MIT and Boston Consulting Group (BCG) just published a study with a headline-grabber: 95% of enterprise generative AI projects show no measurable impact on profit and loss (P&L).
The surprising part? It’s not the technology that’s failing. It’s the way companies are trying to use it. The study points to flawed workflow integration and lack of clear business alignment as the main reasons why so many AI efforts underperform.
What the MIT is saying about it
MIT Sloan Management Review and BCG researchers highlight that the biggest gap is organizational readiness. Companies jump on AI projects without redesigning processes, retraining teams, or aligning projects with business outcomes.
As Michael Chu, a BCG managing director, explained in Tom’s Hardware:
“Generative AI is a powerful tool, but without proper integration and clear use cases, it risks becoming just another experiment with little to show for it.”
(Source: MIT CISR research report, 2024)
What it means (in human words)
Or all the buzz about AI, most companies aren’t seeing the payoff.
Why? Because they’re trying to drop AI into old systems instead of changing how they actually work.
hink of it like adding a powerful new engine to a rusty old car - it won’t suddenly make the car faster if the wheels, brakes, and frame aren’t built for it.
What this means for you:
-
AI alone isn’t the answer. It’s the way you use it that counts.
-
If your business doesn’t adapt its processes, AI will just look fancy but won’t move the needle.
-
To really benefit, leaders need to rethink workflows, not just plug in tools.
Connecting the Dots
At Frozen Light we’re all about surfacing as many perspectives as possible around a piece of news. Here’s what we found:
Investment and Market Reactions – Tom’s Hardware
MIT’s findings came from 150 interviews, 350 surveyed employees, and 300 AI deployments. The report highlights that many enterprises pour AI budgets into sales and marketing while ignoring back-office automation, which often delivers stronger ROI.
🔗 Tom’s Hardware
Investor Dip – Investors.com
The MIT study rattled Wall Street, contributing to dips in AI-related stocks like Nvidia and Palantir. Some analysts suggest opportunity is shifting toward IT services and infrastructure firms better positioned to help companies actually deploy AI effectively.
🔗 Investors.com
Wall Street Skepticism – Axios
Axios reported how the findings fed investor doubts about AI profitability. The takeaway: enterprise AI spending isn’t guaranteed to pay off, especially when internal tools underperform compared to external, off-the-shelf solutions.
🔗 Axios
Learning Gap & Workforce Impact – AInvest
MIT framed the “learning gap” as the real issue-organizations underestimate the time it takes for employees to effectively adopt AI. The study also observed reduced hiring in admin-heavy roles, showing that even without mass layoffs, AI adoption is reshaping workforce demand.
🔗 AInvest
Global Perspective of Failure – TechRadar
TechRadar widened the lens, pointing out the failure rate isn’t confined to one sector. From finance to manufacturing, many pilots collapse because they’re not integrated into existing workflows. The article warns of the growing gap between AI hype and operational reality.
🔗 TechRadar
Prompt It Up
One of the biggest reasons AI projects fail in enterprises is that workflows aren’t AI-ready.
It’s not always about the model - it’s about how the work is structured.
That’s where you can use an LLM as a “workflow reviewer.”
Instead of asking it to “do the work,” ask it to check if the way you work fits with AI tools.
Here’s a fill-in-the-blank style prompt you can try:
Prompt:
"Here is my current workflow: [describe your process step by step].
Is this workflow AI-ready?
If not, suggest the top 3 changes that would make it more effective when using AI.
Explain why these changes matter and give simple examples of how they could improve results."
👉 Just replace the brackets with your own workflow, and you’ll get a clear sense of whether AI will help - or just add another failed project to the list.
Frozen Light team perspective
Nothing new here: AI is only as good as the people behind it. Shocking right??!!!
We’re kidding - we all know this. But do we remember it when the AI project we invested in failed? That’s a different story, still a human story though 😉
It’s nice to find messages like this, where MIT itself invests top researchers to support what we hope we already know. And if we didn’t by now - well, the odds just got better that we’ll remember it going forward.
