Amazon may have learnt a terrifying lesson from its ambassadors project

"I feel like I'm talking to the Borg," wrote exasperated Twitter user Diana Wilde, after falling down a rabbit hole that went viral.

Wilde, owner of the account @rulesObeyer, asked @AmazonNews why they didn't treat their workers better, only to find a flurry of "Amazon FC Ambassador" accounts (Rafael, Brittany, Dylan, Audra, Cindi, Rachel, Carol) lining up to tell her that working in Amazon's "fulfilment centre" is fine, absolutely fine. The hive-mind tag-teaming couldn't help but come over a little creepy-cyborg.

“Are you a robot, did they make a bunch of AI to lie for them instead of helping real workers,” she wondered, sounding equally besieged and bemused. “Rafael - Amazon FC Ambassador” replied that, no, he was genuinely an Amazon FC (warehouse) picker, one of the human ones. “That would be a crazy technology to artificialise thoughts,” he insisted through his Amazon-controlled account.

It was all a bit bleep, affirmative, bleep, system malfunctioning, bleep. Resistance did seem to be futile. But Amazon’s ambassadors are not actually machines. Endearingly, one account, “Brittany - Amazon FC Ambassador”, made the kind of rhetorical error you imagine would be beneath a robot, revealing that her job was to make sure “the good experiences are being heard and not just the bad”.

Tennis and Tokyo: the dream combination

---

Our face-down culture won’t last forever – not if advertisers can help it

---

Donald White House’s genius vision of a world without surnames

---

‘Sportswashing’ is the grim game of our times

---

This allowed Wilde to skilfully ask, “what if, instead, they just fixed the bad experiences?” Alas, Brittany was out of this particular conversation and Rafael was back picking up the slack.

The ambassadors’ existence, though spotlighted by the thread, aren’t new. For a year now, Amazon has had a marketing scheme in operation in which it pays its warehouse employees to go on Twitter and write positive things about working for the company, or “share facts based on personal experience”, as the ecommerce giant puts it in a commonly issued official line.

The trouble here – one of them – is that it’s not such a “crazy technology” to artificialise thoughts. Some social media platforms appear born for bots. On Twitter, humans engage in pointless debates with bots, and that’s only the start of the weirdness: bots have also been known to “argue” with each other in auto-responding loops.

Fluffy messages

The obvious potential for a corporate brand to create a vast program of quasi-real, bot-generated messages that all revolve around its essential fluffiness is still largely untapped – as far as anyone knows, anyway.

Because when it happens, we'll all be able to recognise it for what it is, right? Er, maybe. A parallel discussion about the uncanny phenomenon of comic actor Bill Hader "morphing" into Tom Cruise while doing a chat-show impression of him was last week sparked entirely by the fact that this "deepfake" was just so subtle.

There is no Hader 2020 campaign for the White House to fear, and the harmless clip was labelled for what it was. But the sheer skill involved was enough to worry some. For more sinister applications, we only have to go back three months to the Facebook-disseminated video of House speaker Nancy Pelosi doctored by her political enemies to give the wrong impression she was a slurring drunk.

Amazon’s ambassadors are, to be fair, also clearly marked as such. There has been no attempt to disguise their purpose. And luckily their sheer efficiency in responding to anyone on the internet who might dare to have a low opinion of working conditions at Amazon’s “fulfilment centres” is an ironic form of corporate clumsiness in itself.

The Amazon ambassadors were easily outfoxed, but what really undid them was that they were trying too hard. They’re not to blame. It is an endlessly reassuring fact that big brands often fail at small subtleties. In trying to exercise too much power and control, they draw attention to their efforts.

The scheme appears to have been devised as a crisis-PR response to high-profile criticism of its employment practices, but would any crisis-PR specialist worth their fee recommend they keep at it now?

On the one non-robotic hand, it looks absolutely terrible to have to pay your staff to say nice stuff about their employer, especially if the game is to drown out less flattering takes. On the other, a general propagandist vibe about one’s place of work is far from uncommon across social media. The waters of suspicious enthusiasm are already muddy.

All an “ambassador” has to do is muddy them some more. So what if an undercover journalist found that some Amazon workers were urinating in bottles because they feared being sanctioned for taking a toilet break? Here’s a reasonable-sounding Amazon worker to say, hey, they don’t know anything about that, that’s never happened to her.

Smoothing cracks

Petrifyingly, the lesson Amazon may take from its tweet-force’s interactions with Wilde is not that it shouldn’t be bothering with the ambassadorial scheme, but that it needs to hone it, and smooth out its human cracks.

Public relations professionals who are bearish about employment in their own industry sometimes cite the rise of AI-led marketing as a threat to their own jobs. We could indeed be on the cusp of a new and tedious wave of AI-generated brand messages.

But right now it seems more likely that what we’re in for is the industrial-scale “hey, I know I’m paid to say this, but... no, really” style of marketing perfected by influencers and podcast hosts and amateur ambassadors, where we know we’re being sold to, but still can’t help falling for it.