MHI Solutions Improving Supply Chain Performance
Organizations can gain more value from these collaborations when they understand the capabilities and limitations of the technology and of the people who work with it.
It’s been a milestone year for artificial intelligence (AI), with generative AI technology like ChatGPT, Google’s Bard and Bing Chat capturing the public’s attention and generating a great deal of discussion about their potential for rendering humans in the workplace obsolete. In the supply chain, at least, that’s not going to happen. Although AI, automation and similar technology have become better and better at handling repetitive tasks, such as picking goods in warehouses or generating monthly reports, there will continue to be many activities and responsibilities that require the critical and creative thinking skills that only a human can contribute.
Balancing the human/tech interactions will be an ongoing process. As workplace technologies evolve, people should gain even more opportunities to perform the meaningful and important work that will contribute to their organization’s success.
Human/Tech Collaborations Add Value
To gain the most value from the human/tech collaborations in the workplace, organizations must fully understand what the technology they are investing in is capable of doing—and what it cannot do.
“As a group, we tend to jump to technology, and we think it’s going to fix everything. I think that’s such a fallacy,” said Christine Barnhart, vice president, product marketing and go-to-market at Nulogy. “Just making an investment in a new tool without really considering how that tool interacts both with your people and adjacent processes is very shortsighted, and often yields less than desirable results. You end up with a really expensive tool, a really poor process, unhappy people and a very, very limited return on investment.”
Companies must realize that AI and similar technologies are intended to help people understand the broader context of a situation. They are not good at factoring in the nuances of a particular situation or considering the emotional framework of a decision-making process that may be critical to a successful business outcome.
“While AI can help speed up and optimize decision making, there are watchouts. Consider, for example, a company that invests in a program designed to collect its data and use that information to recommend a solution that optimizes profits. It has a choice between serving two of its customers, Customer A and Customer B. Customer A is a strategic customer at this time but is paying a penny less. The system is going to prioritize Customer B; it misses the point that the company has some other objectives as well, and those objectives change from time to time,” said Ketan Shah, a partner at McKinsey & Company who leads the firm’s North American supply chain practice. When a human manager looks at that analysis, however, she would know that it’s important to serve Customer A.
“So, do you continue to chase that last penny profit? Or do you say, ‘We also need to serve Customer A, because it is a strategic customer,’” Shah added. Unlike an AI program, the humans looking at the data can understand the tradeoffs that can change over time.
“We frequently express it as technology and humans working together, not technology versus humans,” said Scott McArthur, a consultant at McKinsey. “For example, technology can use an optimization model to show a set of scenarios, and then the human, who knows who Customer A and Customer B are, can make the final decision. But they can do it much quicker than if they were doing it by themselves.”
People also bring another skill that AI may not have; they are more capable of understanding data in context. That enables them to determine if it would actually be useful for solving a particular problem. “If we looked at demand data or even consumption data during COVID and we used only that right now, we probably wouldn’t make great decisions, because there was this outside influence that was impacting so many things around our day-to-day lives,” said Barnhart. “Humans are really good at saying, ‘Okay, there’s an issue with this data; it might be valuable, but I might need to utilize it with care.”
Organizations also need to be able to trust the data, and that has proven to be a problem with generative AI. While it does a good job at bringing together a lot of different viewpoints and data, it isn’t very good at checking the gathered data to verify it.
In the warehouse, robotic and AI technology can gather data and identify problem areas, but it is up to the humans working there to determine the best solutions and implement them. A robotic picking system that regularly scans warehouse shelves can send an alert when it finds goods are in the wrong area; that’s a big benefit. “Normally that’s like looking for a needle in a haystack, finding where something has gone,” said Jackie Bibby, director of marketing at MHI member Dexory. But with the information the robot has provided, warehouse managers can resolve the problem more quickly, getting the goods to the right area.
Bibby said that robots often find a number of issues that need correction when they are initially deployed in a warehouse. People use this data to improve warehouse efficiency, developing solutions that robots can’t implement without human input. Collected data can also enable human managers to identify trends and make forecasts, enabling warehouse managers to proactively alert customers when warehouse space is going to become available.
Technology Choice Is Key
Most companies understand that tech is not the whole solution and that they will continue to need people who can work with it. “Organizations need to address how people will embrace technology within their roles and must re-wire existing processes to align with the new technology and evolving business objectives,” said McArthur.
Some executives make the mistake of deciding that they want some technology and then just throw money at it, without thinking it through, said Barnhart. That never works. “The humans have to assess the environment, what’s happening in the market and then strategically make decisions about what problems are the most pressing. From there they can determine the appropriate technology. Is it traditional AI? Is it generative AI? Is it a multi-enterprise business network or blockchain? There are a bevy of tools in the box.”
A company’s success in managing the human/technology equation depends to a great extent on the technology itself. Did designers think about humans when they were designing it? Will the organization’s workers be receptive to the technology so that the company can get the maximum value from their combined contributions?
“You don’t want to deploy flashy tech just because it’s cool. You want to deploy a solution that’s actually going to improve the experience for your associate,” said Joseph Ruck, VP, marketing and communications at MHI member Ambi Robotics. The company has built its brand around connecting people with robots and providing human-centric solutions.
Humans will be more receptive to using AI and technology if they have had some say in the technology choice and if they understand in advance the benefits it will provide to them.
“I’m a big believer in education driving adoption, so our adoption strategy begins before the robots get on site,” Ruck said. Ambi Robotics’ training not only focuses on how to use the technology, but also explains the value that it will bring to the operation and to the workers themselves. Providing people a way to offer feedback on the technology—and ensuring that the feedback is actually listened to and acted upon—helps employees feel involved and part of the process.
Sometimes that means encouraging people to see for themselves that the data the technology provides is accurate. If supply managers are accustomed to compiling reports from numbers they pull from a spreadsheet, their employers can encourage them to compare those numbers with what the new AI software is providing. As their experience with the technology grows, they may discover that it actually does a more accurate job of compiling the information than they were able to.
Click here to read the full feature.