[caption id="attachment_138642" align="aligncenter" width="2178"] Amazon Go[/caption] Amazon has created a new-look convenience store that lets you buy goods without ever going through a checkout. Dubbed ‘Go,’ the store is almost entirely automated. What does that mean for humans, and jobs? Here’s how it works: the Amazon Go app has a QR code you must scan at subway-style turnstiles to get into the store. Once you’re inside, Amazon uses machine learning, computer vision and artificial intelligence (via sensors) to ‘watch’ what you’re purchasing. After you’ve grabbed all the things you need, you simply walk out. Amazon’s tagline for Go is “no lines, no checkouts, no registers.” That’s insinuating you’ll be able to avoid interaction with people, too. It’s worth noting that the space (in downtown Seattle, and currently only open to Amazon employees involved in a beta program) is roughly 1,800 square feet, so it’s not as if Amazon re-created a proper grocery store. But it has prepared meals, ready-to-eat items created by on-site chefs, various sundry and pantry items as well as Blue Apron-esque meal kits for two. Still in beta, Amazon is planning to open Go up to the public early next year. It all sounds sublime, but there’s a lot we don’t know. Is Amazon Go part of Prime? Are there expansion plans to other cities (we’d have to assume so, but who knows)? More critically, who is running the store day-to-day?

Does Amazon Go Put Jobs at Risk?

In its video, Amazon is careful to minimize the presence of actual human employees at Amazon Go, likely in order to appear more futuristic (robots do all the work; just shop!). But even with sensors and monitors watching our every move, things are bound to go wrong – things a machine can’t correct for. An example: Amazon Go uses sensors to know your in-store location and what items are being pulled from shelves. So if you’re near the bread aisle, and take a loaf of wheat from the shelf, Amazon Go assigns that bread to your in-app cart. Conceptually, it's brilliant. But if you’re there with a few others and everyone is picking through the bread, it’s possible you’d be held responsible for bread that you never grabbed. The opposite is also true: what if the various sensors and cameras can’t detect who has taken what (does it send an in-app notification asking if you took the item?), or a customer leaves something they decided not to buy on a random shelf elsewhere in the store? It’s possible there are secondary methods for tracking what you have, but the nature of Amazon Go lends itself to human behaviors like grabbing something off a shelf and putting it into a purse or bag, then walking out. While the barrier for shoplifting is low for Amazon Go (it’s linked to your Amazon account, and charges you automatically), the likelihood machines will make errors is higher. Although Amazon hides away employees, its Go concept is not wiping out jobs wholesale; we just have to evaluate roles differently. Amazon Go will still need a store manager, cooks, kitchen management, receiving staff, people to stock shelves, humans to verify inventory and likely a few bodies strewn about to help people find things, answer questions and verify purchases or lack thereof (as well as troubleshoot in-app issues, should they arise). To a smaller degree, Apple already does this. Apple Stores use the Apple Store app for many off-the-shelf purchases; you scan a barcode on the packaging and pay using Apple Pay. Save for high-dollar items like Apple TV, where the app asks you to ping an employee to continue checking out, you are free to move about the store and buy grab-and-go items without ever speaking to an employee (and there are plenty). Self-checkout at grocery stores is conceptually similar to the execution we should expect from Amazon Go. As customers check themselves out, humans manage queues and troubleshoot issues. As ‘convenience stores’ go, Amazon Go may be the best of the bunch when it’s all said and done. By design, technology should make our lives simpler and more efficient, and Go checks all of those boxes.