AI Does Not Excuse Us From Developing People

I wrote once that American managers were spoiled. I had just come home from Reykjavik, where a client on a cold island with a tiny labor market was outperforming companies that had every advantage we take for granted. I thought the lesson was about geography and scarcity. It was. It is also about what happens when a market has so many people, and now so many machines, that leaders stop doing the one job that cannot be automated: developing the next person who can lead.

The gap that should worry you is not artificial intelligence (AI) versus humans. It is experienced people who already know the work and now have AI as leverage, versus new graduates and junior employees who used to learn that work by doing the mundane, repeatable tasks that AI now finishes before lunch. One group is compounding. The other is watching the on-ramp disappear. If you lead a team, that is your problem. If you are early in your career, it is your opening, but only if you stop treating AI as the thing that will take your job and start using it as the thing that will get you the job.

The Island That Could Not Afford Bad Managers

Imagine managing on an island. Your island has limited human resources, sits hours away from anything, and speaks a language the rest of the world does not. Attracting employees is hard. The climate is cold and damp. Some months barely deliver daylight. Your people grow up learning two other languages because theirs is little known off the island. You cannot casually replace anyone. You invest heavily in salary and in resources. Each employee starts with five weeks of vacation a year. You cannot promote people on a whim, because turnover and resentment would bury the business. Your job is to move people into positions where they can succeed, and to keep them, even though they can leave whenever they want.

The island is Iceland. The city is Reykjavik. 20 years ago, my company sent me there to help a client. From the moment we landed, the organization’s culture, professionalism, and the people’s dedication were unlike any American company I had worked with. Fishing and tourism were the top industries. The island is rich in glaciers, geysers, and lava fields, and the people are rich in culture and history. They also ran a profitable business in an international industry that was floundering almost everywhere else.

They faced far more constraints than we did. Their competitors in our country could go bankrupt as part of a strategic plan. They focused on quality while competitors focused on price. They held long-term strategies while competitors watched today’s stock price. Their livelihood required it, and they delivered. In meetings with dozens of employees, we could not tell who were front-line and who were senior managers. All were knowledgeable, committed, vocal, and engaged.

That is what scarcity does when it is paired with responsibility. You cannot hide a weak manager behind a recruiting pipeline. You cannot treat tenure as a liability. You cannot skip development because someone else will apply next week. In Iceland, the environment demanded better marketers, better business people, and better managers.

In America, we had so much talent in the next cubicle, the next city, the next graduating class that we often didn’t have to manage at all. We changed the environment to mask an inability to lead. In some businesses, turnover even looked like an advantage because it kept pay down. Some leaders still believe it is cheaper to get a new employee than to keep an experienced one.

The Talent Pool Got Bigger, Then AI Arrived

Two decades later, American managers are even more spoiled, not less. The pool of people you can hire, contract, offshore, or staff through a marketplace is larger than anything I saw in 2005. On top of that pool now sits automation and AI. The repetitive work that used to justify a junior seat, a trainee rotation, or an analyst’s first two years now gets drafted, summarized, cleaned, scheduled, and QA’d by a model that does not need health insurance. If you did not like how someone worked in 2005, you could make it uncomfortable until they left. In 2026, you can point at a dashboard and say the tool will cover it.

That is a hell of a luxury. It is also a trap. Productivity in this country has always been famous. I didn’t believe then, and I don’t believe now, that it came primarily from great managers. It came from a vast resource of human talent. When the resource is vast, you can get away with not leading. You can target long-tenured people for their weaknesses instead of treating longevity as an asset. You can skip the slow work of teaching judgment because the next resume is already in the inbox.

AI did not invent that habit. It supercharged it. Experienced operators who already understand the customer, the offer, the data, and the company’s politics can now move at a pace that looks unfair next to someone still learning what a good brief even is. The senior marketer who knows which claim will get legal’s red pen can prompt a model, kill the weak variants, and ship. The new graduate hired to support the team by pulling reports, formatting decks, and writing first-pass copy is competing with a system that does those tasks in seconds and has no instinct for when the answer is wrong.

The result is a widening productivity gap inside the same org chart. It is not a morality play about who deserves a job. It is a skills-and-leverage problem. People who already had context got a force multiplier. People who were supposed to earn context by doing the grunt work lost the grunt work, and with it the classroom. If you are a leader and you treat that as a cost-savings story, you will get a quieter office and a thinner bench. If you are early in your career and you treat it as proof that the ladder is gone, you will wait for a role that looks like 2012 and wonder why nobody is hiring for it.

The Trainee Ladder Is Breaking

For a long time, companies hired inexperienced people to do necessary, boring, educational work. You learned how the CRM actually behaved by cleaning lists. You learned how a campaign was built by assembling the first draft of the brief. You learned how a manager thought by sitting in the room while they tore apart your slide. None of that felt like leadership development. It was. It was how judgment built muscle.

AI is excellent at the boring layer. That is the point. It drafts the email, clusters the comments, writes the SQL, builds the first version of the dashboard, and turns a transcript into bullets. Used well, that is a gift. Used as a reason to stop hiring and stop teaching, it freezes an organization at the skill level of the people who already know how to prompt. The juniors who remain get fewer reps, not more, because the reps moved into a chat window the senior never shares.

You can see the distortion in hiring. Job posts still ask for three to five years of experience in tools a graduate has never been allowed to touch. Interviews still test for polish on tasks a model can fake. Meanwhile, the people who would have grown into those roles are told the entry work no longer exists. Both sides are telling a partial truth. The entry work still exists. It changed shape. The person who used to spend a week on a competitive teardown now needs to spend a day directing AI through that teardown, then a week on the part the model cannot do: what it means for this brand, this quarter, this sales motion, this legal constraint.

That second week is the job. The first week was the apprenticeship. If you remove the apprenticeship and never replace it with a deliberate one, you do not get a leaner, smarter company. You get seniors who are exhausted because they are doing their job plus the teaching nobody scheduled, and juniors who cannot get into the work because nobody will let them near a live system until they already know it.

What Leaders Must Do Now

Leaders, you cannot all be replaced by AI and robots, and you should stop talking as if that were the plan. A model does not sit with a frightened customer. It does not smell a bad vendor deal. It does not notice that the high performer is about to leave because you used them as a utility instead of a future executive. It does not walk a new hire through why last quarter’s win was luck. Those are human jobs. If you stop doing them because the spreadsheet looks cleaner without the trainee budget, you are not becoming an AI-native company. You are eating your seed corn.

In 2005 I wrote that we do not need to manage; we need to lead and view company longevity as an asset. That sentence has aged too well. The Icelandic client could not afford that posture. Their people could leave the island, but replacing them was painful, so managers had to develop them. We have the opposite problem. We have a massive pool of resources, and now we have AI and automation to assume some of the roles we used to staff with trainees. That makes it easier than ever to skip development. It does not make it wiser.

A corporation that does not take the time to develop its personnel will die a slow death. Not because it adopted AI. Not because it automated a reporting queue. It will die because it is no longer developing fresh, dynamic, and innovative leaders. The org chart will still have titles.

The work will still ship, for a while, because experienced people plus models can cover a shocking amount of surface area. Then those people will leave, retire, or burn out, and nobody behind them will have been trusted with a real decision. You cannot hire your way out of a pipeline you refused to build. The companies that win will look, from the outside, almost like that Reykjavik conference room: you will not be able to tell the front line from senior leadership, because everyone in the room will be knowledgeable, committed, vocal, and engaged.

That requires a different definition of productivity. If your only metric is output per head, you will fire the people who are learning and keep the people who are already fast. That improves this quarter and wrecks the next three years. Pair every AI rollout with a teaching obligation. If a senior uses a model to do in an hour what used to take a junior three days, the senior owes the junior a walkthrough of the prompts, the failures, and the judgment calls. Put that on the calendar. Review it the way you review pipeline. If you cannot explain how a new hire becomes dangerous in twelve months in a world where the old grunt work is gone, you do not have a talent strategy. You have a headcount target.

Cutting people to fund tools is the spoiled-manager move with a new logo on the slide. You have more leverage than any generation of American managers before you. You have a deep labor market, global contractors, software that never sleeps, and models that draft. If with all of that you still cannot grow people, the constraint is you.

Cutting headcount is not a growth strategy. Unlocking talent is.

Alex Atzberger, CEO, Optimizely, Camp Opticon keynote

That line from Optimizely Chief Executive Officer (CEO) Alex Atzberger at the Camp Opticon keynote is the whole argument in two sentences. Unlocking talent is slower than a layoff announcement. It is also the only way a company stays alive after the people who already know how to use AI get a better offer. Optimizely’s point was about growth. Mine is about succession. If you unlock talent, AI becomes a multiplier for a bench you actually have. If you only cut, AI becomes a mask for a bench you hollowed out.

What New Graduates and Early-Career Employees Must Do

If you are a new graduate or in a role that used to be the on-the-job trainee seat, the lecture is just as blunt. Nobody is coming to reconstruct 2014 for you. Systems that don’t need a parking pass are doing the mundane, repeatable tasks that used to be your curriculum. Waiting for a manager to give you real work while you refuse to touch AI is how you stay unhireable. The market is not required to preserve a learning path that no longer matches the work.

Stop thinking that AI will take your job. Start using AI to get your job. That is not a motivational poster. It is the new apprenticeship. The junior who can show how they directed a model, caught its mistakes, and turned the output into something a customer could see is already more useful than the junior who is proud of never using the tools. You do not become worth hiring by protecting the old chores. You become worth hiring by learning faster than the chore list shrinks.

That means you have to build the judgment the model lacks. Use AI to draft, then spend your hours on the part that used to come after the draft: why this audience, why this offer, why this risk, why this week. Ask to sit in on the reviews where work gets killed. Keep a log of failed prompts and why. Learn the systems of record, not just the chat box. If your company will not let you near production data, build a sandbox. If your manager will not teach, find one who will, inside or outside the company. Loyalty to a leader who refuses to develop you is not a virtue.

You should also be honest about the fear. It is rational to look at a tool that does your first-year tasks and feel erased. Erasure is what happens if you compete with the tool on the tool’s terms. Do not. Compete on taste, context, relationships, and learning speed. Those still compound. An experienced worker with AI is formidable because they had years of reps before the model showed up. You will not get those same reps unless you manufacture them. AI is how you manufacture them. Run more scenarios. Critique more drafts. Learn adjacent skills that used to be not your job. People who treat the model as a senior intern will outpace those who treat it as an enemy.

None of this lets leaders off the hook. A twenty-two-year-old cannot invent a development culture alone. But you are not powerless. Every company that is about to learn, the hard way, that it still needs humans will hire the humans who already know how to work with machines. Be that person before they post the role.

Development Is the Strategy

In my career, I have met a handful of managers who could compete in that Icelandic environment, where you cannot hide, and you cannot churn your way to a good quarter. Most of the rest, including me on a bad day, were products of surplus. We had people to spare. We had applicants in the hall. We had, and still have, a global marketplace willing to send talent to our door. AI is the latest surplus. It feels like power. It is power. Spoiled managers will use it to avoid the discomfort of teaching. Good ones will use it to give more people real shots at harder work, sooner.

The original piece ended with a warning that our island was getting smaller, that America was becoming a customer of the global marketplace, and that we would not succeed unless we held managers accountable. That was true when the pressure was cheaper labor overseas. It is truer when the pressure is a model that never sleeps. Good employees and good managers will still be taken by companies that value their worth. Bad managers will still ride bad companies into the ground. The difference is speed. A weak bench used to take years to reveal itself. With AI covering the cracks, it can look healthy until the experienced people walk out and the models have nobody left to tell them they are wrong.

The top software developers are more productive than average software developers not by a factor of 10X or 100X, or even 1,000X, but 10,000X.

Nathan Myhrvold, former Chief Scientist, Microsoft

I was fairly certain in 2005 that this could happen in most organizations. I am certain now. Good employees are worth more than other employees. They are worth exponentially more, and AI stretches that exponent. That is exactly why you develop people instead of discarding them, and why you become one of the people who can use the tools rather than a spectator who hopes the tools stay illegal.

Iceland had to be good because the island gave them no other option. We have every option: a deep talent pool, automation, and AI that can assume work we used to hand to trainees. That is the spoiling. The companies that survive it will be the ones that still choose the hard path, developing personnel as if they lived on an island, because in the only way that matters, they do. The next generation of leaders will not appear from a vendor’s roadmap. They will appear because someone sat with them, handed them real work, taught them to use the new leverage, and refused to confuse a smaller headcount with a growth strategy.

Unlock the talent you already have. Hire the people who are trying to get in, and teach them. If you are trying to get in, pick up the tools and make yourself expensive to ignore. Cutting headcount is not a growth strategy. Unlocking talent is. Robots won’t save a company that forgot how to grow people, and they will not hire a graduate who refused to grow.