-human collaboration and make the human staffer reluctant to accept or act on AI outputs.

Because of that, elements that are not good for AI automation include those requiring empathy, human judgment, where a dialogue with subtle cues will be helpful, or where there is a chance that the automation may provide erroneous or misleading information, Elliot said.

“Design of these systems is more craft than a science,” Elliot continued. “Yes, there is advice available on human factors design. But often there is a need for experimentation and feedback from actual deployments. And in some cases this can be an iterative process. Also, not all people have the same response to automation, so there may be a need to enable customization.” 

Organizations Need to Get All-in With AI to Distinguish Themselves

Investing in AI is obviously a big move for any organization, but smart ones know that there is no time like the present, said Tom Davenport, a leading technology consultant and co-author of the new book, “All in on AI.”

Organizations should not fear AI as a job killer, Davenport said, but instead embrace it as a means to improve processes, gain efficiencies and spark innovation. Specifically in the case of customer service, AI can be an invaluable tool at providing customers with prompt attention, valued information and appropriate resources.

Because time is always of the essence in any customer engagement situation, Davenport argues that those organizations that get an early start with AI technologies are most likely to gain competitive advantage and lead in their industries.

Many customer-focused activities and tasks are ripe for automation, Davenport explained. He sees AI as the game-changer that will distinguish leading organizations in virtually all business process areas.

“Companies that do a lot with AI, and are bullish at incorporating it into their processes, are quite successful,” Davenport explained. “The time to stand on the sidelines is over. This is going to be an area that it will be difficult to be a fast follower. And it will be hard to catch up if somebody else in your industry is doing this, and you’re not.”