Amazon opened a convenience store on Monday in downtown Seattle. Known as Amazon Go, it enables shoppers to scan their smartphones with an app, grab what they want from the 1,800-square-foot mini-mart, and leave without paying a cashier. By combining sophisticated computer vision, machine learning algorithms, and sensors, the e-commerce giant tracks what customers have purchased and charges their Amazon accounts  for what they take out the door. Amazon didn’t say whether it will open more “Go” stores, but reports suggest that the company could sell the technology to other retailers.

We asked Bruce Clark, associate professor in the D’Amore-McKim School of Business, who studies marketing strategy and managerial decision-making, to examine the potential impact of automation on the future of commerce.

Grub Street, New York Magazine’s food and restaurant blog, called Amazon Go an “automated 7-Eleven killer.” How do you think convenience store chains should respond to Amazon’s initial push to join the $550 billion convenience store industry? Might they need to rethink their business models in an effort to compete in the age of artificial intelligence?

The short-term consequence to convenience stores is probably low. A lot will depend on local conditions. Convenience stores have very localized trading areas because the primary reason someone goes to a convenience store is—surprise—convenience. Unless you have an Amazon Go next door to your store, I’m not sure this is a big deal. A longer-term threat might be if Whole Foods went cashier-free and you had a Whole Foods next to your convenience store.

At a broader level, convenience stores are already looking at how to incorporate technology into their business models. Many have apps and loyalty programs. Self-checkout or online ordering is increasingly common in food retail of all kinds. Convenience stores have to keep up with technology, but it’s not clear to me the customer experience of an Amazon Go store is sufficient to make me walk past the local 7-Eleven. Geography is destiny in this sense: where you have more convenient stores, you will do well.

Studies show that retail salespeople and cashiers, who make up 6 percent of the U.S. workforce, are particularly vulnerable to automation. In your opinion, what role will artificial intelligence play in their ability to find stable, well-paid employment in the years to come?

The key point to remember is that AI doesn’t automate whole jobs, it automates pieces of them. AI therefore allows employers to change the nature of retail jobs, with unpredictable effects. The famous counterexample here is the relationship between ATMs and bank teller employment. Despite widespread adoption of ATMs, employment of bank tellers increased over much of the past three decades. The driving factor was that while a given bank branch might require fewer tellers, that reduction in labor costs meant that banks could open more branches, offsetting the loss at any given branch. Tellers’ jobs in turn became more like that of a customer service representative rather than a paper- and currency-pusher. More recently, as Starbucks has allowed mobile ordering for pickup it has found it sometimes needs more baristas at peak times to deal with increased demand.

The common thinking is that over time, human jobs will evolve toward processes at which humans remain better value, notably in emotional intelligence and physical dexterity. Retail employees will find their jobs increasingly specified in those terms. You’re either interacting with customers or efficiently managing the stocking and layout of a store. Employees who excel in these areas should be OK. All that said, I’m not sure any of us should be assuming we will have stable, well-paid employment in the future.

Retail chains like Wal-Mart and Lowe’s have turned to automation to boost sales and replace some human employees. Do you see process automation as an unequivocally good move for the commerce sector, or are there some hidden costs to the rise of AI in the workplace?

This depends on retailer strategy. If the plan is to be the low-cost retailer of a certain set of goods in one’s area, then maybe you automate everything and get rid of employees. The problem with that strategy is that only one retailer can be the low-cost producer—and it’s increasingly likely to be Amazon. If instead the plan is to provide a customer experience that is worth driving to the store for, automation is a double-edged sword. Why drive to a store to be served by a robot when you can simply say, “Alexa, re-order diapers”? A superior customer experience may mean less automation as one improves employee training and carefully manages a store’s physical space and product assortment.