Companies Are Getting Burned by Burning Tons of Tokens

Ah, the AI bubble's hangover has arrived, and companies are discovering they've spent a fortune on digital glitter. Gizmodo reports that Uber CEO Dara Khosrowshahi now admits it's 'getting harder to justify' AI costs — which is CEO-speak for 'we've been burning cash on chatbots that suggest meeting times.' Meanwhile, Amazon had to roll back an internal scoreboard ranking AI tool usage after employees apparently started gifting each other token-consuming tasks. Yes, gifting.

Amazon's scoreboard was meant to gamify AI adoption — because nothing says 'strategic implementation' like a leaderboard. Instead, workers gamed the system by assigning meaningless tasks to each other, driving up token consumption while the company footed the bill. This is the corporate equivalent of a crypto 'yield farming' scheme, but with less glamour and more API calls. The rollback came quietly, but the lesson is clear: if you incentivise usage without oversight, people will optimise for the metric, not the goal.

And Uber's Khosrowshahi isn't the only one wincing. The article notes that many companies are realising the ROI of large language models is murky at best. When your CEO publicly questions the cost, you know the spreadsheets are ugly. The industry promised a revolution; what we got was an expensive auto-complete that employees use to prank each other. Still, the bills must be paid — so keep burning those tokens, and maybe next quarter you'll afford a model that can count. 🇬🇧