We continue our wonderful column "Corporations Realize AI Costs Money." Earlier we discussed Uber and Microsoft's AI hangover, but it keeps going.
1️⃣ Amazon Officially Removed AI Leaderboards
Earlier I wrote how Amazon forced engineers to trigger AI agents for pretty dashboard statistics. The result: the company officially shut down the KiroRank platform.
Employees massively ran AI on meaningless circular tasks just to climb the rankings. Amazon saw exponential growth in infrastructure costs with zero business output. Vice President Dave Treadwell had to personally ask staff: "Please don't use AI just for the sake of using AI."
Now Amazon is frantically trying to implement a new metric — "normalized deploys" (assessing real value), because counting raw tokens turned out to be too expensive.
2️⃣ $500,000,000 in 30 Days on Prompts
Why did big tech suddenly start cutting costs so sharply? Axios provides a great example: a client of one AI consultant managed to burn over half a billion dollars in just one month because employees were given unlimited access to Anthropic Claude.
People by default began delegating to neural networks not tasks that bring business value, but those they were simply too lazy to do manually. Now analysts are sounding the alarm, stating that without strict quotas, budgeting, and monitoring, AI integration will simply bankrupt businesses.
It turns out that attaching a powerful LLM to bad processes is not innovation. It's just very expensive bad processes.
Comentários
0Ainda não há comentários.
Entre para participar da conversa.