Reinvention Roadmap
Recently, a C-suite executive I respect told me he had stopped hiring junior analysts. AI does the first pass faster, he said, and the output is cleaner. I understood his logic completely but I also couldn't stop thinking about the implications of what he was actually describing.
The real risk of AI for the next generation is not losing their first jobs. It is losing the experiences inside those jobs that teach them their craft.
Young people understand this well. More than half of Americans under 30 are now more concerned than excited about AI, and nearly three-quarters believe it will reduce jobs over the next 20 years, according to a 2026 Pew Research survey. Gallup found a parallel tension: Gen Z uses AI frequently, but many worry that leaning on it too heavily makes real learning harder.
Their anxiety is well-founded. Most experienced professionals encountered AI after years of building expertise. We had already written the bad first drafts, built the clumsy models, sat through the difficult meetings, and slowly developed the judgment that comes from living with our own mistakes.
Young people are meeting AI at the very beginning of that process, just as the technology has become capable of doing much of the work through which beginners used to learn.
Entry-level work has always served two functions at once: producing useful output for the employer, and developing the person doing it. A junior analyst building a model was learning what mattered. A young lawyer researching cases was learning to recognize patterns. An engineer debugging code was building intuition about how systems fail. Production and development used to happen together, almost automatically.
AI is pulling them apart. If a manager can get the first draft, the analysis, or the research faster from a machine, using it makes complete sense. But if fewer people do the work that once made them experienced, we eventually face a harder question: where will the experienced people come from?
The answer is not to preserve every tedious task in the name of tradition. Plenty of grunt work should disappear. But we have to be deliberate about what actually builds capability. AI can give a young person access to more scenarios, feedback, and expert examples than any previous generation could have imagined; some experience genuinely can be compressed. Trust, responsibility, and judgment under real stakes still have to be earned.
The opportunity is to redesign apprenticeship around that distinction:
Compress the experience that can be borrowed. Increase access to the experience that must be lived.
Companies never had to separate getting work done from developing the people doing it. Those two things happened together. The analyst built the model and learned what mattered. The associate did the research and learned to distinguish signal from noise. The correction from a more experienced colleague was both a quality check and a lesson.
AI separates them. An organization may no longer need a junior person to produce the first draft or analysis, but it will still need someone with the judgment that used to develop through doing exactly that work. That means capability formation can no longer be left to chance. If the work changes, the path to expertise has to change with it.
The old apprenticeship model was limited by whatever experience happened to come your way: the clients you served, the cases you saw, the manager you happened to work for. AI removes that ceiling. A young professional can now work through more scenarios, study expert examples, and practice difficult situations before encountering them in real life. Some repetitions that once took years to accumulate can become available much sooner.
What AI cannot make abundant is real stakes. Exposure to patterns may become cheap. Trust and responsibility will not.
Traditional career ladders used time as a proxy for readiness. Spend enough years doing junior work and eventually you were trusted with consequential decisions. That logic weakens when AI allows people to build pattern recognition much faster. Someone who can demonstrate sound judgment through practice and supervised work should not have to wait the same number of years.
But the progression still has to lead somewhere real. At some point, a person has to make the recommendation, own the decision, and live with the outcome. That is where judgment becomes more than understanding. A better apprenticeship model would use AI to shorten the road to responsibility, not remove responsibility from the road.
Young people will grow up with AI as a normal part of how they learn, create, and work. Trying to keep it out of their lives is neither realistic nor useful. The more important task is helping them notice the difference between using AI to strengthen their own thinking and using it to avoid thinking altogether.
That distinction is subtle but consequential. Asking AI to challenge an argument you have already formed can deepen your reasoning. Asking it to form the argument for you short-circuits the very process you need to practice. The same applies to writing, problem-solving, and decision-making.
The better conversation for parents is less about policing use and more about building self-awareness. One question cuts through the noise: What can you do or understand now that you couldn't before? If AI helped them get there, great. If it only helped them finish faster, that is useful, but it is not the same as learning.
Schools and universities are still spending too much energy debating whether AI should be allowed. That question is becoming irrelevant as the technology embeds itself into everyday work.
The more important question is what the assignment is actually for. If the goal is to teach a student to build an argument, the student still needs to practice building one. If the goal is to develop judgment, they need situations where they have to make and defend a choice. When AI can produce the essay, the code, or the analysis, the artifact stops being a reliable signal of what the student actually knows how to do.
Some old assignments should disappear. Others should be redesigned. But the developmental purpose has to stay clear: What capability are we trying to build, and what practice does that capability require? That question should drive every decision about how AI is used in education.
Companies face a harder version of this problem because the economic incentive to automate junior work is real. In many cases, they should automate it. There is little reason to preserve tedious tasks simply because previous generations had to do them.
The risk is that some work looks low-value only after you already know how to do it. A senior executive may look at a first-pass analysis and see something that AI can produce in seconds. What they may forget is how much they learned from doing that analysis themselves years earlier. Reviewing cases builds pattern recognition. Writing the first draft teaches you what matters. Sitting in difficult meetings teaches you how decisions really get made.
Before removing a task, employers need to ask what the person doing it was learning along the way. If the answer is "not much," automate it. If the task was building judgment, intuition, context, or relationships, then the learning needs another home.
The new apprenticeship model asks employers to stop assuming that experience will accumulate automatically through time served. Instead, employers must start using AI to move young professionals into better experiences sooner.
The test is simple: Does AI leave the next generation with less experience, or with better experience?
Before automating a task performed by someone early in their career, run it through this prompt:
I am considering using AI to automate or substantially reduce the following task:
Task: [describe the task]
Role performing it today: [role]
What AI would do instead: [describe proposed AI workflow]
Analyze this decision from two perspectives:
1. Production value: What useful output does this task produce, and which parts can AI perform more efficiently?
2. Developmental value: What capabilities might a person currently develop by doing this task? Consider domain knowledge, pattern recognition, judgment, problem framing, critical thinking, communication, relationships, accountability, and understanding consequences.
Then classify each part of the experience as:
AUTOMATE — little meaningful developmental value is lost.
AUGMENT — AI should assist, but the person should still perform enough of the work to develop the underlying capability.
PRESERVE — direct human experience is important because the capability depends on judgment, responsibility, relationships, or real-world consequences.
For anything classified AUGMENT or PRESERVE, recommend a better way to develop that capability if the original task is automated.
Finally, answer:
If AI does this work now, where will the human capability this work used to develop come from?
Run it on one entry-level role in your organization. The goal is to make sure we don't automate the pathway to expertise along with the work.
The Skill Code by Matt Beane
Matt Beane spent years studying how people actually become skilled when intelligent machines enter the workplace, including environments as different as robotic surgery and warehousing. What makes his work especially relevant to this conversation is his focus on what can disappear when technology makes experts more productive.
Beane identifies challenge, complexity, and connection as important conditions for developing skill. Novices need difficult work, enough exposure to understand the messiness behind the textbook version, and proximity to people who already know what good looks like. When technology allows experts to accomplish more on their own, those opportunities can shrink.
The Skill Code is useful because it explores how to still develop capabilities when the work itself changes.
Changing Higher Ed with Dr. Drumm McNaughton
My conversation with Dr. Drumm McNaughton on Changing Higher Ed feels particularly relevant to this issue because universities are actively confronting the apprenticeship problem.
We talked about the changing value of a degree, and why universities may need to place much greater emphasis on mentorship, judgment, synthesis, ethics, and learning through real problems. The value of education cannot rest primarily on transferring information or producing artifacts that AI can already create.
The opportunity for higher education is to rethink its role around the experiences and capabilities students will still need when they enter the workforce. That makes this conversation a useful companion to the apprenticeship question: how do we prepare young people for a world of work that is changing before they have even had the chance to enter it?
Watch the episode: Changing Higher Ed with Dr. Drumm McNaughton
What experiences did you need to become good at what you do, and which of those would still exist for someone starting today?
Until next time...stay curious!
Cheers,
Nikki
© 2026 NikkiBarua.com
Privacy Policy