Follow ZDNET: Add us as a preferred source on Google.
ZDNET’s key takeaways:
- Just 5% of enterprise customers are profiting from generative AI.
- A bottom-up versus top-down approach can improve implementation success.
- AI companies are making big promises in a bubble, most of which are unfulfilled.
Investment in generative AI may be booming, but most individual businesses using it have yet to see the payoff. In fact, a new MIT study found that 95% of enterprises attempting to harness the technology aren’t seeing measurable results in revenue or growth.
Also: Gen AI disillusionment looms, according to Gartner’s 2025 Hype Cycle report
The study, conducted by MIT’s Networked Agents and Decentralized AI (NANDA) project, was based on interviews with over 150 business leaders and an analysis of 300 business deployments of generative AI.
“Just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact,” the authors write in the report.
It paints a stark contrast between promises and reality: while tech developers are selling AI tools like agents as productivity boosters, NANDA’s new report indicates that for all but a vanishingly small minority, the technology is having little to no effect on businesses’ bottom lines. What accounts for the huge disparity?
What isn’t working – and what could
It largely boils down to a matter of bureaucratic inefficiency. Generative AI tools can provide efficiency gains in the hands of competent individuals, but when business leaders attempt to integrate them into existing, company-wide operations and workflows, they tend to throw a wrench into the organizational machinery.
Also: 71% of Americans fear that AI will put ‘too many people out of work permanently’
The main reason for this, according to the report, is that the generative AI systems that most businesses are attempting to deploy internally and at scale lack the ability to seamlessly adapt with existing organizational workflows, ultimately making them more of a hindrance than an accelerant.
“The core barrier to scaling is not infrastructure, regulation, or talent. It is learning,” the authors write. “Most GenAI systems do not retain feedback, adapt to context, or improve over time.” While an ability to remember past interactions, customize outputs to different contexts, and learn over time are all key traits of AI, the authors are specifically referring to the context of the technology’s use within enterprise-scale operations.
One of the implications of the new study therefore seems to be that in order for businesses to make the most of generative AI, they’d do well to take a bottom-up (allowing employees to experiment and discover their optimal mode of human-AI collaboration) as opposed to a top-down approach (forcing all employees to use a particular tool in a manner that’s tightly controlled by executives and supervisors).
Also: Stop using AI for these 9 work tasks – here’s why
Another trend that emerged from the study was flawed prioritization in the application of generative AI. Many businesses that were failing to profit from the technology were using it for marketing and sales, while the 5% that were using it successfully tended to do so through the automation of more fine-grained and mundane “back-office” tasks.
Based on their study, the authors predict that future success will belong to those businesses that deploy agentic and adaptable models in the right places, while those that choose a general, top-down approach will continue to be frustrated.
“The next wave of adoption will be won not by the flashiest models,” they write, “but by the systems that learn and remember and/or by systems that are custom built for a specific process.”
AI hype and cultural pressure
On its surface, the NANDA study seems to lend support to the belief that generative AI is nothing but a massive hype bubble that will soon pop, not unlike the short-lived corporate rush into the metaverse that preceded it. If such a massive proportion of businesses aren’t seeing results, then surely that means the technology is being pedaled on empty promises, right?
Time will tell. For now, companies across the board are doubling down on their investments in AI, promising customers and investors that the rise of more agentic systems will usher in a golden age of prosperity, creativity, and leisure. At the same time — and on the heels of a GPT-5 launch that received mixed reviews — OpenAI CEO Sam Altman himself said he sees an AI bubble taking shape.
Also: 5 ways automation can speed up your daily workflow – and implementation is easy
Meanwhile, the widespread cultural embrace of AI means that companies are facing huge pressure to integrate the technology quickly — or risk looking like dinosaurs. As NANDA’s study indicates, this rush is, in many cases, apparently taking place at the expense of any kind of well-calculated plan, and as a result, investments in generative AI are leading many companies nowhere.
Even at the individual level, generative AI can be counterproductive in the long-term — even while boosting productivity in the present. A recent study conducted by Workday, for example, found a correlation between heavy use of AI at work and employee burnout, while other studies find evidence that AI use degrades critical thinking skills.
(Except for the headline, this story has not been edited by PostX News and is published from a syndicated feed.)
