On Wednesday evening, conversations between members of the general public and Amazon’s so-called Fulfillment Center Ambassadors began to circulate on Twitter. One particularly bleak tweet from the account @AmazonFCHannah got the most attention, displaying all the hallmarks of someone afflicted with corporate Stockholm syndrome:
The ambassadors are individuals who work in or have experience working in Amazon fulfillment centers. The individuals helm the accounts, which are maintained by Amazon, on a rotating basis for four months each. They were launched in 2018, and were previously most notable for their hard anti-union stance.
But Wednesday’s exchanges, which went viral on Twitter for their sheer oddity and invited speculation that the ambassadors are actually bots, showed another purpose of these accounts: to provide insight into how good the conditions are in Amazon fulfillment centers. Did I say good? I meant GREAT. Things are fine there! There are pirate hats! Sometimes they’re not worried about being fired! At other times they are not worried about being fired! Here’s a coworker! Whoops they get written warnings! Things are fine!
For a nation swamped in conspiracy theories, the mystery of whether these are flesh-and-full-bladdered humans typing somewhere in a high-vis vest as a conveyor belt full of slogan T-shirts and bulk lube rattles beside them is tantalizing. A little poking around invites a lot of seeming evidence to the contrary. The ambassadors seem aware that they have to prove themselves to be human and provide plenty of photos unique to these accounts—but when people began scrolling through the accounts’ old tweets, they noted instances where the purported employees call themselves by another name or mention family details wildly incongruent with a specific account’s photo. (Amazon maintains tweets from past ambassadors while changing the account name, leading to the confusion.) And this being Twitter, parody accounts joined in and muddied the waters before they were ultimately nuked.
Are these real people? (Yes.) If they are (they are), why are they so deeply unnerving? Why do they set off our uncanniness alarms? Maybe we really are hearing the dry, phlegmy rattle of chatbot code, but I think not. Yes, the diction in the tweets is strange sometimes, but no more so than, say, Jeff Bezos’s sexts. The more interesting question is not whether these are people made of meat and bad debt like the rest of us (they really are), but why it’s so difficult to imagine that they might be.
Amazon has spent years feeding us a steady diet of impractical tech—not impossible to technically execute, but not easy or worthwhile either. Think of the flying delivery drones. Think of the robotic sidewalk pack mules. Think of the glassy garden orb in downtown Seattle. Think of the rockets and the quantum computing and Jeff Bezos’s new biceps. They’re all very shiny (or lightly furry in the latter instance), but none of these have arrived on your stoop yet. Tech like this is just wild enough to capture the imagination—and just plausible enough to obscure the reality of Amazon.
Amazon isn’t a drone company. Amazon isn’t a sidewalk robot company. Amazon is a company that hires other companies to hire people to move a box, with their hands, into and out of a truck, over and over and over again. Amazon’s task is logistics. Its ethos is customer service. Its proven method is pointing a fire hose of human labor at any problem, as cheaply as possible.
Using shipping floor staff to do PR work is in line with that mission. What it lacks in polish it makes up for in “authenticity” and, crucially, its cheap, precarious labor. The same method is behind content moderation at Facebook and image detection for Ring’s home surveillance tools (newly acquired by Amazon, naturally). What presents itself as an automated process is in fact just very many humans, working together. In Ring’s case, image detection tools aren’t robust enough yet to supplant humans at those tasks. Amazon faces the same issue in its fulfillment centers generally. Humans are just better—and more inexpensive—for certain work.
The Amazon Twitter accounts sound strange because they’re shills and it’s hard to say anything worthwhile through a rictus grin. But also, they’re not really using Twitter. Their access to Twitter is mediated—and overseen—through Sprinklr, a tool that lets brands treat social media as a customer service problem. When they swarm responses to an @Amazonnews tweet, the ambassadors aren’t holding a conversation—they’re working their way through tickets, like at the DMV.
Usually that works fine, and some ambassadors keep up good-spirited conversations for long threads, but on days where the number of tickets spike they step on each other’s toes a bit. The fact that many of them take turns replying to one person in one thread would be really odd behavior for a bot if it were trying to pass as human, but makes sense if their task is just to answer questions, one at a time, from anyone engaging with Amazon’s latest publicity tweet.
The skepticism about these ambassadors has something more to it: It’s hard to believe that the second-most valuable company in the world might take to Twitter to defend its image and ... not do a good job. Why can’t Bezos buy a slicker army of bootlickers? If Steak-umm—a brand I refuse to look up but which I assume involves some kind of meat—can tell a good joke on Twitter, how can Amazon’s Fulfillment Center Ambassador program struggle to be effective?
This is where the conspiracy theories are unhelpful. The uncanny facade, the sheen of surreality, benefit a company that is really built on cardboard and human labor. Suspicions of a bot army hide Amazon’s actual army of poor precarious souls. The less comforting truth is that when a company has a trillion dollars and commands the lives of thousands of people, it doesn’t have to tell jokes.