Tech CEOs Have Caught 'AI Psychosis' and Workers Are Paying the Price

I called my old colleague Frank at a systems integrator last month. He's seeing the same dynamic play out with warehouse automation clients. "They want to cut headcount before the robots are even commissioned," he told me. "Then they're shocked when throughput drops 30%." Frank's been doing this for 25 years. He sounds tired.

The TechCrunch piece frames this as a debate over whether tech CEOs are uniquely prone to this kind of thinking. I'd argue it's not unique to tech. It's just more visible there because the layoff announcements come with press releases bragging about AI transformation.

Is the technology actually ready?

This is where it gets complicated. Some AI applications genuinely work. I've seen machine vision systems that outperform human inspectors on specific, well-defined tasks. But "specific and well-defined" is doing a lot of heavy lifting in that sentence.

The gap between demo and deployment remains enormous for most applications. We don't have good data on how many of these AI-driven workforce reductions actually achieve the promised productivity gains. The companies aren't exactly rushing to publish those numbers, which tells you something.

It's too early to say whether the current wave of cuts will look smart or catastrophic in two years. My gut says it's going to be a mix, with the catastrophic cases quietly reversed and the smart ones loudly celebrated. Survivorship bias will make the whole thing look more successful than it was.

What happens to the workers?

This is the part that bothers me most. When I started in this industry, automation displaced manufacturing jobs, but it created new ones in integration, maintenance, programming. The net effect was, well, complicated but not purely destructive.

The AI displacement seems different. If you're cutting knowledge workers because you think an LLM can do their job, what new jobs are you creating? Prompt engineers? That's not a one-to-one replacement.

I don't have answers here. Nobody does, really. But I know that making major workforce decisions based on technology that's been deployed at scale for maybe 18 months seems, in a way, reckless. The executives making these calls won't be the ones scrambling when the AI systems hallucinate something important or fail in ways nobody anticipated.

Levie called it psychosis. I might call it something less clinical. But he's not wrong about the diagnosis.

Robert "Bob" Macintosh · 20 hours ago · 4 min