Will automation make retail teams less efficient?
Knowledge@Wharton staff
Presented here for discussion is an excerpt of a current article published with permission from Knowledge@Wharton, the online research and business analysis journal of the Wharton School of the University of Pennsylvania.
While the arrival of AI-supported automation in the workplace has led to concern about the potential loss of jobs, a new paper by a Wharton marketing professor explores the fate of those who keep their jobs but count robots instead of humans among their co-workers.
In “Men and Machine: When Should a Firm Adopt Automation?,” Wharton’s Pinar Yildirim and co-author Mustafa Dogan, a postdoctoral researcher at Carnegie Mellon University, show that in some cases, automation can actually lead to increased costs and inefficiencies because of its dysfunctional impact on human teams.
The professors explored a scenario in which a manager oversees a task completed by two humans and one in which it is carried out by one human and one machine. Automation was found to be the lesser of the two approaches if it makes the remaining human workers “more expensive” or requires changes in the way human workers are assessed and compensated in order to create gains.
Workers who aren’t displaced by automation may react with feelings of powerlessness or fear, Ms. Yildirim notes, or they could become less motivated because they are unable to perform tasks as flawlessly as a machine. “When a machine is working next to you, the incentive for you to exert effort changes,” she points out.
For managers, pairing a human with a machine will likely reduce their ability to understand whether any underperformance is due to a lack of effort or external factors (i.e., recession). She adds, “And knowing that the monitoring ability of a manager is reduced also increases the incentive the employee has to shirk because he or she knows it will be less likely to be detected.”
Companies that do rely on a lot of team interaction, the research shows, can mitigate the human costs by setting up compensation structures that are more competitive rather than cooperative. And yet the downsides to humans working in more competitive environments would have to be explored.
Said Ms. Yildirim, “People need to realize that organizations will have to be redesigned around automation. We have to think about what’s happening to the individuals who keep their jobs — their lives are changing, too. Nobody is talking about that.”
Discussion Questions
DISCUSSION QUESTIONS: Do you see increased automation and AI having an adverse effect on the performance of human associates on selling floors? How might management need to change the way it motivates, assesses and compensates employees? What potential side effects should retailers be most concerned about?
Poll
BrainTrust
Ken Lonyai
Consultant, Strategist, Tech Innovator, UX Evangelist
Ken Cassar
Principal, Cassarco Strategy & Analytic Consultants
As noted, the potential negative impacts on human associates of deploying automation is hardly understood. While most discussions about automation and AI focus on the potential savings and productivity gains, it’s helpful and mildly disturbing to see a study that attempts to understand the unintended consequences these systems may have on human workers. Given the stage of development that automation is at in retailing, I believe human associates will be required for some time to come and therefore the potential negative impacts described by the researchers is valid.
Retailers who choose to deploy automation and AI in their stores need to proceed with caution. Using automation to reduce/eliminate repetitive, mindless tasks or to facilitate smoother transaction processing and to free-up associates’ time to engage with shoppers and deliver better in-store service makes complete sense.
Automation is already heavily used behind the scene in online retailing. The question is if automation will help in brick and mortar interactions where the primary value proposition is that you get to interact with the product and a human being to answer questions. Using a food service example, if I buy a bowl of ramen from a vending machine, I am not looking for human interaction (yes, they have one now near where I live; need to try it…), but if I go to a Japanese restaurant, part of the experience is to interact with the staff. Blindly adding automation to brick and mortar interactions to work with humans without being under the entire CX would be very negative for the employees and the customer.
Information has always been the business enabler, and as data and physical automation increasingly solve challenges, the role of the worker changes. Those who embrace this enabling effect, while still understanding the logic that underpins its application, will be the talent moving forward. Automation is intended to contribute to efficiencies, so such workers will also be the high performers, enjoying the recognition that comes with it.
I agree with the potential, Lyle. But I’d stop far short of “always been.” Information and tech have too often slowed and harmed processes. They are neutral factors — whether they are applied well or poorly delivers the end result of improvement, no change, or harm.
Fully agree with you, Doug! It’s not the technology that is inherently contributing to efficiency or productivity — it’s the application of the technology by the people controlling it!
This has already happened in at least one industry — automotive. It sounds like (due to lack of reference to it) we may not have learned anything from shifting to a largely automated working environment. Anybody familiar enough with this business?
Two huge differences between the auto and retail sectors: unions and the nature of the work. Carmakers would undoubtedly be utilizing much more automation now if the unions hadn’t fought for retaining jobs. In some areas like welding and main structure assembly, it’s a 100 percent robotic world because it’s both a hazardous and precisely repeatable work. Except for the loss of some colleagues, I don’t believe any autoworker cares that robots do that work.
Retail/warehouse has neither the equivalent unions nor precisely repeatable functions, save for maybe some stock work. Retail can use “automation” for knowledge related functions like interacting with consumers whereas there’s little need for that in manufacturing. Apples to oranges.
There’s no logic in humans and machines doing similar work. Each has strengths and weaknesses which have been expressed in RetailWire before. So sure, if automation is a near direct replacement for some human tasks, staff will likely feel uneasy, angry, disassociated, etc.
Obviously, any new technology requires rethinking roles and responsibilities that it affects. Some forms of automation will undoubtedly eliminate jobs (it is happening) so the remaining positions will need adaptation to make them valuable to the employer and (as much as possible) satisfying to the employees.
That’s what I was trying to figure out. Why would you have workers doing the same work alongside robots, ever? Like, in a retail context, are we talking about baristas working alongside a robotic barista? If the business case is worth it enough to buy the robot in the first place, why would you keep human baristas around at all?
Robots, like any technology should enhance human efforts. Whether that is a robot barista who makes the drinks while the human takes the orders and maybe steps in to supplement when demand is higher than the robot can produce, or the human who can now go around to customers at tables and ask them if they need anything else or chat about how their day is going and establish better relationships with the regulars, or whatever, it’s about freeing up human time for the complex, higher-value, non-repeatable tasks.
For heaven’s sake, would you buy price optimization software and then ask the human to keep on doing it by hand? No!
One aspect of this discussion is that humans are social beings. We enjoy interacting with other humans. As automation reduces the number of humans with whom another human can interact, there will be more negative impact on the associates that remain on the job.
Given the general state and perception of sales associates on retail selling floors, the increased use of automation and AI for supporting the sales process should be celebrated and accelerated. Any technologies that can empower sales associates can only help meet the expectations of the digitally-empowered shopper that has already started her shopping journey using online (ZMOT) automation and AI long before she enters your store.
The closing line in this article is the key point. Organizations need to be reorganized around automation. When the chainsaw came around, logging companies needed to reallocate resources from cutting trees down to stripping bark and planing boards. Costs of lumber dropped and more people could afford to build homes, which expanded the industry and jobs in the industry. Automation is scary and requires adaptation across the organization, but I don’t know of a historical instance where automation didn’t lead to greater opportunity at the macro level.
AI by all accounts is in its infancy. This week Paul Allen invested another $125 million into his AI2 Labs to teach machines “common sense.” AI still lacks what most 10-year-olds possess: ordinary common sense. There are so many variables — human variables, human outcomes — only touched upon by the researchers in this article. At this point, the supposition that AI can be anything, do anything, in any use case is merely uninformed conjecture. AI is a tool, not a panacea. A tool of business for which leaders/retailers will need to invest the time to understand it.
This is not tech driven by technologists. The human aspects of AI will require business leaders to get their hands dirty to ensure the proprietary programs each retailer deploys using AI balances both human and business needs. AI is not a one-size-fits-all. Nor are we really using “AI” yet. Buzz …
Looking specifically at the human performance metrics, there is no intrinsic reason why productivity should decline, simply due to increased labor automation. Human task assignments should be even more focused and increasingly dependent upon humans to execute, since there will continue to be a need for human touch. Those retailers that completely eliminate the human touch will experience the decline in productivity, in my view because there are indications that humans still drive a lot of the components that drive brand loyalty. That capability is not easily replaced by automation. And that will not drive increased revenue, so overall productivity may not rise.
No I do not. If AI is implemented correctly and with consideration for the eco system it’s to be used in, it can be a positive but if deployed badly then as with any automation system it could harm performance. Retail will evolve as a result of AI or any other automation, we can see this in history with examples in office automation and manufacturing automation. Jobs that existed then no longer exist today; retail will surely have the same evolution with new jobs being created to train and curate AI inference engines, for example.
I found an old article for the manufacturing sector: “In our view, the oncoming transition will probably be less dramatic than the impact of office automation over the same period.” This describes in many ways the same cycle and the original fears for manufacturing with lessons that can be rolled into retail 40 years on!
This is interesting, but seems to be a narrow evaluation by design (for success in doing the research).
In the broader picture, seeing AI making things less efficient would be no surprise. Remember the 1980s paperless theory? In the category of unintended consequences, paperless efforts have driven far more consumption of paper.
Technology also releases the ability of paper clip counting bureaucrats to micro-manage to within an operation (do we need to track paper clips in Excel?) rather than simplify and improve. It allows things to be tracked that offer no business advantage other than political satisfaction of saying “yes, we track that.”
Technology often has the opposite effect of what was intended and that very well might be AI’s end result — to increase cost and time to complete work.
Two points come to mind from reading this article.
First, it’s worth noting that almost every discussion around AI and robotics to date centers on a black and white world in which tasks and processes are either completed by a machine or by a human. While this may work for the simplest and most repetitive of tasks, our world really isn’t that black and white. So why aren’t we considering tasks and processes in these analyses that reflect humans and robots working together towards a common goal? At least this research attempts to do so.
Second, if you are going to attempt to evaluate this gray area, then you need to use tasks and processes that make sense to be evaluated this way. From what I can see, the research still looked at relatively simple tasks for the comparisons and I don’t feel this produces a relevant conclusion. There is a future where AI and robots are a part of a process involving human thinking and control that leads to more productive and efficient results. I’m not sure AI is at this level just yet, but we will get there.
Narrowly, the research is not very relevant to the retail (or any) space due to the basic flaw that an organization would continue to utilize a human for a task that has been determined to be better done by a robot or other form of machine learning or AI. However, the valid discussion is in thinking about how an organization thinks about how the humans will interact with, access, and leverage these evolving tools (or not). If thrown in haphazardly, I would predict lowering productivity and reduce results. If thoughtfully deployed, with full participation of the teams who will be impacted, it will be value-additive and drive higher levels of job satisfaction, engagement and results.
In the retail space, I have already seen my former company (DFS Group) carefully assess how merchants, for example, would utilize machine learning and big data analytics for work they used to do manually. By including the users in the process, the implementation was not flawless, but much more intentional and effective. This same effort must be applied by retailers when contemplating the integration of all high-level AI enabled tools — in the warehouse, in the merchandising arena, and on the sales floor.
A: Who’s the new guy?
B: Watson. I hear that these robots just showed up today in all of our stores.
Watson: “I will be leaving for a few hours to go and find executives who thoughtlessly dropped all of us in the stores, if you have any stock in this retailer, I suggest selling it now.”
A: Cool, I like this guy.
(One year later at an employee training offsite…)
A: Hey, how are you!
B: Excellent, what classes are you taking?
A: “Robot collaboration for managing merchandise returns” how about you?”
B: “Agent, AGENT, AGENT!” It’s a conflict resolution class for customer/robot issues.
(One year later…)
A: Wow, glad to see you still have a job, how did you pull that off?
B: I decided to go to Robot programming school, check this out: “Hey Watson, I’ve been waiting here for help, what’s going on?
Watson: “Playing, “What’s Going On” by Marvin Gaye.
A: Now that’s cool!
The moral of the story: I will be the first to be replaced!
Video of robot collaboration from the test store – here.