Creating an employee-centric culture of innovation needs to come before data collection, algorithm training or use case research when developing an enterprise AI strategy. Not only can employees offer important insights into business processes and what can be automated, they are also the primary users of the technology. Integrating their feedback into AI strategies from the research stage increases the chance of AI implementation success.
Several experts at the Intelligent Automation Week conference in New Orleans spoke on the importance of recruiting employees into your AI strategy from the start. They have the day-to-day insights necessary to create a comprehensive strategy. Relying on one or two data scientists to identify processes, choose what to automate and provide feedback is becoming increasingly difficult as the gap between available data scientists and desire to automate continues to widen.
When deciding what to automate, companies often look at lagging business processes or holes in production times or outsource to a third-party vendor for suggestions. However, Jason Kammer, director of intelligent automation at Fifth Third Bank said that talking to employees should be your first step in business process automation.
Employees have first-hand knowledge of what needs to be automated or where automation can easily fit into an existing process. AI should enable your employees to do the highest-value tasks they are capable of -- and accelerating your enterprise AI strategy should start with employee discussion.
Fifth Third Bank chose to automate customer support in its equity line division and mortgage liens. Cloud-based servers process, store and evaluate data immediately. Employees at the bank are then free to perform higher-level or personalized customer support. This automation resulted from employee feedback, alongside the classic process automation analysis and use case research.
The maturity of your team and employee feedback should guide your AI strategy, Kammer said. New hires or citizen data scientists on your team may have less experience with internal processes, but more experience around issues in current processes, ideas for automation and familiarity with new technologies.
AI projects are often framed by financial metrics like how much money can we save. But Kammer suggested assigning value to automation that goes beyond strict monetary value.
"If you have 10 AI processors that can free up an hour of time each week, that's a straightforward calculation. What [you] are doing with that time, though, is creating a positive internal outlook," Kammer said.
Putting the emphasis on employee success rather than AI process success will also help you allay worries of AI destroying jobs, said Dana Zeller, senior vice president and head of banking operations at Bank Leumi.
Employee buy-in leads to AI success
Integrating employees in every step of the AI and automation process allows for employees to be exposed to the technology earlier and can reduce the friction of AI being inserted into their workflows.
"There's somewhat of a resistance until they see [AI] in action and they realize this is a tool for them to do their job better," said Katrina White, senior manager of business process automation at Boeing.
Raj PolankiHead of analytics and data science, Wacker Chemie AG
Alongside initial employee buy-in, Jason Conde, director of intelligent automation at Houghton Mifflin Harcourt, said that having an environment of innovation and exploration is key to a successful enterprise AI strategy.
"Our bosses gave us the space and support to explore technologies. They stopped training us like computers and let the computers do some work," Conde said.
While the C-suite is an important part of creating an AI strategy -- as it controls funding and high-level decision making -- AI is truly a cross company effort that requires employee support. Conde advised having conversations with employees that work directly with the processes that are being automated.
"Go face to face. Don't do a Webex [video conference] with 2,000 employees, because you aren't going to get the right feedback from them."
The feedback that follows can range from recommending parts of day-to-day work that are ripe for automation, potential issues with implementation or thoughts on the integration of tools.
The data scientist problem
The most typical approach to creating an AI infrastructure involves defining your AI strategy, hiring data scientists and creating models from there. This leaves your employees to come in after the strategy has been built and implemented without their feedback.
But in order to get from insights to action, companies need to build AI competency without data scientists at the center, said Raj Polanki, head of analytics and data science at Wacker Chemie AG.
With a nationwide shortage of data scientists, building a strategy based on existing employees and their work processes makes companies less dependent on data scientists.
"Avoid the wish list approach to forming an AI strategy. You won't get the rock star data scientist," Polanki said.
In order to avoid making your enterprise AI strategy reliant on a data scientist that may be impossible to acquire, Polanki recommended taking a jump-start approach. This approach melds together existing employees and research to focus on creating a company culture to allow the easy integration of AI across company market silos.
AI strategies need to feature employees at the center to offer insights, feedback and ideas and help streamline implementation. While employee buy-in has historically been about automating processes and giving higher-level work, experts are encouraging enterprises to create a strategy that uses their existing teams to develop AI processes. Keeping employees in the loop and allowing them to help define processes and bring AI into the workplace reduces the friction of implementation and forgoes fears of new technology.