4 reasons AI projects fail that have nothing to do with technology

Having worked with dozens of companies in various stages of AI adoption, I’ve had a front-row seat to the myriad reasons (and sometimes excuses) why AI projects fail to launch, fail to make it past pilots or fail to deliver business value and ROI.

While every organization’s circumstances are unique, the root causes are often surprisingly familiar. Like so many technological leaps that came before AI, fear, culture and competing priorities are often the biggest barriers to enterprise success.

1. Fear of job replacement

It’s no secret that employees across industries, roles and seniority levels can see the writing on the wall: AI will affect their careers. According to Pew Research, 52% of workers are concerned about AI’s future impact on the workplace, and 32% believe it will reduce job opportunities in the long run.

As a result, there may be resistance, or just a lack of enthusiasm, to AI initiatives. This can cause AI success to stall in the form of slow adoption, low engagement and knowledge hoarding. One Writer study even found that 29% of employees (and 44% of Gen Z) admit to sabotaging their employer’s AI strategy.

There is a common refrain, and new research from Gartner to boot, that beginning in 2028, AI will create more jobs than it eliminates. Even so, such assurances can ring hollow to employees. The bitter pill for tech leaders to swallow is that there is little certainty that the jobs to be created will be well-paid or accessible to workers whose roles were eliminated.

Tech leaders are often surrounded by high performers, innovators and professionals who naturally view change as an opportunity. In these environments, it’s easy to overlook that many workers experience transformation differently—and would prefer predictability over a disruption to their routine or simply don’t have the bandwidth to pivot.

Take secretarial work, which was once a well-compensated role, especially for women without an advanced degree. Technology—namely computers, email, software and virtual assistants—enabled the reduction in demand for these professionals, not overnight but over the course of several decades. More than 2.1 million administrative and office support jobs have disappeared in the U.S. since 2000, according to Labor Department data. While there are many professionals who upskilled or changed careers, The Washington Post reports that middle-aged and older workers have had a hard time finding work within their skill set with similar pay and benefits.

On the other end of the spectrum, AI is empowering many employees to lift the ceiling on their potential by expanding what we can do and who can contribute high-value work. A rising tide may lift all boats, but those who Microsoft dubs “Frontier Professionals,” who are the most advanced AI users, are most likely to benefit from new job opportunities created by AI. The Writer study shows that 92% of the C-suite are actively cultivating “AI elite” employees, while 60% plan layoffs for non-adopters.

No leader can promise what the labor market will look like a decade from now. What they can do is provide clarity about the next six months to two years. Moreover, supporting employees with tools that help them prepare for the future is more valuable than trying to offer certainty about the future.

Provide a transparent roadmap for your organization’s AI implementation goals. Acknowledge the fear, but also the possibility, and help employees process the changes they are living through by providing access to information, continuing education, redesigned workflows and sandbox environments for AI learning and experimentation. The exact way your organization approaches the fear of job replacement will depend on the nature of your industry and its professionals. Some roles will change dramatically in a few years, while others may change slowly over decades, as secretarial roles did.

Ironically, despite fears of job displacement, AI workforce impact remains low, according to The State of AI in the Enterprise Report. The most immediate barrier to AI adoption is often the opposite: a shortage of AI skills and systems. 

2. Lack of AI-first culture

Many organizations purchase AI technology without redesigning current business processes and workflows around it, which can lead to failed adoption. AI adoption is less like a software rollout and more like an organizational transformation initiative that requires “cultural openness” to a process or workflow reset.

Despite the anxiety around AI at work, the Microsoft Work Trend Index Annual Report found that “In many cases, people are ready. The systems around them are not.” The research shows that 65% of AI users fear falling behind if they don’t adapt fast. Yet 45% say it feels safer to stick with current goals than to redesign work with AI—and only 13% are rewarded for reinventing how they work, even when results fall short. This demonstrates a paradox where organizational metrics, incentives and norms keep employees anchored to the past way of doing things.

There is no universal blueprint for an AI-first culture. What it looks like will vary by organization, industry and workforce, and it will continue to evolve as AI capabilities mature. But a common thread is prioritizing a growth mindset. As Microsoft Chief People Officer Amy Coleman and WSJ Leadership Institute President Alan Murray discussed in a recent interview, “Stop being a know-it-all company and start being a learn-it-all company.” That means encouraging experimentation despite imperfect conditions, permitting employees to fail, rewarding those who succeed, and ensuring leaders model the behaviors they want to see.

Learning and development alone are not enough. An AI-first culture must also prioritize strong data foundations and workflows, which may be one of the most challenging barriers to overcome. The State of AI in the Enterprise Report found that although 42% of companies surveyed believe their strategy is highly prepared for AI adoption, they feel less prepared in terms of infrastructure, data, risk and talent.

For leaders who view culture as a secondary concern, the numbers tell a different story. The Microsoft report revealed 67% of AI impact comes from culture, manager support and talent practices, which is more than double the 32% tied to individual mindset and behavior.

3. Competing priorities and misaligned incentives

One of the least discussed reasons AI projects fail is that different stakeholders are optimizing for fundamentally different definitions of success. Consider an ITSM AI initiative: the CIO is tasked with reducing technology costs, the service desk wants faster ticket resolution, builders want scalable systems and the legal department is concerned about compliance and liability. Each group may support the project in principle, but they are measuring success through entirely different lenses.

Without alignment on a shared business objective, teams might struggle to balance the inevitable trade-offs AI projects require. Teams optimize for their own priorities rather than a common outcome, resulting in slower decisions, competing incentives and a lack of ROI. They might also be working off of incentive structures that reward the old way of doing things. For example, if an IT team is rewarded based on tickets resolved, there is little incentive to drive down ticket volume in the first place.

In some organizations, the problem runs even deeper. Rather than optimizing for a business outcome, they’re optimizing for appearances. 75% of executives acknowledge their company’s AI strategy is more performative than practical—existing primarily to signal innovation rather than to provide meaningful business results. Much like offices that touted high-end photocopiers in the 1980s that nobody knew how to use, investments in this vein can end up costing way more than they’re worth.

Unlike underutilized photocopiers, the stakes of failing at AI adoption are high. Though the underlying challenges are nothing new, what is new is the scale of AI’s impact and the risk of falling behind competitors that get it right. (Yes, I recognize the irony of referencing photocopier technology while writing about AI.)

4. Excuses

When explaining why AI projects stall, there are sometimes excuses:

• The vendor overpromised
• We chose the wrong model
• The technology wasn’t mature enough
• Compliance and legal slowed us down
• We didn’t have the right talent
• The market changed

These concerns are valid but rarely insurmountable. Nearly every successful AI program has had to navigate some combination of imperfect circumstances. It’s important to treat these challenges as hurdles, not dead ends, and find ways around them by having a growth mindset culture and bringing in expertise where needed.

I’ve yet to see a project fail because leaders cared too much about communication, culture, alignment or commitment over the long-term. More often, the opposite is true. AI may be one of the most significant technological shifts of our lifetime, but success still depends on fundamentals: strong leadership, adaptable culture, clear objectives and a willingness to act.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?