The financial model looked promising. A junior consultant had used AI to assemble a first draft for an unfamiliar assignment. Then a senior colleague began asking about the assumptions. The consultant could present the result, but could not fully explain how it worked.
An entrepreneur had almost the reverse experience. He spent considerable time documenting how his business operated so that an AI system could help with customer proposals and follow-ups. Once it became useful, he found himself taking on more projects, not enjoying more leisure.
These are two of ten anonymised, interview-based accounts from AIGW Policy Observatory’s research into how people learn, work and make career decisions with AI. The participants include students, consultants, a professional working across China and the United States, a researcher and an entrepreneur. Their situations differ, but a tension runs through the stories: producing more is not necessarily the same as learning more, progressing further or gaining control over one’s working life.
The question matters well beyond these interviews. The World Economic Forum’s Future of Jobs Report 2025 found that 86% of surveyed employers expected AI and information-processing technologies to transform their businesses by 2030. OECD research on AI in the workplace likewise documents reported gains in performance alongside concerns about autonomy and work intensity. The studies provide context; the accounts below show some of the trade-offs up close.
When work gets easier, does learning get easier too?
1. A financial model its author could not explain
The junior consultant welcomed AI when a project required him to learn unfamiliar concepts. It helped him research terminology and prepare explanations. He also experimented with automating a routine status email—the kind of repetitive task he was happy to spend less time on.
The modelling assignment was different. AI helped him produce a convincing first draft before he understood the logic well enough to defend it. When a senior colleague pressed him on the assumptions, the gap became apparent. The difficulty was not necessarily the quality of the model; it was his ability to stand behind the work.
He still saw AI as a useful tutor. But getting an answer, understanding it and taking responsibility for it were no longer the same thing.
2. Better grades, less certainty about what was learned
A postgraduate student moving from furniture design into information technology has found new possibilities in AI. In part-time retail work, image tools can help customers imagine furniture in different colours or rooms. In their studies, AI can explain unfamiliar concepts or help with code.
Yet success raised questions. Preparing for an examination with AI-generated practice questions helped the student perform well, but the result did not necessarily mean the material had been mastered. When AI supplied code the student did not fully understand, debugging it became much harder.
Earlier design training offered a tangible comparison. “If I design a chair, I need to see whether it is actually stable and whether it is ergonomic,” the student explained (translated from Mandarin). Generating a persuasive image is one thing; knowing whether it can be built and used is another.
Approaching the labour market, the student could imagine several directions between design and technology. AI opened possibilities without deciding which one was worth pursuing.
3. The question that never reached the classroom
Another undergraduate described asking AI when a question arose in class instead of immediately raising his hand. It was quick and convenient, but sometimes the question was never brought into the room.
That could mean missing a discussion that would have helped other students too. In technical subjects, he also noticed that moving quickly through an explanation sometimes left foundations he needed later.
The same problem surfaced in research. AI could summarise material quickly, but the hardest part of a poorly documented question was still determining which sources deserved trust. He felt that AI had “largely levelled the advantage that comes from knowledge” (translated from Mandarin)—his interpretation of a shifting competitive landscape, not an established labour-market outcome.
Instant answers did not remove the slower work of developing judgment.
4. A junior consultant who wants the right tasks automated
A junior consultant had seen an internal AI tool speed up the first pass over large volumes of client material. Yet she still spent time checking folders and chasing missing documents. She wanted that repetitive, low-learning work automated.
What worried her was the possibility that firms might instead automate assignments through which beginners learn to detect mistakes, recognise patterns and develop professional judgment.
She had helped with a training exercise in which new associates produced polished presentations quickly with AI. The slides looked good; she was less certain the material had been examined carefully. She worried about losing the pattern recognition that develops through repeated experience.
Her enthusiasm for learning was equally clear: “As long as I’m learning, I’m happy.” The two observations belong together. She wanted AI to remove work that taught little, not the experiences that might prepare her for greater responsibility.
5. If routine entry-level work disappears, what replaces it?
A student with experience in nonprofit and project-support work had performed the kind of administration that often fills an early career: tracking information, preparing correspondence and helping projects run. He did not believe all of it deserved to survive automation.
Still, some apparently minor assignments create access. A junior colleague may learn by seeing how decisions are made, being trusted with a small responsibility or watching a more experienced person handle an unexpected problem. Remove the task without replacing that exposure, and the organization may inadvertently remove part of the path to more demanding work.
He wanted AI literacy to mean more than knowing how to prompt. A junior professional should be able to check a source, recognise an implausible answer and decide when a tool is not enough.
This dilemma also appears in AIGW’s The Missing First Rung and the expanded policy brief Designing the First Rung.
A faster workplace does not always mean a clearer career
The interviews also reached beyond classrooms and entry-level assignments. People already working with clients and colleagues were wondering what AI would do to experience, responsibility and the way careers advance.
6. A China-linked professional questions the value of experience
A professional coordinating technical customer requests across teams in China and the United States was interested in how AI might shorten slow exchanges. In his work, a customer question may need input from several people before anyone can offer a complete answer. AI could potentially classify inquiries, retrieve information and route requests faster.
But he drew a line between moving information and getting people to act. Someone still has to decide whether a reply is good enough, chase a missing answer, reconcile competing responses and take responsibility for what the customer receives.
His deeper uncertainty concerned experience itself. If AI can retrieve knowledge that once took years to accumulate, what advantage will an experienced worker retain? He could see the technology helping him connect ideas and learn faster, even while it made him question the future value of what he was learning.
He was not describing a role AI had already eliminated. His concern was what professional experience might be worth as information becomes easier to retrieve.
7. A broad job that stopped feeling developmental
A young professional moved from relatively narrow technical work into a role covering programming, coordination and communication. At first, that breadth was exactly what she wanted. She could see how different parts of a project connected, and AI helped her understand unfamiliar code rather than constantly waiting for someone else to explain it.
Over time, the experience became less positive. Competing demands and limited influence over the work made it difficult to turn that variety into sustained development. She eventually left the role because of concerns about management and work design—not because AI had replaced her.
She also disagreed with treating AI-assisted learning as inherently shallow. “If you ask AI about a problem but cannot turn what it tells you into your own knowledge, you’re just copying and pasting,” she argued (translated from Mandarin). What mattered was whether an explanation became understanding she could actually use.
A demanding assignment can build capability when it comes with feedback, ownership and time to reflect. Without those conditions, a broad job can simply mean being stretched thin.
8. The European consultant who finishes faster but stays at the same level
A consultant in Europe described work that had changed dramatically in pace. Tasks involving charts, research and presentation materials could sometimes be completed far faster with AI-assisted tools. Yet that speed did not automatically translate into a different title, more client responsibility or a redesigned job.
Finishing earlier often meant receiving the next assignment. Faster execution was useful, but it had not automatically altered his authority or eliminated the routine work still waiting to be done.
Outside his assigned tasks, the same tools offered a different kind of freedom. Without a traditional software-development background, he could prototype small applications and test his own ideas. “You have more agency to do things because you’re now able to build things that you previously can’t as a non-technical person,” he explained.
His experience contained both realities: a faster workload inside an institution and more freedom to build outside it.
Who decides what to do with the time AI saves?
For the final two participants, the important question was not merely how quickly a task could be completed. It was whether greater capacity translated into choices they could make themselves.
9. A researcher considers opportunities beyond a formal career
A university researcher involved in an independent project had spent time in different institutional and international settings. In a wide-ranging conversation about work and future choices, the researcher returned to the value of people, networks and projects pursued beyond a formal job description.
That perspective matters when thinking about AI-enabled opportunity. A new tool may make an independent experiment easier to attempt, but access to professional networks, time and the freedom to take financial risks still shapes what someone can do with it.
The researcher’s experience adds a question to the stories of faster tasks and new ventures: who has the room to pursue an opportunity in the first place?
10. An entrepreneur saves time—and starts more work
An entrepreneur in his early thirties had spent years developing the customer knowledge and routines behind his businesses. He began gathering that knowledge into documents and workflows an AI system could use to help prepare proposals and follow-up communications.
The result was not instant automation. Organizing the material required substantial effort: deciding what mattered, describing unwritten procedures and making his own preferences explicit. Once the system became useful, however, work that used to involve repeated assembly could move much faster.
He had expected efficiency. What surprised him was the effect on his workload. The new capacity made previously impractical projects feel possible, so he undertook more of them. He became busier.
His experience also reverses a fear heard elsewhere in the interviews. AI did not make his accumulated knowledge irrelevant. His understanding of customers, processes and quality made the technology much more useful. The difference was that he could decide how the resulting capacity would be used.
What changes when AI makes work faster?
There was no clean division between people who embraced AI and people who feared it. The consultant who struggled to explain a model still valued AI as a tutor. The student who worried about shallow learning also saw new possibilities in combining design and technology. The professional concerned about the value of experience used AI to deepen his own understanding. The entrepreneur who benefited from automation ended up working more.
A few distinctions are difficult to ignore. Output is not capability: someone may produce a convincing result before learning to defend it. Productivity is not progression: faster delivery may simply lead to more assignments. And access is not agency: people differ in how much control they have over the time and possibilities that AI creates.
These conversations cannot establish what young people as a whole think or how common these experiences are. They do show why someone can welcome AI while feeling uncertain about the job, skills or career path taking shape around them.
For employers and educators, this suggests a practical test: when a task becomes faster, what happens next? Does someone gain the understanding to check the result? Do they get a harder problem, useful feedback or meaningful responsibility? Do they have room to pursue an opportunity of their own? Those outcomes—not just the speed of the original task—will help determine whether AI becomes a shortcut through work or a route to becoming more capable.
Stay informed on AI governance, workforce transformation, and future-of-work policy
Policy briefs, event invitations, and analysis — delivered weekly.