Crossing the GenAI Divide: Why 95% of AI Projects Fail (and How to Get It Right)

When OpenAI’s CEO, Sam Altman, uttered the word “bubble” in reference to artificial intelligence, headlines exploded. Investors panicked, executives hesitated, and the chatter around AI shifted from excitement to skepticism almost overnight. And now, even Mark Zuckerberg is saying a ‘collapse’ is ‘definitely a possibility’. Soon after, MIT’s GenAI Divide report revealed that 95% of enterprise AI pilots have produced no measurable return. For many, that was all the confirmation they needed: AI was already overhyped.

But that’s not what Altman meant. And it’s not what MIT’s research uncovered. The truth is more nuanced and requires a deeper dive into this topic.

The Misleading Narrative

As Fortune’s Jeremy Kahn noted, a bubble doesn’t always mean collapse. It often signals where breakthroughs emerge. The dot-com bubble, for example, gave us Amazon, Google, and the modern internet economy that transformed even our basic social fabric. Froth may fade, but it often reveals companies that define the next generation. At the same time, it also leaves behind many skeletons in its wake. Over 50% of the dot-com companies have failed and are no longer in business because they failed to truly impact the business.

MIT’s research was also misconstrued. AI isn’t failing — executives are failing to apply it effectively. The technology is capable. What’s missing is focus, execution, and design thinking.

The GenAI Divide

At the heart of MIT’s findings is the GenAI Divide:

• 95% of organizations are stuck in outdated approaches by funding siloed pilots, chasing flashy demos, and pouring budgets into front-office use cases like marketing, even though the bigger ROI is in back-office and production automations.
• 5% of organizations are pulling ahead by aligning AI with workflows, empowering domain experts to drive adoption, and focusing on measurable business outcomes.

This isn’t a technology problem. It’s an execution and design problem. Leaders are repeating the same mistakes of the digital transformation era: too many scattered pilots with no clear link to value and no clear link to business-centric problem solving.

What Leaders Get Wrong

Three common mistakes explain why AI efforts stall:

1. Hype over context – Pilots are launched because demos look impressive and UI looks beautiful, not because they solve pressing business problems. Most vendors do not even bother learning about the client’s business and their core challenges before they try to jump into their sales pitch and demo.
2. Cost-cutting mindset – Leaders see AI as a way to reduce headcount instead of an opportunity to innovate, open new markets, or design better workflows. Back in the early days of the dot com frenzy, many so-called ‘experts’ predicted that the internet revolution would bankrupt the postal mail service and Xerox printers because everything would be digital. Fast forward to 2025, we still have the need for post service workers and we still print and copy on paper.
3. DIY over partnerships – MIT found that vendor partnerships succeed twice as often as internal builds. Yet many large firms insist on building their own proprietary tools, wasting years and millions. They forget that there is power in focus and depth. The vendors who are focused and dedicate their entire existence into one specific discipline will always outpace and out innovate any internal ‘multi-utility’ IT departments. Not to mention, they accumulate many valuable lessons-learned from many of their clients.

In short, leaders confuse activity for progress. As one CIO in the MIT study explained: “We’ve seen dozens of demos this year. Maybe one or two are genuinely useful. The rest are wrappers or science projects.” It’s no surprise that when faced with polished sales pitches backed by multi-million dollar marketing budgets, even seasoned executives can struggle to separate genuine value from glossy hype.

Hype, Mergers, and the Reality of Adoption

The recent announcement of Workday’s merger with Sana Learning AI is a case in point — it shows how competitive and crowded the AI landscape has become. On paper, these combinations of enterprise platforms and AI startups sound like silver bullets. But anyone who has lived through a large-scale acquisition knows the reality: integrating two technology stacks and two different organizational cultures can take years before delivering measurable impact.

This is not unique to Workday. Oracle recently struck a deal with OpenAI, and countless other companies are racing to align themselves with AI through acquisitions, partnerships, and rebrands. It feels reminiscent of the late 1990s, when any company with “dot com” in its name attracted sky-high valuations. Fast forward, and only a handful of true innovators thrived while others faded into the history book. Remember AOL?

Like Workday, many big tech companies like Google, Apple, Meta, Oracle, and Microsoft are all now racing to become the “front door to work”. AI agent platforms are designed to sit at the center of the employee experience, orchestrating tasks, knowledge, and workflows. But enterprises will not need multiple competing “front doors”, each requiring employees to toggle between different systems. If your enterprise is using Workday and Microsoft, which AI agent will become your single source of truth? Which one will you spend time and resources to train? Ultimately, there will be consolidation and only a few platforms will emerge as the true entry point to work and become the overarching AI agent platform. For now, though, the market is noisy and crowded, with vendors jockeying for position and making bold claims that may take years to prove out.

The lesson is clear: hype can excite investors and customers in the short term, but only those who translate AI into sustainable business value will endure.

What the 5% Do Differently

The few organizations seeing impact from AI are taking a different path:

The L&D Parallel: Why This Matters Beyond AI

For those of us in Learning & Development, this pattern sounds familiar. Too often, organizations chase the latest shiny object — LXP, VR, AR, gamification, now AI — without connecting it to business outcomes. The result? Investments that look good in a pilot but fail to scale or demonstrate real value.

At Seertech, we’ve seen this firsthand. Our mission isn’t to sell technology as a widget; it’s to help organizations solve real business problems. From my years leading learning at Texas Instruments, I learned that real transformation requires a disciplined approach to change management, design thinking, and smart application of lean principles to iterate improvements. That’s why Seertech’s platform is built not just to deliver learning content, but to close skills gaps, improve manager’s productivity, and measure impact on business performance.

The AI conversation reinforces a lesson L&D leaders already know: technology is only as powerful as the strategy and real human-centric partnership behind it.

Final Thought

The GenAI Divide isn’t proof that AI is hype. It’s proof that leadership hasn’t caught up. AI isn’t failing us, we’re failing to use it effectively.

The real question for leaders is this: are you using AI to improve the past, or to invent the future?

Like the dot-com bubble, this moment will create winners and losers. Those who stay grounded in solving real business problems, who measure impact, and who choose partners that understand their journey will be the ones who cross the divide and thrive in the next era of business.

If this article resonated with you, please reach out to us for a free initial business consultation.