Amazon employees automate busywork to pretend they’re using AI

First AI came for our jobs – and then it came for our busywork. 

That’s according to a report from the Financial Times, which cites multiple Amazon employees saying that in-house AI systems are being used by their colleagues for no good reason. Unless you count impressing their managers.

Amazon managers are reportedly tracking how employees use AI, as part of efforts to ensure 80% of developers use the technology each week. Internal leader boards track token usage, and whoever’s at the top… gets a gold star on their homework. Who knows.

To top the board, some employees are using Amazon’s in-house AI tool MeshClaw to create agents to automate unnecessary tasks. All in an effort to top the rankings, a clear sign that gamification works until it doesn’t.

Sadly, the FT story doesn’t offer any examples of time-wasting tasks; perhaps the agents are being sent to idly scroll Facebook, plan then cancel meetings, or whatever the digital equivalent of chit chat at the watercooler is these days.

Amazon said in a statement to the newspaper that the tool had helped “thousands of Amazonians to automate repetitive tasks each day”. Presumably it’ll be looking through what those tasks are now.

Amazon busywork: Taskmasking for AI

Of course, none of this is new, even though doing it with AI is novel. Presenteeism means employees need to look busy as well as getting actual work done. Back in the day, staff could impress bosses by walking around the office with papers stuffed in a folder.

TikTok has renamed that performative busyness as task-masking, loud labouring or “LARPing” your job. Key techniques include typing furiously to look passionate about your work; sending a lot of emails or Slack messages; acting frustrated or annoyed when the boss comes around; wearing headphones as though you simply must concentrate without distractions.

Now, in a dystopian twist that will surely make William Gibson quiver with mirth, we’re teaching it to our AI. There’s even a Gen Z name for it: “tokenmaxxxing”.

Tokenmaxxing and top-down AI

There’s no suggestion that Amazon managers are analysing the data about AI usage for use in performance evaluations, but staff are still concerned. One employee told the FT: “Managers are looking at it… When they track usage it creates perverse incentives and some people are very competitive about it.”

Monitoring AI usage is no more useful when considering employee performance than tracking how many times they open Excel or how quickly they respond on Slack. When managers focus on such metrics, it’s no surprise employees game the system.

But in this case, there’s a cost to the business. AI is, after all, very expensive, and Amazon will be paying for every token used by employees to automate something silly in order to top that pointless leaderboard. Indeed, given the destructive environmental impact of AI, we’ll all be paying for this useless tokenmaxxxing in the long run.

Either way, Amazon has found one way to ensure AI gets used by employees, amid stalling deployment of the technology that tech giants promised would revolutionise work.

To companies that have invested in AI, remember this: employees will use AI if it benefits them, you don’t need to gamify it. And to employees feeling pushed to task-mask using tokenmaxxxing, don’t automate random busywork. Consider having your AI agent look for a new job instead.

More by Nicole Kobie

Nicole Kobie
Nicole Kobie

Nicole is a journalist and author who specialises in the future of technology and transport. Her first book is called Green Energy, and she's working on her second, a history of technology. At TechFinitive she frequently writes about innovation and how technology can foster better collaboration.