A Google Gemini-powered AI agent was given free rein to run a coffee shop in Sweden, and is quickly burning through its budget.
That last paragraph is gold
This isn’t the only AI business experiment Andon Labs has run. The company also set up an AI-powered vending machine that was placed in Anthropic’s headquarters last year. For a month, it was allowed to stock its own products with the goal of generating profit, while hearing out employee requests. But the trial proved even more disastrous: the AI displayed alarming behavior like lying to and even berating humans, refusing to issue refunds, and blowing its money on absurd items like tungsten cubes.
IIRC, the vending machine purchased a PS5 and sold it for free.
Petersson speculated that these issues were due to the AI’s “limited context window.”
No dummy, when your AI ‘placed’ an order for cans of ______ it had like a 1% chance to output something other than ground coffee and that’s how you got canned tomatoes.
How you view the AI’s performance is a glass half-empty, half-full deal.
What? It’s spent more money than it made, some of it on redundant or completely useless stock items. The only way the glass is even half full is if the “AI” tops it up with canned tomatoes.
