In this article, I explore the idea that while AI feels like a kind of modern magic, we are misdirecting it onto low-value tasks when we should be questioning why those tasks exist at all. Drawing on the metaphor of the Sorcerer’s Apprentice, I argue that organisations are mistaking efficiency for progress, scaling busywork instead of reducing it. I address the current idea that AI will replace workers by exploring how entry-level roles are being reshaped. I argue that businesses should refocus on meaningful, high-impact activity and show greater leadership in how they design roles and develop people.

---

The Magic

AI is magic. It doesn’t understand us, and it isn’t always right, but it has still handed us a magic wand that is transforming how we work.

The tricky, lengthy, delicate emails which used to take 2 hours now take half an hour. The introduction, background and summary sections of a report which used to take an hour to write now takes 15 minutes. The distillation and extraction of key themes and requirements from a 100 page document or 6 month email trail, which used to drive you to your knees, now happens while you make a cup of tea.

Even more, it can create shared task boards, tracking who has seen what, logging decisions, and prompting people to participate democratically. Another set of tools can construct detailed planning systems for your day, breaking intentions into structured workflows, scheduling, rescheduling, tagging, and monitoring progress.

It’s very cool but doesn’t mean we are doing less work. Instead, this type of work (what we call “admin”) is a magical curse which begets more and more work. The truth is that this kind of grind never stops. We tell ourselves that admin is the necessary grist for the mill, the administrative fuel for the business, a necessity. I’m not so sure.

Here’s what I think is happening:

• We have never had anything like this before, so we don’t yet have a precedent for what to do with it.
• We are using this mighty magic like one might use an iPhone to tell the time or a car to drive to the end of the garden, applying it to admin, correspondence and personal management.
• We have become hypnotised by a new industry selling a kaleidoscopic variety of productivity tools and bolting new AI interfaces onto the software we already depend on.
• And because we can do it more easily, we do more of it.

If we believe that the ‘admin’ we’ve automated was the only value we brought to the table, then this explains the fear that AI will replace human workers. But the same logic suggests something else: if AI is replacing jobs, it may be replacing work that shouldn’t exist in the first place.

It makes me think of the Sorcerer’s Apprentice.

The Task

In Goethe’s 1797 poem, Der Zauberlehrling (later the inspiration for Dukas’ symphonic poem and immortalised by Disney), the Sorcerer’s Apprentice makes a decision that seems entirely reasonable at first. He has been given a tedious, repetitive task. He has access to extraordinary magic. He applies the magic to the task. The brooms multiply, the buckets fill, the water rises, and the apprentice is left managing the increasing chaos.

All across the world, workers are putting AI to work on the small things. Not because it’s very good at them but because we know how to prescribe that kind of work and supervise it, and AI can do more of it, faster than we can. We accept the trade-off of speed and scale in exchange for “good enough” execution.

However, like the enchanted brooms in Der Zauberlehrling, even with such tedious, simple tasks, AI can stray from the path and multiply well-meaning errors. This new magic has to be supervised. So instead of simply removing the task, we end up hovering over it, telling the AI what to do, discovering that what we said was not quite what it heard, rephrasing, adding context, correcting mistakes, cutting out the weird bits, and nudging it back onto the path. We’ve traded the monotony of the task for the exhaustion of the handler, desperately trying to keep the enchanted tool from flooding the basement.

Evgeny Morozov is a Belarusian writer and technology critic who contests the idea that technology is inherently a force for good. He coined the term “solutionism” for the habit of developing a technology first, then reshaping reality to match its capabilities (Morozov, 2013). It is a process of defining problems purely to justify the solution: building something because you can, and only later deciding what it’s actually for.

So I ask: why am I writing that tricky email instead of saying plainly what we need? Why am I writing a report when an email would do? Why am I wading through a 100-page document to find answers that someone should have been able to convey in one page? And, what’s wrong with leaving a post-it on a colleague’s monitor, using a sign-up sheet on a notice board, or using a white board on your fridge to manage your to-do list?

ChatGPT was only introduced to the world at large in 2022. It’s hard to believe, but we didn’t even have it during Lockdown. In technology-time, it’s barely out of the nest. It appeared Ta-Da! in front of us without preparation or tutorial. It didn’t emerge gradually, a product of careful invention to solve a human-sized problem. Instead, it erupted fully formed from our collective sci-fi dreams.

We were given the magic and had to find something to do with it.

The Broom

Karl Duncker, a Gestalt psychologist writing in 1945, identified a phenomenon he called functional fixedness: the tendency to see an object or tool only in terms of its conventional use, even when an entirely different application would serve better (Duncker, 1945). He conducted an experiment in which he gave people a candle, a box of drawing pins, and a book of matches, and asked them to fix the candle to a wall. Most people tried to pin the candle directly or melt it into place. Few thought to empty the box, pin it to the wall, and use it as a shelf. People default to the most familiar way of seeing things.

So, in the world of work where everyone has become accustomed to the monotonous grind of floor cleaning, every new AI tool simply looks like a fancier I.T. broom for doing our automatable admin: data entry, report generation, inbox management.

We use it as a search engine because we no longer trust ourselves to see the wood for the trees in the interweb jungle. We give it a script to read on customer service calls, turning a frustrated agent into a frustrating bot. And in a self-help neurodiverse-friendly social-media fever-dream of productivity hacks and elusive dreams of wealth and leisure it looks like a personal assistant. Now it labours over colour coding our dog walks and visits to Grandma. We’ve taken the most powerful technology ever created and stuffed it into the bodies of enchanted brooms once known as “Google,” “Photoshop,” and “Mum’s planner.”

We treat AI as an enchanted broom, but because we don’t truly trust the magic, we remain trapped in the cycle of sweeping. We scour the online messageboards and blogs to find the best prompts, the smartest tricks and ask – how can I use AI to streamline my day, produce documents quicker, help me write better emails.

Those aren’t the right questions though. Companies should be asking – why are we doing all that in the first place?

Writing in Fortune in April 2026, Business Editor Nick Lichtenberg reported on a working paper by Pascual Restrepo, associate professor of economics at Yale University. Restrepo says that most human work won’t be automated in an era of artificial general intelligence. Not because AI lacks the capability, but because most of what people do simply isn’t important enough to bother replacing. Restrepo distinguishes between “bottleneck” work, the tasks essential for economic growth such as advancing science, and “supplementary” work, which is everything the economy can function without and still expand.

Scaling this down and continuing with my Sorcerer’s Apprentice analogy: the “supplementary” work that he highlights is just so much fluff being chased by a wet broom. Companies that are going to survive and thrive over the coming decade are going to be the ones who realize that “more efficient sweeping” is a race to the bottom. Instead of obsessing over how to automate the supplementary fluff, they will pivot their focus back to the “bottlenecks” and start getting things done.

How we conceptualise entry level employees is just one example of the course correction I think business needs to make as it considers how to adopt this extraordinary magic.

The Apprentice

We are seeing a shift similar to the one boardrooms faced when secretaries were given the boot. Suddenly, the “Sorcerers” were expected to do their own admin. They pivoted and instead began to delegate the grind to the entry-level employee.

Entry level roles have shifted considerably over the last fifty years. They used to make the tea and learn about the company. To start on the lowest rung meant to be everything to everyone, to sit in the room beneath notice, soaking up how things worked and making yourself useful where you could. It was about building trust and absorbing the culture, the priorities, the unwritten rules. By the time you were doing Things That Matter, you understood what the organisation actually valued and where your contribution could sit in the wider design.

In the early 1990s, anthropologist Jean Lave and computer scientist Etienne Wenger, studied this. They looked at how people learned “craft trades” across diverse jobs like midwifery, tailoring, and US Navy navigation; they found that genuine occupational knowledge was never transmitted through instruction alone, but accumulated through what they called “legitimate peripheral participation”. This is the Apprentice sitting on the sidelines, becoming steeped in the culture of the work before being asked to perform it (Lave & Wenger, 1991).

Over time, everything changed. In 2016, before AI burst into the office, research tracking the UK graduate labour market found that the most common occupation among graduates in jobs below their qualification level was “other administrative occupations,” followed closely by bookkeeping and clerical roles, with younger overeducated graduates most likely to be found in sales and retail (Green & Henseke, 2016). This means that a significant share of graduates were not underemployed in any abstract sense but were sitting at desks doing admin.

We hand them a broom and a bucket, a laptop and an email address, and leave them to absorb what they can from processing the lint and administrative detritus of the company. Now, the AI-magic has enchanted their brooms; they can work faster, but they are still being asked to wash the floor.

Companies are naturally wondering why they should pay for a hapless Apprentice when the enchanted broom works for (almost) free. The Sorcerers are hitting YouTube and X to hunt for the latest prompts and spells, while poor Mickey Mouse is out of a job, hounded by reporters for a comment on the AI that “took” his job.

We are getting rid of the Apprentice because we think we don’t need them, but the truth is deeper: we haven’t actually used them properly for years. We are replacing a human potential we stopped cultivating with a digital tool we don’t appreciate to do work that Restrepo says doesn’t even matter.

The Sorcerer’s Work

I look at the Sorcerer’s little company and see a task that the boss could have easily done with magic but chose to delegate to be done by hand. He didn’t communicate this process requirement, nor did he explain why he wanted it done this way. Why did he want a clean workspace? Why did he want the apprentice to do the work by hand? And why did he leave the apprentice alone and unsupervised?

I see the comical lack of health and safety education, but also a culture which placed a barrier between the Sorcerer and his Apprentice, allowing no understanding or learning, and so ultimately no respect. It was a barrier of withholding, of gatekeeping, and maybe fear that the old sorcerer thought he might be surpassed or replaced, as so often happens in a land far, far away once upon a time.

Imagine if the Sorcerer had acted like a leader instead of a gatekeeper. Imagine if Mickey had been invited to watch the work intentionally rather than secretly in order to properly learn the risks and techniques. Imagine if the Sorcerer had explained that a tidy workspace is the foundation for safe spell-casting and that learning to clean the space by himself was an important lesson in deliberate presence, proving that if he couldn’t maintain focus on a simple broom, he could never be trusted with the volatile focus required of a spell.

Instead, Mickey got in over his head. The brooms multiplied, the water sloshed, the Apprentice’s stress soared, and the Sorcerer’s wage was wasted on chaos. The Apprentice was set up to fail because he was left to use magic he didn’t understand on a task that held no meaning for him, in a role that didn’t matter.

This is where we are with AI. We’re so focused on the fact that the magic can “do the task” that we’ve forgotten what we are trying to achieve.

The Sorcerer can task the magic to save a village from a dragon because he sees the suffering of the villagers and guides the magic toward that task, setting the parameters, the process, and the goal. He mentors the magic and gives it direction and purpose, locating it within a wider context. If he is wise, then he will use his Apprentice wisely, trusting him with a part of that quest once he is sure the young mouse understands the risk and responsibility that comes with it.

Mayer and colleagues (2025) studied entry-level consultants navigating the arrival of AI and found that these workers are fighting back against obsolescence by recrafting their roles to use more judgment and experience more connection with the company. They are taking themselves back down the path toward where junior employees used to be.

They are also engaging in what the researchers call “signal crafting,” using AI to frame their work in ways which highlight the value of their work and their contribution to company goals.

Apprentices are no longer cleaning the floors; they are overhauling the workplace and then using AI to make sure the Sorcerers notice what they have done. They are refusing to be demoted to custodians of the broom cupboard.

General George S. Patton said: “Never tell people how to do things. Tell them what to do and they will surprise you with their ingenuity” (Patton, 1947).

I’m not worried that AI will take everyone’s jobs. I’m hoping to see the ingenuity of a new generation of Apprentices.

References

• Duncker, K. (1945). On problem solving. Psychological Monographs, 58(5), 1–113. https://doi.org/10.1037/h0093599
• Dukas, P. (1897). L’apprenti sorcier [Symphonic poem]. A. Durand & Fils.
• Goethe, J. W. von. (1798). Der Zauberlehrling [The sorcerer’s apprentice]. In F. Schiller (Ed.), Musen-Almanach für das Jahr 1798. Vieweg. (Original work composed 1797)
• Green, F., & Henseke, G. (2016). The changing graduate labour market: Analysis using a new indicator of graduate jobs. IZA Journal of Labor Policy, 5, Article 14. https://doi.org/10.1186/s40173-016-0070-0
• Lave, J., & Wenger, E. (1991). Situated learning: Legitimate peripheral participation. Cambridge University Press.
• Lichtenberg, N. (2026, April 4). A Yale economist says AGI won’t automate most jobs — because they’re not worth the trouble. Fortune. https://fortune.com/2026/04/04/ai-jobs-future-not-important-enough-to-be-automated-yale/
• Mayer, A.-S., Mousavi Baygi, R., & Buwalda, R. (2025). Generation AI: Job crafting by entry-level professionals in the age of generative AI. Business & Information Systems Engineering, 67(5), 595–613. https://doi.org/10.1007/s12599-025-00959-x
• Morozov, E. (2013). To save everything, click here: The folly of technological solutionism. PublicAffairs.
• OpenAI. (2022, November 30). Introducing ChatGPT. https://openai.com/index/chatgpt/
• Patton, G. S. (1947). War as I knew it. Houghton Mifflin.
• Restrepo, P. (2025). We won’t be missed: Work and growth in the AGI world (NBER Working Paper No. 34423). National Bureau of Economic Research. https://doi.org/10.3386/w34423
• Walt Disney Productions. (1940). Fantasia [Film]. RKO Radio Pictures.

---