AI is only as effective as your human-led processes
By Pamela J. Gallagher
Last year, I wrote an article lamenting the impact of interacting with robotic, pre-recorded “representatives” on our ability to communicate interpersonally. Today, the human representative is scarcer as AI’s capabilities and usage continues to expand. Unlike the previous use of robots with pre-recorded lines “listening” for keywords to prompt certain scripted responses or relying on the caller to “press 1” for certain information, now AI bots-as-representatives are imitating more critical thinking functions.
I wonder whether the increase in AI-dependent customer service is growing because the technology is becoming less frustrating for customers to use, or whether we’re just getting used to the reality that if we have a question, we’ll have to be content with a robot being the first to try to answer it.
AI technology in the healthcare industry is becoming more sophisticated by the day, able to perform an increasing number of complex functions. But AI or any other technology will not work well without solid processes in place to ensure that it is serving the mission and the patient. For an AI technology to add value, organizations need:
a clear picture of the problem you are trying to solve and
a process that pairs AI capabilities with a clear structure of human responsibility and accountability for that technology.
Pinpoint the problem
Are you looking for ways to cut costs? Decrease human error? Automate basic job functions to give overwhelmed employees their time back? The technologies an organization chooses to utilize among the endless options available and the ways they deploy those technologies should be rooted in a clear understanding of the problem you are trying to solve. You need more than gas in the tank and a driver in the front seat to realize the potential of a car; you also need to know where you want to go. Without a clear direction, AI cannot add value and often serves to make the problem even worse.
Develop human and AI processes in tandem
Often, organizations either create an organizational process, then figure out where in that process they can plug in AI, or they adopt a particular AI technology and work backward to design a process to make use of it. Both of these approaches are ultimately unhelpful. It is most effective to develop human and AI processes together. This ensures the most effective use of technology and clarity on which employees are ultimately responsible for stewarding those technologies.
For example, AI representatives with the most impressive critical “thinking” functions are useless if the organization employing the technology does not have complete clarity on which departments various requests should be funneled to and exactly which human beings are responsible for overseeing, executing, and following up on that AI-performed function.
A few questions organizations should ask themselves as they are building a human-led and technologically-robust process are:
Where in this process do we still need humans in the loop?
Where are judgment calls or decision making required to serve the customer well?
What are low-risk functions for AI to perform?
What tasks are repetitive?
What tasks do our employees say is a drain on their time and energy?
What are the routine questions we receive from patients?
Strategic implementation of AI technology can make room in a day for humans to do the things that only humans can do. But for a technology to work well and not exacerbate the problem it is supposed to solve, organizations need to create strong processes with clear human accountability structures. Organizations that do this well are more likely to utilize AI in a way that adds value for patients and employees and supports the mission of the organization.