MIT Says 95% of AI Investments Fail to Deliver ROI. Here’s What They Missed.

In July, MIT’s NANDA Initiative released a new report, The GenAI Divide: State of AI in Business 2025. Its headline finding is hard to ignore: despite $30–40 billion poured into generative AI pilots, 95% of companies have seen no measurable return on investment.

It’s a striking number, and one that has added fuel to the growing debate about whether AI is living up to the hype. The MIT researchers point to a fundamental limitation in the technology: today’s AI models don’t learn, and they don’t remember. That’s true in a narrow sense, but it’s not the whole story. In my experience working with executives and leadership teams, the bigger issue isn’t the machines. It’s the humans who use them.

The MIT Explanation: A Technology Gap

MIT’s research team concluded that today’s AI models don’t deliver ROI because they don’t learn and don’t remember. From their perspective, the tools reset with every prompt, generating fluent language in the moment but failing to retain organizational knowledge or adapt to workflows over time.

That claim already feels shaky. Models like GPT-4 are updated constantly, and they do learn, just not in the same way humans learn from each lived experience. And persistence isn’t a hard technical limit. It’s a design choice. With memory features now standard in leading platforms, and with cognitive architectures like the Second Brain framework I’ve developed at haveLAB, AI can indeed adapt and evolve alongside the humans using it.

The real bottleneck isn’t the tool’s ability to learn, it’s the human’s ability to transform how they work, and to design workflows that turn AI into a thinking partner instead of a novelty.

The Missing Half of the Story

The uncomfortable truth is that many leaders want AI to be a “magic button.” They want to type in a question and get a perfect answer. They want ROI without the hard work of developing new skills, new habits, and new ways of thinking.

But that’s not how transformative technologies work. ROI only emerges when humans change their practices to match the potential of the tools. And the greatest intangible benefit of AI when used correctly is that it doesn’t just help you think faster, it rewires your own brain to think better.

When leaders treat AI as a Second Brain, they engage in new cycles of attention, intention, and reflection that reshape their First Brain through neuroplasticity. Over time, that rewiring makes them sharper, more creative, and more adaptive, benefits that compound far beyond any single pilot or quarterly ROI measure.

Lessons from the PowerPoint (and Excel) Revolutions

When Microsoft PowerPoint was introduced in 1987, few expected it to transform business. Within a decade, it had done more than replace overhead projectors, it rewired how organizations and executives thought and communicated. MBA programs and consulting firms built entire methodologies around slides and pitch decks. For many of us, building a deck became more than a way to present, it became a way to think.

Excel had a similar impact. It didn’t create ROI on its own. The breakthrough came when people learned to think in spreadsheets by reframing problems into rows, columns, and formulas.

The lesson is clear: technology itself doesn’t deliver value. ROI comes when humans transform their way of thinking to match the tool. And just as PowerPoint reshaped communication and Excel reshaped analysis, generative AI is now reshaping leadership. But only for those willing to adapt.

Why Are We Holding Back?

So why are we so slow to evolve in the AI era? A few patterns stand out:

• Surface-level engagement. It’s easier to consume AI outputs than to process them critically. Many leaders stop at the first draft instead of iterating.

• Fear of falling behind. Executives are reluctant to admit they don’t know how to use AI well, so they avoid experimenting in public.

• Overwhelm. With hype everywhere, many leaders don’t know where to start, so they dabble in pilots that never scale.

• Old habits. Leaders cling to legacy ways of working, assuming AI should fit into their way of thinking rather than adjusting their thinking to leverage what AI offers.

The result is predictable: organizations run “pilots” without real behavioral change. They test the tools, but they don’t retrain the humans. And so ROI doesn’t materialize.

The Path Forward: Humans + AI

The way out of this ROI problem is not to wait for better technology. Yes, memory-enabled models are coming. Yes, persistent agents will improve. But if leaders sit back and expect the tools alone to deliver transformation, they will stay with the 95% who see no return.

The real path forward is strengthening human intelligence and building new capabilities.

Leaders must learn how to:

• Offload overload. Use AI systems to capture and structure inputs, so their First Brain can focus on higher-order thinking.

• Process with precision. Sort AI outputs into insights, actions, and references instead of letting them pile up as noise.

• Build memory systems. Create external architectures that can be used to retrieve knowledge, when it matters, even if the model itself forgets.

• Rewire their First Brain. Develop new habits of attention, intention, and reflection so working with AI becomes second nature.

This is exactly why I built the Second Brain framework at haveLAB, to help leaders make that leap. The framework gives structure to the chaos, turning AI from a distraction into an extension of your cognition.

Your Second Brain

MIT’s report says 95% of companies aren’t seeing ROI from AI. They’re not wrong about the symptoms, but they missed the cure: most organizations are still stuck in surface-level engagement. Leaders haven’t yet built the habits and new ways of thinking to truly partner with AI.

That’s exactly what the Second Brain program is designed to solve. Its core premise is simple: AI’s greatest value is how it can help you think, and over time, how it rewires your brain to think better.

In four live sessions, I walk leaders through creating a personal AI collaborator that mirrors how their mind works. We build a system that helps them reduce mental clutter, see choices more clearly, and step into decisions with calm confidence.

This isn’t about productivity hacks or shiny new apps. It’s about upgrading human intelligence and building new capabilities, the very things the MIT report overlooked.

By the end, participants have:

• A personal Second Brain inside ChatGPT, built around their identity and priorities and habits of thinking and working.

• A repeatable way to process AI outputs into insights, actions, and references.

• A memory system that keeps ideas alive and actionable.

• Daily practices that ensure their Second Brain grows alongside them.

The real path forward isn’t waiting for the next model upgrade. It’s learning how to use today’s AI to upgrade how you think.

Join the 5%

MIT is right: the technology isn’t perfect. But that’s only half the story. The other half is you.

The organizations that will win in the age of AI aren’t the ones with the biggest budgets or flashiest tools. They’re the ones with leaders willing to transform their thinking. Leaders who put in the work to rewire their minds and build a Second Brain alongside their First.

If you’re ready to stop dabbling in AI pilots and start driving real results, now is the time to act.

Explore Working Together

We are offering special Back2School pricing for our fall Second Brain sessions. Seats are limited and filling up fast. Enroll now and take the first step toward becoming the kind of leader who thrives in partnership with AI.

Get the Book: The AI-Enhanced Leader: How to Upgrade Your Thinking and Leading

Now available on Amazon.