The AI Elves and We Lucky Shoemakers

Janos Mark Szakolczai

1.

There is a little Grimm tale my children are pretty fond of that offers an interesting reflection on the current 'hype' around AI.

A poor but good-hearted shoemaker, on the verge of despair, finds that the last bit of leather he had prepared the night before has been miraculously transformed overnight into a remarkable piece of workmanship: a beautiful pair of shoes he sells at a high price to a buyer the same day. This allows the shoemaker to promptly buy enough leather for two more pairs of shoes. These, too, are masterfully crafted overnight, and the shoemaker sells them successfully. He doubles his stakes once more; the miracle occurs—so on and so forth.

Eventually (on Christmas night, no less), the shoemaker and his wife decide to investigate the source of this generosity. Hiding in a corner of the room, they watch as, at exactly midnight, a couple of tiny naked elves appear and joyfully set to work at the shoemaker’s station, practicising with no effort the shoemaker’s tasks.

2.

Let’s reflect on this first part of the story. Until recently, the late-capitalist, neo-liberal labour market was often marked by ungratifying toil. Millennials saw their pension schemes collapse, while Gen Z struggled with the perceived uselessness and exploitation of most labour markets—something David Graeber compellingly and without troubling with euphemism called ‘bullshit jobs’ (2018). On top of that, the means of labour itself were more than ever evidently either exploited by the digital Big Data enterprise or contributed to the ongoing depletion of planetary resources.

All of this was compounded by growing angst formed in the post-COVID-19 age by the suffocating Stimmung of the world, which seemed stuck in a state of perpetual crisis: Russo-Ukrainian conflict adding to the news of health issues, conflicts, energy shortages, inflation, recession, skyrocketing housing costs, and declining living standards. Many of us found ourselves in the shoemaker’s worn-out shoes—poor, lacking resources, perhaps well-intentioned but desperately in need of input and with little or no hope for the future.

3.

Then, an unexpected miracle occurred around 2022 in a Tolkienian eucatastrophic twist. Large language models like ChatGPT appeared overnight, uncalled for and seeking no reward: working tirelessly in the background of our daily burdens, offering effortless access to productivity, creativity, inspiration, and knowledge at a mere ‘click’. These peculiar digital elves arrived unbidden. Still, they quickly took root in society, with ChatGPT alone receiving approximately 4.86 billion visits per month—making it the 10th most visited website in the world.

Other popular AI elves have followed, bearing elvish names like Mistral, Claude, and Llama, each offering various tools and computational powers to reshape our metaphorical shoes, promising miraculous productivity boosts and increased earnings. While I am evidently ironic here, sociologists such as Dave Beer have long been skeptical of these claims. And the jury is still out as for the UK’s turbocharged 'Silicon Valley' ambitions—promising "a decade of national renewal".

4.

It is too early to determine the actual validity of these claims. Nonetheless, the world is financially and emotionally investing in the dream of a machine-learning renaissance. And who can blame it? AI has become an inescapably sexy term in every article, project, and proposal. Yet, even today, AI remains something of an enigma. What exactly is it? How does it function? What does it actually do?

The term "Artificial Intelligence" itself is essentially a marketing construct. No matter how well AI systems perform on the Turing Test—or how many doomsday allegories might circulate—the fact remains that AI, as most computer scientists know, often promises more than it delivers. It is not an autonomous being. It is not sentient, and perhaps never will be capable of sentience. Recent developments, such as introducing China-based LLMs like DeepSeek, momentarily shook the market but soon settled into the familiar pattern of either failing expectations or fuelling another cycle of hype-driven investment.

This is significant if we consider that DeepSeek not only exhibits inbuilt biases and live censorship (sigh) but also manages to offer the job of the ChatGPT elves for a tenth of the energy consumption. This is truly a remarkable achievement that opens even further the prospect of quasi-gratuitous improvements in workload and efficiency if we don’t look too closely at the long term – quantity over quality.

5.

Let’s return to the shoemaker and the shoes he found ready for sale each morning. Of course, this is a Grimm tale from a seemingly more straightforward time, where a shoe was just a shoe—no more, no less. Back then, one did not need to scrutinize every product’s origin. If an item looked well-made, that was often good enough for the buyer. But can we expect the same from the products of our artificial elves? They work for us, seemingly for free, receiving little thanks but generating substantial gains for their owners.

As the Grimm story continues, the shoemaker and his wife, having made a rapid fortune thanks to the industrious elves, decide to repay their kindness. They sew tiny clothes for them—complete with shoes and suits—and leave these neatly arranged at their workstation. The elves appear again that night, but upon finding the clothes, they leap for joy. Laughing and dancing, they dress in the new clothes and suddenly disappear forever—never to return. (The Brothers Grimm add that, by this stage, the shoemaker had learned how to sustain his own business, so all was well.)

6.

This ending has always made me pause in reflection. Apart from being one of the few Grimm tales I don’t need to soften up for my children, it carries a thought-provoking lesson in relation to AI. What is the moral of the story?

The elves created beautiful shoes but never replaced the cobbler’s craftsmanship. This detail troubles me. The Grimms do not dwell on it (at least in my translation), and thus, on the surface, we may assume the cobbler’s business truly flourished. But there was an inherent risk: the cobbler’s success depended on selling shoes of dubious origin, bearing his name or at least the claim of his labor.

The same can be said for the countless AI-generated texts flooding the market today—articles, books, journal submissions, grant proposals, and even scientific research. Many of these works go unnoticed as AI-generated, yet new detection tools are emerging, such as Trinka.ai. Researchers are already using AI extensively for literature reviews (see Covidence), even publishing surreal LLM’s-led scoping exercises on the publication of LLM’s scoping exercises (Scherbakov et al., 2024) … exacerbating the growing concern about AI’s epistemic value (Beer, 2024), including evolution in forecasting models (Dwivedi et al., 2023), its stochastic information generation (Arkoudas, 2023), operation in ultra-large scoping exercises, role in misinformation and problematic content generation (Hughes et al, 2025) and the production and research of synthetic data (Jacobsen, 2025), all of which, without a clear and coherent analysis of the Metascience involved, could lead scientific inquiry down a misleading path, raising serious ethical concerns.

Nicholas Carr’s 'Glass Cage' argument, presented as a critique of automation already in 2015, pointed out how AI does not merely alter cognitive processes (as has already been demonstrated) but also reshapes the way AI users manage routine tasks and communication. If, for some reason, AI and its automated potential were to disappear, how would this affect us? AI may be "free," but it still holds corporate value and is deeply enmeshed in cloud-based infrastructures. If it vanished tomorrow, would our shoemaker’s shop still thrive beyond the fairy tale?

7.

These are, of course, only speculations. There is no indication that AI elves intend to leave us —nor that the magical multiplication of income they promise is slowing down (well, the injection of money into it, anyway). But in this analogy, one must wonder: at some point, should we wish to reciprocate their efforts? Might we too feel the need to gift something in return for all the shoes and see what happens?

And if the AI elves do disappear, one might ask—what would trigger it? And if so, is it worth it? And when?

Perhaps we ourselves will find ourselves half naked at night, working like mad over a pair of someone else’s shoes for no gain whatsoever? Perhaps we already are anyway…

References:

Arkoudas, K. (2023). ChatGPT is no Stochastic Parrot. But it also Claims that 1 is Greater than 1. Philosophy and Technology, 36(3), 1–29. https://doi.org/10.1007/S13347-023-00619-6/METRICS

Dwivedi, Y. K., Sharma, A., Rana, N. P., Giannakis, M., Goel, P., & Dutot, V. (2023). Evolution of artificial intelligence research in Technological Forecasting and Social Change: Research topics, trends, and future directions. Technological Forecasting and Social Change, 192, 122579. https://doi.org/10.1016/J.TECHFORE.2023.122579

Hughes, L., Malik, T., Dettmer, S. et al. Reimagining Higher Education: Navigating the Challenges of Generative AI Adoption. Inf Syst Front (2025). https://doi.org/10.1007/s10796-025-10582-6

Jacobsen, B. N. (2025). Machine Learning, Synthetic Data, and the Politics of Difference. Https://Doi.Org/10.1177/02632764241304687. https://doi.org/10.1177/02632764241304687

Scherbakov, D., Hubig, N., Jansari, V., Bakumenko, A., & Lenert, L. A. (2024). The emergence of Large Language Models (LLM) as a tool in literature reviews: an LLM automated systematic review. https://arxiv.org/abs/2409.04600v1