AI Mania Is Eviscerating Global Decisionmaking

Nikhil Suresh has a front-row seat most people don't: hundreds of catch-ups with executives and workers across industries, plus hands-on work leading technical engagements for his firm's clients. What he describes isn't a mixed bag of AI wins and losses. It's something closer to a company-wide fever, one where admitting a project isn't working can get you fired, and where the people signing off on strategy sometimes haven't used the tools they're mandating.

His core claim is blunt: every AI project his team has observed, whether they were hired to help or just happened to be nearby, has failed. Not most. All of them, over a year and a half of watching. He's careful to note this isn't always the technology's fault. Plenty of companies are simply bad at running software projects in general, and AI work carries the added risk that even a well-executed rollout can flop because the whole approach is still new. Internal chatbots go unused because the underlying documentation is garbage and an LLM can't know what was never written down. Customer-facing bots, he says, rarely deliver a good experience either — he cites a Mitsubishi support call where a smooth-sounding AI promised a callback that never came, and no one can say whether that failure even registered anywhere as a problem.

What's more unsettling than the failures themselves is the culture that's grown up around hiding them. Suresh describes an environment at large companies where continued employment increasingly depends on publicly professing faith in AI's transformative power, regardless of whether anyone can point to an actual transformation. He recounts an executive at a company with $2B-plus in revenue building an entire AI-centered strategy without ever having used ChatGPT personally. Engineers, meanwhile, have started