Employees Are Using Their Jobs’ Super-Expensive AI Tokens for the Most Hilariously Pointless Tasks Imaginable

Under all the bluster of AI hype lies a real conundrum: companies are charging out-of-this-world prices for a tool that still can’t match the value of a competent human. As AI companies and their financial backers continue to dump billions of dollars into AI development, the consequences of that basic contradiction are spilling out into offices and shops throughout the US.

The fintech firm Slash, for example, recently encouraged its employees to start using AI coding tools as much as possible, a phenomenon known as tokenmaxxing, with the ultimate goal of boosting productivity and lowering costs. Unfortunately, LLMs are too expensive — and not quite useful enough — to really make that a reality.

At Slash, one employee wasted an astonishing $80,000 in AI tokens to vibe code a lackluster video game called “brainrot shooter.” As Business Insider describes it, the first-person shooter is a barren experience where the player runs around shooting at enemies inspired by viral internet memes.

“Pls play it so we can write this off as a marketing expense,” Slash wrote on social media.

Elsewhere, office workers who’d been told to use AI as much as possible are incinerating incredible sums of cash on tasks that didn’t need LLMs in the first place.

At consultant firm Accenture, for example, 404 Media reports that non-tech workers are using corporate AI budgets to do things like convert PDF files into PowerPoint presentations.

“We’re seeing from some of the data internally at least that it’s actually not our engineers that are driving the token consumption,” Accenture’s head of AI strategy Justice Kwak said in an internal meeting, according to leaked audio files obtained by 404. “It’s a lot of the non-engineers that are doing some of those behaviors.”

It’s a particularly ironic development, because workers are doing exactly what they were told — using the exciting new technology at all costs, thereby exposing just how little they’re getting in return.

For years, the titans of the tech industry have been subsidizing the mammoth costs of LLMs in a bid to spread their buzzy new tools to corporations far and wide. Once LLMs advance to a point where they can actually make companies money — a scenario which is becoming increasingly difficult to imagine — the market will theoretically settle on a price that makes everybody happy.

Proponents of AI support this delayed approach. They point to examples like Uber, which operated at a loss for years in order to undercut the price of taxi…