Every rollercoaster has its dip. According to Gartner’s 2025 Hype Cycle, generative AI now sits squarely in the Trough of Disillusionment. Despite investments averaging nearly $1.9 million per initiative, fewer than 30% of CEOs are satisfied with ROI (Gartner).

MIT research found that 95% of AI pilot programmes never make it past incubation (Windows Central). Meanwhile, a survey this year revealed that 42% of companies are abandoning most of their AI pilots, up sharply from 17% a year earlier (Jon Krohn).

The headlines talk of “AI disillusionment” as if this is something unique. But what strikes me is how familiar the reasons for it sound. That's because many AI projects aren’t failing because of AI, but for the same reasons technology projects have always struggled.

• Unrealistic expectations. AI was sold as the silver bullet. Boards expected breakthroughs in months, not years. When the hype doesn’t match reality, interest fades.

• Weak foundations. Research suggests up to 80% of projects falter due to siloed or poor-quality data (The Economist). The smartest model in the world won’t deliver if the plumbing is faulty. Garbage in, garbage out.

• Lack of adoption. A Guardian report highlighted rising scepticism as employees encounter hallucinations, bias, and patchy rollouts that undermine trust (The Guardian). Tools rolled out without context quickly become shelfware.

• Transformation fatigue. As my former colleagues at TechRadar note, “transformation fatigue” is becoming the silent barrier to AI success: too many top-down initiatives, not enough engagement (TechRadar).

I’ve seen this before: from my past life, TV news's shift to digital in the 2000s, the digital transformation in media more widely in the 2010s , and in my consulting practice, clients' transformation programmes in the 2020s. The same patterns repeat.

The trough of disillusionment isn’t a dead end, though. It’s a clearing-out phase. The hype burns off, and the valuable use cases survive. The organisations that succeed are those that:

• Reset expectations with the right strategic narrative, treating AI as a long-term capability, not a quick fix.

• Invest in the plumbing (the not so glamorous part): clean data, governance, integration.

• Tell the story so employees understand not just how to use AI, but why it matters to them.

• Embed the learning into the culture, to make innovation repeatable, not just improvised.

In other words: the same lessons that apply to every technology transformation.

So perhaps the real surprise isn’t that AI is in the trough of disillusionment, it's that we appear to keep forgetting the lessons of history on this particular rollercoaster.

Read Positive Momentum's research paper on technology transformation here.