MIT NANDA’s The GenAI Divide: State of AI in Business 2025 studied over 300 AI initiatives, conducted 52 structured interviews, and surveyed 153 senior leaders across four industry conferences. Its headline finding:

95% of organizations are getting zero measurable return from GenAI investment.

That stat made the rounds online, usually compressed into “95% of AI projects fail.” The framing felt alarmist, and for a while, I set the report aside.

But a recent client conversation brought it back. “We have ChatGPT, Copilot, and two other AI tools. But honestly? Nothing has really changed.”

That sentence captured exactly what the report describes, and it forced me to look again.

The headline may be imprecise. The underlying diagnosis is not.

Three Findings That Hold Up Under Scrutiny

The report’s most viral stat deserves context. It measures a specific slice of enterprise deployments within a limited timeframe, not all AI investment everywhere. But strip away the headline, and the patterns underneath are well supported.

The Pilot-To-Production Gap

60% of enterprises evaluated AI. 20% piloted it. Only 5% reached production. The technology worked. The organizations did not. They could not figure out how to move from experiment to operating reality. That is a resourcefulness failure, not a technical one.

The Shadow AI Economy

Only 40% of companies have an official LLM subscription. But workers from 90% of companies already use personal AI tools for work. The most resourceful employees are not waiting for permission. They are solving real problems with AI through unofficial channels. Call it shadow IT if you prefer. I see shadow resourcefulness, hiding in plain sight.

The Learning Gap

Most GenAI systems do not retain feedback, adapt to context, or improve over time. The report calls this the learning gap. Resourceful organizations close it: they capture what works, feed it back in, and compound the learning. The rest start from scratch every time.

The Common Thread: Organizational Resourcefulness

All three patterns point to the same underlying issue. The divide between the 5% getting value and the 95% getting nothing runs deeper than AI strategy, tool selection, or budget.

It is about organizational resourcefulness: the ability to find what is already working inside your organization, make it safe, and scale it deliberately.

The employees using personal AI tools on their own time are the signal. They are showing you where the friction is, and what the organization has not equipped them to do officially.

Read that as a compliance issue if you prefer. For L&D and talent development leaders, it is better understood as an enablement gap waiting to be closed.

Ending thought:

The GenAI Divide is real, but it is not a technology story. It is a resourcefulness story. The organizations on the right side of the divide are not the ones with the biggest budgets or the best models. They are the ones that noticed what was already working, made it safe, and scaled it deliberately.

For leaders in L&D, talent development, and HR, the opportunity is clear: stop waiting for perfect conditions and start building from what your people are already doing. The signals are there. The question is whether your organization is resourceful enough to learn from them.

At Radiant Institute, we work with organizations across Malaysia and APAC to build structured AI enablement programs that close exactly this kind of gap. If this resonates, we would welcome a conversation about how to make your teams more resourceful with AI.