The End of "Own Your Own Development": Why AI Enablement Can't Be Self-Directed

Should a company leave AI training to its employees? No. AI is the first workplace technology in a generation where the hands-off approach corporate America has relied on for twenty years stops being a tolerable shortcut and becomes a genuine strategic liability.

Somewhere around the early 2000s, a quiet philosophy settled into corporate human resources: employees are responsible for their own development. Companies dressed it up as empowerment, pointed to digital learning libraries, and celebrated self-directed learners as model employees. What most of them were really doing was offloading the cost and complexity of workforce development onto individuals. For soft skills and routine upskilling, the world tolerated it. For artificial intelligence, it does not hold.

Proxigee Services helps organizations move from AI access to AI fluency, advising leadership teams on the strategy, governance, and enablement that turn AI from a collection of tools into a capability the company owns. The pattern we keep seeing is that AI is not a feature an employee can discover on a Saturday afternoon and apply on Monday morning. It is a shift in how work gets done, and it touches governance, data strategy, risk, and competitive positioning all at once. Leave it to individual initiative and the bill arrives in forms that are hard to unwind.

Why doesn't self-directed learning work for AI?

Self-directed learning fails for AI because the skill is contextual, fast-moving, and specific to your organization, and the aggregate evidence shows the consume-it-on-your-own model is already underperforming on far simpler skills. Josh Bersin's firm, drawing on data from more than 800 organizations, found that 74 percent of senior leaders believe their companies lack the skills to compete, even as employers collectively spend over $400 billion a year on corporate training.¹² The money is enormous. The results are not following it.

Bersin's explanation is that the prevailing model treats learning as something employees consume alone, disconnected from the workflows where the skill gaps actually live. His research found that companies practicing what he calls dynamic enablement, structured and employer-led learning embedded in how work happens, are six times more likely to exceed their financial targets and twenty-eight times more likely to see their people reach their potential.¹ Fewer than five percent of organizations have built it. When the capability at stake is AI fluency, a rapidly evolving and organizationally specific competency, the distance between what is possible and what most companies are doing turns dangerous.

Why can't AI fluency be built one employee at a time?

AI fluency cannot be assembled from individual effort because the risks of getting it wrong are organizational rather than personal, and they show up in three specific ways. When an employee independently earns a certification in financial modeling, the learning is self-contained and the company bears little risk from the process itself. AI breaks that pattern.

First, AI fluency is inseparable from organizational context. Knowing how to prompt a general-purpose tool is not the same as knowing how to use AI inside your company's workflows, data environment, governance policies, and risk tolerance. An employee learning on their own can develop habits that quietly conflict with how the organization needs AI to work. McKinsey makes this explicit, noting that frontline workers, managers, and domain experts each require fundamentally different AI learning experiences, so generic literacy training misses most of what matters.³

Second, without structured guidance, AI use drifts toward overreliance. Gartner predicts that by 2030, 30 percent of enterprises will see their decision-making quality decline because employees lean on AI outputs instead of applying their own judgment.⁵ Gartner separately expects the atrophy of critical-thinking skills from generative AI use to push half of global organizations to require AI-free skills assessments through 2026, testing whether people can still think when the machine is out of the room.¹⁰ Critical thinking and AI fluency have to develop together, which takes deliberate design, not individual discovery.

Third, the stakes of misaligned AI use are systemic. When one employee underuses a CRM, the consequence is local and correctable. When people across a function prompt inconsistently, push unvetted outputs into customer-facing decisions, or feed sensitive data to outside tools without grasping the implications, the exposure is enterprise-wide. Governance cannot be crowd-sourced from individual initiative. It has to come from the organization and be built into how employees are trained from day one.

What does the research say companies should do instead?

The research points to one answer: treat AI enablement as a structured, employer-led investment built into the flow of work, not a catalog of courses employees browse on their own. McKinsey lays out a three-tier framework. The first tier is AI literacy, a shared baseline of fluency that reduces fear and creates a common language for AI decisions. The second is AI adoption, embedding tools and behaviors into core workflows by redesigning roles, processes, and incentives. The third is AI domain transformation, where technical and domain experts build function-specific use cases that extend competitive advantage.³

Each tier demands intentional design, executive sponsorship, and a tie to business strategy. McKinsey reports that organizations with rigorous human capital development practices are 4.2 times more likely to outperform their peers financially, and that productivity can fall by as much as 22 percent when employees lack the right skills.⁴ Gartner adds the talent angle: it predicts that by 2027, half of enterprises without a people-centric AI strategy will lose their top AI talent to competitors who have one.⁶ Workers in roles requiring AI skills already command an average wage premium of 56 percent over comparable roles that do not, according to PwC's 2025 Global AI Jobs Barometer.⁷ A company that will not cultivate those skills internally simply loses the people who build them on their own.

Is AI training an HR program or a boardroom decision?

AI enablement is a boardroom decision, because it sits at the intersection of technology, talent, risk, and competitive positioning, and none of those belong to HR alone. Defining acceptable use, setting data-handling standards, and deciding where human judgment outranks the machine are organizational choices that have to be made deliberately and then built into how people are onboarded to AI tools. The business strategy for AI and the workforce strategy for AI must be developed together, not in separate rooms on separate timelines.

"AI enablement is not a training program. It is a strategic function, and the conversation belongs in the boardroom, not just in human resources."

Gartner's work on CHRO priorities for 2026 names three tactics for building the human capabilities AI cannot replace: peer learning channels that move tacit knowledge between colleagues, formal infrastructure to transfer expertise from experienced to emerging talent, and AI simulators that let employees practice judgment in high-stakes situations.⁸ None of those appear on their own. McKinsey's recommended shifts run the same direction, from episodic training toward continuous development in the flow of work, with skill-building measured as rigorously as financial performance.⁴ The common thread is that AI enablement requires active organizational ownership.

What has thirty years taught us about enablement that actually works?

Thirty years in the field taught me something the surveys only gesture at: when the subject is complex, changes how a team works together, and moves the culture, coordinated enablement beats self-directed learning every time. I have watched it from the inside, and some of the most valuable training of my career was formal, company-paid, and conducted in a room with my colleagues.

A leadership retreat with my peers at FreeMarkets more than twenty years ago built relationships I still treasure today. I still lean on that kind of structured, in-person development for skills like influence and navigating international business cultures, because those are learned with other people, not from a screen alone. When SmartOps needed the world to understand what inventory optimization was, we did not send everyone a link. The entire company learned it together, the same material, in the same room, alongside the people they would have to explain it to. Later, when new market conditions demanded a higher and more consistent standard of presale leadership at SAP, we gathered from around the world to hear it, consider it, and debate the way forward together. The shared understanding that came out of those rooms could not have been downloaded.

AI checks every one of those boxes. It is complex. It changes how teams interact. It will move culture. That is exactly the territory where individual discovery breaks down and coordinated enablement compounds. The parts of AI fluency that matter most, when to trust the machine, when human judgment has to lead, and how a team agrees to use it, are learned the same way we learned to sell inventory optimization: together, on purpose, with the organization standing behind it.

What does building real AI capability require?

Building real AI capability requires people who know how to develop and tend it, which is why enablement is inseparable from the larger project of owning your intelligence rather than renting it. We have made this argument from several angles. In Stop Renting Intelligence we laid out why treating AI like a subscription leaves a company with no lasting capability and no differentiation. In The Sovereign AI Operating System we showed why governments spending $100 billion to own their AI infrastructure are applying the same logic every company should. In Agents Are Not Enough we argued that an agent is only as valuable as the intelligence it carries.

The workforce is where all of those arguments converge, and it is where they meet the human stakes we explored in Maximizing Human Potential with AI: the goal is to enhance human judgment, not retire it. Building an Intelligence Estate takes people who understand the difference between prompting a borrowed model and training intelligence on your company's own rhythms, failure modes, and institutional knowledge. Self-directed learning produces individuals who can use AI. Company-led enablement, wired into your governance and strategy, produces teams that can build AI that belongs to the organization and grows more valuable over time. That capability cannot be outsourced to individual initiative any more than the estate itself can be rented. It is the same conviction behind how Proxigee helps companies take their first strategic steps into AI ownership.

When does a company need to act on this?

The time to build is now, because the gap between organizations deliberately building AI fluency and those waiting for it to appear is widening on a four-year clock. The World Economic Forum's Future of Jobs Report 2025 projects that 59 percent of the global workforce will need training or retraining by 2030.⁹ That is not a slow demographic drift. It is a deadline.

The era of own-your-own-development made sense when the cost of underdevelopment was mostly personal. AI changes that arithmetic. The cost of an underdeveloped workforce in the age of intelligent systems lands on the company: competitive capability lost, decision quality degraded, risk exposure widened, and talent walking out the door. The organizations positioned well in 2030 are the ones building structured, governance-aligned AI enablement now and managing workforce fluency as the strategic asset it has become. The window to build something differentiated is open. History says it will not stay that way.