Why MIT’s AI Failure Study Is Misleading Leaders

MIT claims 95% of GenAI investments failed.

Immediately, the headlines flooded in: "95% of projects crash and burn!"

Meanwhile other sources such as S&P claim 42% of AI projects fail.

Apparently, we live in a world where both can be true — depending on semantics, and which number gets you more clicks (hint - the 95% failure rate gets the most attention).

And of course, a wave of takes explaining why the lucky 5% succeeded.

Here’s the problem: most of those takes are just narrative theater.

It appears those posting about it either haven’t read the article or are purposefully misrepresenting it.

Even fewer acknowledge its glaring limitations:

“Failure” simply means no measured P&L gain - some gains are hard to measure

Sample size wouldn't pass a freshman stats class

Complete blindness to explosive growth in areas like agentic coding

Yet overnight, it became an echo chamber. AI is failing almost everywhere. While at the same time, AI is succeeding. And causing layoffs. And saving the economy. Take your pick.

The truth is that it's complicated. 

Some initiatives collapse because leadership and workflows break down, while others flourish when timing, culture and execution align perfectly. 

And some "research findings" are just marketing dressed in academic clothing.

So no, neither you nor I can definitively explain the 5% success stories (because it’s not 5%). 

Pretending otherwise isn't thought leadership — it's fear manipulation with a LinkedIn filter.

The real MIT lesson isn't hidden in percentages. It's that surface-level narratives provide zero value to leaders building serious AI strategy.

Want meaningful results? Stop the headline addiction. Dive into context, workflows, and execution.

That's where the signal hides while everyone else chases noise.