Predicting the future is an impossible task. Not only do we lack the ability to foresee what will actually happen—whether it’s new innovations, regulatory changes, or social upheavals—but we also can’t predict how the technologies we adopt today will actually be used by those who adopt them.

There is something we can predict, though—and that’s what happens when commercial technologies fail to live up to the hype. The tech sector, in particular, is riddled with ideas pitched as “the next big thing,” only to crash into fiascos and take millions of dollars down with them. 

Think Google Glass or the ongoing struggle for relevance of the Metaverse—unheeded cautionary tales that continue to rattle the industry and, ultimately, shape all our relationship to these companies.

Make the hypes—and the industries—big enough, and you get a bubble. Bubbles are the dangerous economic fallout that occur when investors can’t extract anything tangible from speculative tech.

Surprisingly enough, the AI sector is no stranger to the fallout that comes when promises go undelivered. Historically, these downturns are known as “AI winters”—periods when both public and private investment dries up due to unmet expectations and overhyped potential.

As the hype around GPTs begins to fade—and companies continue to fail at replacing humans with chatbots—we now find ourselves at the edge of a possible new AI winter. The public has adopted the technology quickly, but the pipe dream of a Cortana or HAL 9000 is dying faster than marketing departments can print new feature leaflets.

This collective dream, however, might be exactly what’s driving the AI winter to the forefront this time. A few years ago, the capabilities of models like ChatGPT were said to be limited by the sheer volume of digital content available. It was even claimed that a true “V5” wouldn’t be possible—simply because there wasn’t enough content in the world to train it. So when the latest release arrives rebranded as “agentic AI,” it feels more like a PR maneuver to fend off mounting competition—much like Apple repackaging years-old features as exciting new perks.

With this, OpenAI risks signaling even greater appeal to audiences not well-informed enough to realize that we are still nowhere near achieving true interactive intelligence. This, in turn, may unwittingly encourage misguided companies to replace their human talent without proper substitutes—leading to both job insecurity and growing public distrust or even scorn toward the technology itself.

Maybe it’s time we ask ourselves whether the hype is worth risking the collapse of the entire industry—setting us back decades just to boost this quarter’s earnings.

My book:The Pocket AI Guide: A Practical Introduction for Professionals launches in August 2025.