In the mid-1460s, Leonardo da Vinci entered the Florence workshop of Andrea del Verrocchio, one of the city’s most sought-after painters and sculptors.1[1] He would remain there for roughly a decade, and documentary evidence places him in the workshop even after he joined the painters’ guild in 1472. For most of that time, he was working within the workshop rather than as an independent master.
According to Giorgio Vasari, whose 1550 Lives of the Most Excellent Painters, Sculptors, and Architects provides one of the earliest major sources on Leonardo’s early years, an apprentice’s early years in a workshop such as Verrocchio’s involved the unglamorous infrastructure of the craft: grinding pigments, preparing wooden panels, mixing varnish, and assisting with the physical work behind a finished painting.2[2] Only gradually did an apprentice take on more consequential parts of a commission. Vasari tells the story of Leonardo painting an angel in Verrocchio’s Baptism of Christ, saying that the figure was so far beyond anything Verrocchio himself could produce that the older man never picked up a brush again.3[3]
Embellished or not, the anecdote captures something real: a structure in which a gifted boy spent years working inside a master’s workshop, contributing to real commissions while learning under an established standard before beginning an independent career. That structure was not unique to Verrocchio’s shop; it was one of the principal ways that skill was transmitted across the premodern trades. And it is worth asking why we rarely build skills that way today and what, exactly, is lost when we don’t.
The System Behind the Story
Florence’s painters belonged to the Arte dei Medici e Speziali (one of the city’s guilds), and guild apprenticeship followed a broadly consistent pattern: An apprentice was bound by contract, often for several years, to a master who owed him training and discipline in exchange for his labor.4[4] Its form and length varied considerably from place to place. In London’s livery companies, the seven-year apprenticeship became the standard formal term, reinforced by the 1562 Statute of Artificers, although many apprentices left before completing it.5[5] The length was not arbitrary, nor was it a hazing ritual; it reflected a widespread recognition of the fact that some forms of competence could not be acquired quickly or transmitted through instruction alone.
The point is not that seven years is a magic number. It is that certain kinds of competence require sustained study and practice under someone who already knows the difference between work that looks right and work that is right. In all fields of human endeavor—and particularly in craftsmanship-focused fields such as painting, sculpture, and writing—theory is necessary but not sufficient; mastery requires ongoing, reciprocal cycles of both study and practice.
The Death of Apprenticeships Is, in Large Part, the Death of Real Mastery
Contrast this with the dominant model of skill acquisition offered today. Coding bootcamps, for example, advertise “job-ready” developers in twelve to sixteen weeks, a fraction of the time required to obtain a traditional computer science degree; and a degree is no guarantee of mastery either, let alone the kind of mentored apprenticeship a junior engineer might once have served.6[6] Generative AI tools can now produce terrible writing that appears fluent and powerful to people who have never revised an awkward paragraph by hand enough times to learn the difference between a sentence that works and one that doesn’t. Online platforms host a vast array of “learn guitar in thirty days” and “become fluent in German in six weeks” programs, each promising to exponentially compress learning that was once understood to take years.
These tools and platforms are popular because, broadly speaking, we have been progressively trained over several generations to believe that meaningful proficiency doesn’t necessarily require lots of hard work over a long time. What has changed is the correction loop. Leonardo’s errors could be caught and corrected within the workshop—with real consequences: He might have to skip dinner or stay up all night to redo a poorly prepared panel, or he might have to suffer a little embarrassment at seeing a clumsy error corrected by the master himself.
The modern learner’s errors, by contrast, are often simply absorbed without comment, correction, or understanding. A tutorial video does not notice that you’ve misunderstood it. An AI draft does not flag that it has papered over a gap in your own thinking that will resurface the first time someone asks you a hard question about your work. To be fair, not every modern program is so passive. The better ones are not mere information consumption; they ask the learner to speak the language or write the code, and some can correct pronunciation and grammar in real time. Even so, no technological tool or system can come close to equaling the mind of a human who has mastered his craft—a human who can discuss, debate, learn, understand, guide, and correct in real time, in active, back-and-forth cooperation with his student.
What Gets Lost?
One clear illustration of the correction loop’s purpose comes from a source that may be surprising to some: sushi. At Jiro Ono’s restaurant in Tokyo, an apprentice does not simply decide when he is ready to advance. According to director David Gelb, one apprentice had worked there for ten years before being allowed to make the restaurant’s egg sushi, and then he had to make it roughly two hundred times before Jiro accepted the result.7 The repetition alone is not the point; each attempt is judged by someone who can tell the difference between right and wrong. This is where judgment and precision are built: repeated attempts, rejection, adjustment, and more attempts until the work meets a standard.
But the long apprenticeship built a second, subtler thing, and it is easy to miss if skill is thought of only as the ability to perform a task. In Rembrandt van Rijn’s Amsterdam studio, pupils and assistants worked closely with the master. They produced copies and close variants of his work, executed parts of compositions, and sometimes had their work retouched by Rembrandt himself.8 The result is that scholars are still disputing, four centuries later, which paintings are Rembrandt’s and which are his students’. The Rembrandt Research Project, a team of art historians formed in 1968 specifically to resolve these attributions, has spent decades using X-ray, infrared, and dendrochronological analysis and still has had to reverse several of its earlier rulings.9
The attribution problem reveals how thoroughly Rembrandt’s workshop trained artists to work within his visual language. They had learned, as far as a pupil could, to understand a master’s standards: when a passage of paint was finished, when a shadow was true, when a likeness had been captured. That is a different kind of competence than rice-preparation technique or panel preparation. Those are skills of making; this is a skill of judging, of knowing whether what has been made is good and when it is finished. It is the capacity to judge one’s own work by a standard one did not invent and could not yet fully articulate, absorbed from the master through years of his corrections, until one could apply it unaided.
This is precisely what no tutorial or AI tool can offer, no matter how good its output or feedback is. A writer trained on such systems can (sometimes) produce a passable paragraph in seconds without understanding and validating the evidence-based standard that enables him to recognize, on his own, whether, why, and how that paragraph is good. He has a superficial facsimile of the product without the judgment that makes the real thing possible: Rembrandt’s brushwork, so to speak, without Rembrandt’s eye. The apprenticeship model existed to close this gap before the stakes got real.
Where Apprenticeship Survives
The apprenticeship system has not disappeared entirely from fields in which failing to meet a certain standard carries obviously serious consequences, and the reasons are instructive. For example, surgical residents in the United States still spend years progressing from supervised practice toward independent responsibility. General surgery residency requires five years of training, with increasing responsibility as residents advance, a structure rooted in the model William Halsted established at Johns Hopkins University in the 1890s.10
Commercial aviation offers a starker lesson because it shows what happens when new practitioners carry major responsibility without commensurate experience. In February 2009, Colgan Air Flight 3407 crashed outside Buffalo, New York, killing all forty-nine people aboard and one person on the ground. The National Transportation Safety Board’s stated probable cause was the captain’s inappropriate control inputs, which led to the aircraft stalling, with contributing factors that included inadequate training, procedures, and monitoring.11 Before 2013, first officers at U.S. regional airlines could enter airline operations with substantially less flight experience than the fifteen hundred hours later required for an Airline Transport Pilot certificate. After the crash, Congress enacted the Airline Safety and Federal Aviation Administration Extension Act of 2010, implementing a higher standard.12[12] The change embodied a policy judgment that formal certification alone was not enough: Before assuming the responsibilities of airline operations, pilots would have to accumulate substantially more flight experience. Aviation thus offers a useful reminder of a key premise underlying apprenticeship systems: Competence cannot be inferred from a credential alone.
Law offers a subtler but no less consequential example. English barristers still complete a period of supervised practical training known as pupillage, normally lasting twelve months, shadowing a senior barrister, drafting, and arguing under close supervision before practicing independently.13[13] Judgment under courtroom pressure, like surgery, has never been teachable from a book alone.
What unites these fields is that the cost of error is immediate, visible, and sometimes fatal; failure cannot be hidden behind a superficially polished output the way a mediocre essay or an underprepared presentation can be.
None of this means that every skill demands years under a master; a task whose rules can be stated fully and whose results can be checked at a glance, like changing a tire or tying a bowline, can be learned quickly, but in cases in which competence rests on firsthand experience and subconscious integration of skills, in which error is costly or hard to see, nothing replaces the years.
In cases in which the stakes are undeniable, the apprentice system has proven hard to abandon. Apprenticeships may be long, but the experience they build enables the mastery that’s necessary for excellence.
Why This Is a Moral Matter
It would be easy to end there, with a case for better training programs. But the reason this issue matters is not only economic. It bears on the values a person’s life requires if he is to live it well.
A human being does not merely need activity to live well. He needs what Ayn Rand called productiveness: purposeful, productive achievement that is genuinely his and that he has earned in full rather than approximated or simulated.14[14] Sustained practice and correction build competence, competence builds independent judgment, independent judgment makes achievement fully one’s own, and earned achievement is the foundation of productiveness and of justified pride. Break the progression at the start, and a person has no rational basis for confidence in his own ability, confidence that cannot be faked into existence. Leonardo’s decade under Verrocchio, the ten years and two hundred eggs in Jiro’s kitchen, the pupils working in Rembrandt’s studio—each was the price of being able to stand behind one’s own work without the corrosive suspicion that it wouldn’t hold up under real scrutiny. Apprenticeship builds the rational basis for self-esteem: the knowledge that one’s ability has been earned by one’s own effort and tested against reality.
This is what a culture of convenience disincentivizes when it hands people a credential, a polished portfolio, or an AI-assisted draft: the chance to have genuinely earned a skill—and the judgment to know when one has. We can’t shortcut our way to competence.
None of this is an argument for reviving guilds or restoring seven-year terms by law. No one should be compelled to serve an apprenticeship; but we should understand why this model is so important in so many fields, and we should regain the respect for it that we once had. A rational person should want to know whether his abilities are real and should, therefore, seek conditions that can expose his errors before those errors become his habits. That means deliberately seeking out the correction that the old system provided naturally: a mentor honest enough to tell you when the work isn’t right, a standard exacting enough that meeting it means something, and the patience to stay in the gap between knowing and doing for as long as closing it takes. That is why apprenticeships are worth reviving: because a productive purpose, honestly earned, is one of the central things a good life is built on, and no shortcut can ever deliver it.
Jill Dunkerton, “Leonardo in Verrocchio’s Workshop: Re-examining the Technical Evidence,” National Gallery Technical Bulletin (2011); see also the National Gallery’s page on Andrea del Verrocchio, on Leonardo’s training in Verrocchio’s workshop.
Giorgio Vasari, Lives of the Most Eminent Painters, Sculptors, and Architects (1550; trans. Gaston du C. de Vere, 1912–1915), entry on Leonardo da Vinci.
Vasari, Lives of the Most Eminent Painters, Sculptors, and Architects. The “Verrocchio never painted again” claim is widely treated by art historians as Vasari’s embellishment.
Maarten Prak and Patrick Wallis, eds., Apprenticeship in Early Modern Europe (Cambridge University Press, 2019); see Wallis’s chapter on England for the contract-based structure and its variation.
Patrick Wallis, “Apprenticeship in England,” in Prak and Wallis, eds., Apprenticeship in Early Modern Europe, 247–81, on the 1562 Statute of Artificers, the seven-year term, early departures, and the absence of a formal end-of-term examination. On London’s livery companies, see also Ian Archer, The History of the Haberdashers’ Company (Phillimore, 1991).
Figures drawn from advertised program lengths of major bootcamp providers (e.g., App Academy, Hack Reactor).
David Gelb, quoted in “David Gelb, Jiro Dreams of Sushi,” Filmmaker Magazine; David Gelb, dir., Jiro Dreams of Sushi (2011).
Ernst van de Wetering, Rembrandt: The Painter at Work (University of Amsterdam Press, 1997); National Gallery of Art, “Issues of Attribution in the Rembrandt Workshop.”
Rembrandt Research Project, A Corpus of Rembrandt Paintings, vols. I–VI (1982–2014), https://link.springer.com/series/6494.
Peter Olch, “William S. Halsted and the Surgical Residency System,” Journal of the American Medical Association (1976); Johns Hopkins Medicine, “History” (on Halsted’s principles of progressive responsibility); general surgery residency program requirements.
National Transportation Safety Board, Loss of Control on Approach, Colgan Air, Inc., Operating as Continental Connection Flight 3407 (NTSB/AAR-10/01, 2010).
Airline Safety and Federal Aviation Administration Extension Act of 2010, Pub. L. No. 111-216; FAA, Final Rule on Qualification, Service, and Use of Crewmembers and Aircraft Dispatchers, 78 Fed. Reg. (July 15, 2013), implementing the fifteen hundred-hour ATP requirement.
Bar Standards Board, Bar Qualification Manual, on the normal twelve-month duration of pupillage.
Ayn Rand, “The Objectivist Ethics,” in The Virtue of Selfishness (Signet, 1964), 26; see also the Ayn Rand Institute’s Lexicon entry on Productiveness.