# In the age of AI, Raphael wins, not Michelangelo **Authors:** Floriano Bollettini **Categories:** Data & AI **Tags:** art, artificial-intelligence **Last Updated:** 2026-06-04T15:00:00.041Z **Reading Time:** 5 min read --- ## Summary Floriano Bollettini, GM for Italy at Albert School, argues that AI hasn't ended work: it's ended the tolerance for slow professional maturation. The scarce resource now is judgment, not output. --- The numbers are frightening. Dario Amodei, CEO of Anthropic — one of the world's most consequential AI companies — has publicly stated that AI could eliminate up to 50% of entry-level white-collar roles within five years. Goldman Sachs already documents a nearly 20% drop in employment among software developers aged 22 to 25 since 2022. McKinsey — the firm that for decades defined the prototype of the global professional career — has cut around 5,000 roles since 2023 and deployed 12,000 internal AI agents to perform tasks that until yesterday were the exclusive territory of its junior analysts. These numbers shock. But it's worth asking: what exactly is shocking about them? We are living through one of the deepest structural shifts in the history of work. Organizations have not yet had time to fully understand how they operate today, let alone imagine how they will operate tomorrow. What we see in the data is not a photograph of a world ending. It is an X-ray of an organism in full transformation, taken at the most chaotic moment of the transition. Solow's growth model — Nobel Prize in Economics 1987, a cornerstone of every macroeconomics course — tells us something simple and powerful: in the long run, the only engine of sustained per capita GDP growth is technological progress. Technology shifts the production curve upward. It generates more output from the same inputs. And historically, every time this has happened — from the steam engine to electricity, from the internet to industrial automation — work did not disappear. It transformed, expanded, multiplied into forms no one had predicted. The World Economic Forum estimates that AI will create 97 million new roles by 2030, against 85 million displaced. Goldman Sachs projects a potential 7% increase in global GDP. This is not surface-level optimism. It is the logic of every technological revolution before this one. The problem is not the long term. The problem is the transition window — and above all, what it reveals about a system that was already struggling before. ## The slow cursus honorum and its hidden cost For decades, the professional world operated on an implicit logic: time was the proxy for maturity. Analyst, Senior Analyst, Associate, Manager. Five, six, seven years to grow. To make mistakes safely. To sharpen judgment. A slow system, but one with its own logic: professional maturity was assumed to build through accumulated experience, one step at a time. But that system concealed a silent cost. Those who were already mature at 23 — already capable of reading a complex situation, of deciding with incomplete information, of holding their own in a room where the stakes were real — were held back anyway. Stuck doing "just the analysis." Waiting for a title to catch up with their actual capabilities. AI has changed the equation irreversibly. Being a great analyst — gathering data, synthesizing information, producing solid output — is no longer a differentiator. It is the baseline. And a model does it in seconds. What cannot be automated is judgment. Posture. The ability to read a situation, to decide without complete information, to know what matters and what doesn't. To be present in a room where the stakes are real. Some people develop all of this at 24. Others never do. ## Raphael's workshop There is a historical image that comes to mind when thinking through all of this. Raphael had a workshop. He painted with a team of artists — Giulio Romano chief among them — who physically executed significant portions of his frescoes. Raphael thought, directed, composed. He was the intelligence at the center of a distributed execution system. In modern terms: he was an extraordinary prompter. Perhaps one of the greatest in history. Michelangelo was the opposite. On the vault of the Sistine Chapel he worked almost alone, obsessively, every centimeter painted by his own hand. An act of will and genius without precedent — but also a system in which everything passed through a single human bottleneck. Both were absolute geniuses. But in the age of artificial intelligence, the Raphael model wins by a distance. Why? Because today a single individual with the right capacity for thought, direction, and judgment — and the right tools — could produce a thousand Sistine Chapels. This is not hyperbole: it is the logic of an era in which executive power has become virtually unlimited, and the true constraint is the quality of the thinking that governs it. We tend to look at this and feel afraid: "yes, but where is the limit?" — as if the absence of a limit were something to worry about. But look at it from the other side: there is no limit. How many possibilities. How many opportunities. How many things that once required superhuman effort are now within reach of anyone with the right mind. This is not the end of work. It is the beginning of an era in which thinking is worth infinitely more than execution. ## What education owes this moment It is often said that AI threatens the education system. The opposite is closer to the truth: AI is finally calling it to account. Companies can no longer afford to wait years for young professionals to mature through a slow cursus honorum. That budget of time and patience is gone. But this does not mean maturity no longer matters. It means it must arrive earlier. And not by accident, but by design. This is where educational institutions come in. We cannot limit ourselves to transmitting concepts. Our job — and our responsibility — is to build the muscle of judgment, posture, and decision-making before students walk through the door of their first office. This is not a nice-to-have. It is the entire point. The data that initially looked like an alarm reveals itself, seen from this angle, as a direction. For those who take that direction first — students, institutions, companies — there will not be less room. There will be more. >*[This article was first published in the print edition of La Repubblica in May 2026](https://www.repubblica.it/economia/2026/05/27/news/con_l_intelligenza_artificiale_raffaello_batte_michelangelo_e_il_pensiero_vince_sull_esecuzione-425363720/).* ## Key Takeaways 1. The transition window is the real risk, not the destination. Economic history consistently shows that technological revolutions expand total employment over time. What they also produce is a disruptive interim period — and managing that gap is the defining challenge for institutions and individuals alike. 2. Execution is no longer the edge — judgment is the scarce resource. When a model can gather data, synthesize information, and produce clean output in seconds, the baseline shifts. What remains irreplaceable is the capacity to read situations, decide under uncertainty, and act when the stakes are real. 3. The slow cursus honorum extracted a silent tax from the most capable. A system that equated title progression with maturity systematically held back those who were already ready. AI hasn't created this problem — it has simply made the cost of that system impossible to absorb. 4. The Raphael model is the architecture of the AI era. The ability to think, direct, and govern distributed execution — whether human or machine — is the dominant capability structure of the coming decade. Lone-genius execution, however brilliant, hits a ceiling that networked intelligence does not. 5. Education's obligation has shifted from transmission to formation. Passing on concepts is no longer sufficient. The institutional mandate is now to develop judgment, posture, and decision-making capacity in students before they enter the workforce — not as enrichment, but as the core deliverable. --- *Article from [Albert's Deep Dive](https://deepdive.albertschool.com) - Albert School's Journal*