AI Doesn’t Fail, Environments Do

Who’s to blame for these findings?

• Gartner projects that nearly one-third of all generative AI initiatives will be abandoned after proof-of-concept
• McKinsey’s “State of AI” report finds that the majority of enterprises say that their AI initiatives have not impacted earnings
• MIT says that almost all generative AI pilots are failing

Artificial intelligence skeptics might lay the blame on the technology. But the failure of an AI initiative in a corporation may have more to do with the environment of the organization than the technology itself.

The article below takes a look at what’s causing AI failure in so many organizations and what can be done about it.

Environmental causes of AI failures

Fragmented company systems

An organization with a set of fragmented systems is not AI-ready. Plenty of companies don’t know how many employees they have or how many contractors they have. They have multiple job descriptions for the same job. Their job titles differ for the same job. For example, they have “senior software engineers” and “developers” or “programmers” who have very similar tasks. Or, they have 100 people with the title of “engineer,” all with different tasks.

These are just examples of the many ways companies’ data is disorganized, unstructured, and confusing.

Wrong expectations about AI 

In many organizations, AI is sitting on the sidelines suggesting or scoring. People think of it as a co-pilot that can be turned on or off, or looked to for a suggestion when they wish. In successful organizations, AI sits alongside people. It prepares, coordinates, and accelerates decisions in the flow of work.

Poorly organized work structures

In many companies, work is still organized around job structures designed a century ago for stability, not speed. These structures are too static for modern business cycles and too coarse to reflect the real capability companies need.

Organizations full of silos

Skills, performance, mobility, and hiring all speak different “languages,” preventing a unified view of capability or the ability to make real-time decisions

Businesses moving faster than they can hire

Demand shifts in days or weeks, but talent processes still run on cycles measured in months. The consequences are economic: opportunity lost, innovation delayed, and growth constrained. Stores open later than planned due to workforce issues. Construction projects run long. Products miss their launch dates.

An environment for AI success

An unprepared corporate environment leads to a poor AI implementation. 

But there are companies that are “doing AI right,” laying the groundwork for successful AI for talent initiatives. Here’s a look at what they have in common.

They’re rethinking work and jobs. Everyone’s jobs are changing often, and job titles no longer have the meaning that they did a couple years ago. In other words, the job as a unit of work no longer reflects how work actually happens. Capability is the new organizing principle. Companies that are trying to organize work around tasks and skills are more ready for artificial intelligence than companies that are organizing themselves around jobs and job titles.

They’ve cleaned up their data. The leading-edge organizations are defining what it means to have customer service skills, to do engineering work, or to manage a product. They no longer have multiple job titles for the same job. They’re not making job descriptions on the fly. Their systems can be integrated and then no longer have overlapping and contradictory pieces of data. 

They’ve solved integration. As the SmartRecruiters’-sponsored AI Momentum Model research with Kyle Lagunas points out, “Integration connects systems, processes, and data, allowing AI to scale beyond isolated pilots. Without integration, AI remains fragmented — promising in pockets but unable to deliver enterprise impact. The hard lesson many early adopters have learned is that when HR systems are disconnected, or data is inconsistent, or processes vary, even the best tools remain stuck at pilot stage. Integration provides the connective tissue that allows AI insights to flow across functions, fueling scale and sustainability.”

They’re moving beyond the co-pilot mindset. There is a “shadow economy” where employees experiment with AI tools such as ChatGPT and become more efficient and effective. However, these improvements are uneven throughout these employees’ organizations. Ultimately, as those Gartner, McKinsey, and MIT studies mentioned above describe, the organization as a whole does not experience the payoff from AI.

The most effective organizations are redesigning their connective tissue at the core — the foundational layer where data, workflows, and decisions must move cleanly, continuously, and predictably. Employees are expecting to work in tandem with AI. Agentic AI becomes part of the workforce, not to replace people, but to expand what teams can deliver and accelerate how capability meets demand.

They’re emphasizing governance. As the AI Momentum Model research suggests, “Governance is often assumed to slow progress, but in practice it accelerates adoption.” 

That’s because guardrails give companies and leaders the confidence to move faster. They’re not stuck worried about risk or unsure if they are protecting data privacy, compliance with rules and regulations, or acting ethically. They know that they are. The AI Momentum Model says that “with strong, function-relevant governance, leaders gain a framework to test responsibly, scale confidently, and prove credibility to the enterprise.”

They’re investing in training employees. “It’s when organizations redesign work with AI at the core — and invest in learning — that they begin to see measurable results within a few quarters, not years,” Jennifer Ives, a leading AI market and growth strategist, tells SmartRecruiters. Ives says that “organization-wide investments in skilling and reskilling are not keeping pace with AI advancements.” 

As Pearson’s CEO recently said, “The pace and direction of progress will depend on how effectively we invest in human learning. Every positive scenario for an AI-enabled future is built on human development. Every negative one stems from neglecting it.”

Technology is not the stumbling block to AI success

Certainly, not every organization is finding success when they try to implement artificial intelligence in their enterprises. But the reason is typically not the fault of the technology. It’s their data. It’s the way they think about work, jobs, tasks, and skills. It’s a lack of governance and worries about risk slowing things down. And it’s the silos they operate in, with artificial walls between HR, recruiting, learning, and other departments.

All of these problems are solvable by laying the groundwork for AI success. When that happens, companies find that humans and AI agents are working together to Improve their efficiency, productivity, innovation, and drive actual success for their organizations