Soon after ChatGPT became publicly available, a senior vice president walked into his CEO’s office and said he needed to resign. The technology felt too significant to learn in the margins of an already demanding leadership role, and he did not want to be left behind.
The CEO offered another path. Step away from management. Spend meaningful time learning the technology. Then return and teach the rest of the organization. The vice president accepted. Today, he leads AI implementation for the company.
It is the kind of story people often share as evidence that organizations are adapting. Ten years ago, a senior leader voluntarily moving out of management and into an individual-contributor role might have been interpreted as a step backward. Today, it can look like strategic foresight.
But consider what had to happen for that story to work. A senior executive who recognized a genuine capability gap still needed permission from his CEO to address it. Had the answer been no, the likely outcome would have been resignation rather than reinvention.
The new capability depended on someone with authority noticing the need, valuing the person and trusting his judgment at exactly the right moment. Remove that individual from the conversation, and the outcome could have been entirely different.
That is the distinction between infrastructure an employer grants and infrastructure an individual owns. Granted infrastructure depends on another person’s support. Owned infrastructure stays with someone regardless of the employer, manager or decision-maker in the room.
Behind every headline about AI and jobs sits a more important question: Which kind of infrastructure do most people actually possess today? The honest answer is very little.
Every period of economic change asks people to adapt. This one is asking them to reinvent their skills, careers and professional identities at a speed previous generations did not face, while offering most people too little time, funding, protection or support to do it on their own terms.
AI Is Redefining What Career Security Means
For most of the past century, the employment bargain was relatively clear. Organizations created jobs and people filled them. In exchange for their time, expertise and loyalty, workers received income, structure, protection and some expectation of progress.
Governments built labor policy around that relationship. Education systems prepared people to enter it. Careers were expected to develop within it. Benefits, status and even professional identity were often tied directly to it.
That model is now being challenged from two directions at the same time.
The first source of pressure is coming from workers themselves. Since the pandemic, more people have reconsidered what they are willing to exchange for a paycheck. They still want income and meaningful work, but they also want greater influence over their time, energy, learning, identity and the shape of their lives. Remote work was simply the most visible expression of that change. The deeper shift is that people began seeing every workday as a portion of life being spent, and many are no longer willing to spend it without questioning the return.
The second source of pressure is AI. Inside organizations, AI is often positioned as a productivity technology: a way to accomplish more with fewer people. For individuals, however, it represents something broader. It lowers the barrier to creating value without waiting for an employer to provide the role, budget, team or approval. A person with access to the right tools can now write, analyze, design, code, market, research, teach, advise, build and sell in ways that once required the resources of an organization.
What separates this period from earlier waves of workplace change is the tension beneath it. AI shortens the time between a skill becoming valuable and becoming outdated. At the same time, it gives individuals access to tools that once belonged almost exclusively to companies. Capabilities expire faster than before, while people increasingly have the means to continue developing without relying entirely on an employer’s infrastructure.
Taken together, those two forces suggest that the anxiety surrounding AI is not only fear of job loss. It is fear of dependence. Dependence on an employer to decide whether a skill still matters. Dependence on the training and subscriptions a company is willing to fund. Dependence on titles, organizational structures and gatekeepers to confirm a person’s value.
Employability Alone Is No Longer A Sufficient Goal
This is where the language of career security needs to evolve. Conversations about AI and careers still focus heavily on employability, as though the main objective is to remain appealing to an employer. The broader goal should be personal infrastructure: AI literacy, consistent learning habits, professional networks, a portable reputation, financial breathing room and benefits that do not disappear when someone changes how they work.
In an AI-shaped labor market, these are no longer optional career advantages. They are what allow people to continue generating value regardless of how they are paid.
Adaptability cannot be treated only as a personal quality. It must become part of the infrastructure surrounding work. Learning a new capability takes time. Experimentation requires tools and psychological safety. Opportunity often comes through networks. Credibility depends on proof of capability that can move across roles and industries. People also need enough financial stability to take a calculated risk before a crisis forces them to act, or before the right CEO happens to approve the move.
Personal infrastructure is what an individual requires to remain economically viable. Social infrastructure is what makes that possible at scale: public policy, employer practices, education systems, portable benefits and financial support for career transitions. One cannot function effectively without the other.
AI Literacy Should Be Owned By The Individual
For individuals, personal infrastructure means treating access to AI tools, dedicated learning time, professional relationships, visible evidence of capability and financial runway as essential parts of career security rather than optional extras.
For policymakers, AI literacy is an obvious place to begin building the social infrastructure around that need. It should be treated in the same category as reading, writing and mathematics: a foundational economic capability, not an employee benefit granted by a company.
When responsibility is left entirely to employers, AI training will naturally focus on what the business needs from a person today, rather than what that person may need to remain economically relevant over the next decade.
Singapore offers one useful design principle: fund the individual rather than the job. Its SkillsFuture Credit provides every citizen aged 25 and older with a personal training account that belongs to them, not to their employer. A company cannot redirect the funding toward its own priorities, and the credit remains with the person whether they are employed, self-employed or between opportunities. Learning stops being something an individual waits for an employer to authorize and becomes something they can manage directly.
Time is just as important as money. People cannot easily rebuild careers using only the exhausted hours left after a full workday, family responsibilities and everything else life requires. If society expects people to adapt continuously, then adaptation itself needs to become part of the infrastructure, just as education, healthcare, retirement support and paid leave became part of the social contract when earlier generations recognized that markets alone could not provide equal access to opportunity.
That might include transition accounts people can draw on between jobs, publicly funded career-reinvention credits or a new form of re-education leave modeled on parental leave or paid time off: protected time away from work for a purpose society has decided is valuable enough to support.
People often change faster than organizations, and organizations usually change faster than policy. That gap has consequences. Social infrastructure, new funding models, portable benefits and redesigned safety nets may take years to catch up with what individuals are already doing. People cannot afford to postpone building their AI literacy, professional networks, visible evidence of capability and financial resilience until policy is ready. But policy will determine whether that infrastructure becomes widely available or remains a privilege reserved for those who already have enough time, money and confidence to reinvent themselves.
Return to the vice president who nearly resigned. His story ended well, but it should not require a fortunate conversation with an unusually supportive CEO. AI is giving people a clearer view of economic agency beyond the traditional employment bargain. Once that becomes visible, the conversation shifts from jobs to capability, from employment to earning, from training to reinvention and from protection attached to status to protection designed around the person.
The future of work requires systems that help people remain economically secure with or without a traditional employer. Reinvention should not depend on another person’s permission or on being fortunate enough to have the right conversation at exactly the right time. That is a much larger issue than employability, and it requires a much more ambitious response than most current workforce-policy debates provide.
