Insynergy Insights | Second in the Judgment Series
On a Tuesday morning in a mid-sized financial services firm, a junior analyst arrives at her desk and opens her laptop. Three years ago, her first task each morning would have been to compile overnight market summaries, synthesize analyst notes from four different data vendors, and produce a two-page briefing for the managing director's 8:30 review. She would spend ninety minutes on this work. She would get things wrong. She would be corrected. She would learn what the managing director actually cared about, what he ignored, what made him stop reading, what made him call her over.
This morning, the briefing has already been generated. It is accurate, well-structured, and delivered to the managing director's inbox at 6:47 a.m. The analyst is now free to work on higher-value tasks.
This is the story organizations tell themselves. It is true, and it is incomplete.
I. The Historical Function of Junior Work
Entry-level analytical labor has never been primarily about the output it produced. It has been about the formation it enforced.
There is a tendency, when examining the history of professional development, to misread the apprenticeship model as a kind of inefficient staffing arrangement that predated automation. In this misreading, junior analysts, junior associates, and junior consultants were simply cheap labor performing tasks that senior professionals could not be bothered to do themselves. Automation arrives, the tasks disappear, and the organization becomes more efficient.
This reading misses the structural logic entirely.
The apprenticeship model existed because judgment — the specific capacity to act under uncertainty with incomplete information and accept accountability for the outcome — cannot be transmitted through instruction. It can only be formed through exposure. The junior analyst preparing the morning briefing was not merely producing a document. She was developing a working model of what constitutes signal versus noise in financial markets. She was learning, through repeated correction, what it means for information to be actionable versus merely interesting. She was acquiring, through a thousand small interactions, a tacit map of where organizational authority actually sits, as distinct from where the org chart says it sits.
This distinction between tacit knowledge and explicit instruction is not new to organizational theory.[^1] What is less often examined is how tacit knowledge forms — specifically, how the conditions required for its formation are structural, not pedagogical.
Judgment forms through consequence. An analyst who summarizes a report incorrectly, and who then watches the managing director make a poor recommendation on that basis and face criticism in the executive committee, has learned something that no training curriculum can deliver. The consequence closes the loop. The loop creates the formation.
Repetition matters because it is not the individual failure that teaches. It is the pattern recognition that emerges from many cycles of attempt, feedback, and adjustment. The tenth briefing is different from the first not because the analyst has learned new facts but because she has developed a felt sense for the structure of the problem — what a good briefing feels like before it is evaluated, what questions to anticipate, where the gaps tend to be.
This felt sense is judgment in its early form. It is not yet authority — the capacity to act on judgment in the face of institutional resistance or high-stakes uncertainty. Authority requires additional formation: exposure to escalation dynamics, to political friction, to decisions that cannot be reversed. But the felt sense is the prerequisite. Without it, authority becomes performance. The person can speak with confidence, but the confidence is not grounded in consequence-based formation. It is grounded in facility with process, fluency with tools, and familiarity with the vocabulary of the domain.
The apprenticeship model, properly understood, was never sentimental. It was structural. Junior work existed because organizations needed a mechanism to grow the next generation of people capable of bearing the weight of complex judgment. The work was the mechanism.
II. AI Compression of Entry-Level Work
The economic logic of eliminating junior analytical work is straightforward and largely irresistible.
AI systems in 2025 can perform research synthesis, first-draft document production, data aggregation, and structured reporting at a quality level that meets or exceeds what a first or second-year analyst produces in many domains, at a fraction of the time and cost. For organizations under margin pressure, the calculation is not complex. The junior analyst who spent ninety minutes on a briefing can now contribute to work that AI cannot yet do, or the headcount can be reduced, or both.
This is not a technology hype narrative. The productivity gains are real, measurable, and in many domains already realized. Industry surveys and firm-level data consistently show meaningful reductions in the time required for due diligence preparation, first-draft document production, and research synthesis across financial services, legal, and consulting practices. In one documented case at a major US law firm, a complaint response process was reduced from sixteen hours of associate time to a matter of minutes. These gains compound at scale.
The automation is also not uniform across all entry-level work. Work that requires political navigation, client relationship management, and real-time adaptive judgment in high-stakes conversations remains largely human. What is being automated is the analytical foundation layer — the structured, repeatable cognitive work that historically required trained humans because it required intelligence, pattern recognition, and language ability that machines could not yet provide.
The rationality of this automation, viewed at the level of the individual task or the individual quarter, is genuine. The organization that does not automate these functions faces a competitive cost disadvantage. The managing director who waits for her junior analyst to produce the briefing is consuming ninety minutes of senior attention overhead that her competitor has already eliminated. Individual organizations cannot afford to preserve inefficiency as a training mechanism when their competitors are eliminating it.
This creates a collective action problem at the level of the profession and the economy, but that is not the frame that concerns us here. The concern here is structural, and it operates at the level of the individual organization's long-term formation capacity.
The short-term productivity gain and the long-term formation loss do not appear on the same balance sheet. The gain appears immediately, in measurable output metrics. The loss accumulates slowly, invisibly, in the quality of judgment that enters the organization's senior leadership pipeline eight to twelve years later.
III. The Formation Gap
Skill, intelligence, judgment, authority, and formation are not synonyms, and conflating them is the conceptual error at the center of most organizational responses to AI displacement.
Skill is domain-specific competence: the ability to perform a defined task correctly. Intelligence is the general capacity for pattern recognition, reasoning, and synthesis. Judgment is the capacity to act under genuine uncertainty — where information is incomplete, consequences are significant, and the decision cannot be fully specified in advance by any rule or protocol. Authority is the institutionally recognized right and personally internalized willingness to exercise judgment on behalf of an organization in high-stakes conditions. Formation is the process by which judgment and authority are developed.
AI can accelerate skill acquisition dramatically. A junior analyst using AI tools can develop domain knowledge in weeks that previously required months of document review. AI can also augment intelligence — providing synthesis, pattern recognition, and second-order analysis that extends the reach of human cognitive processing. These are genuine benefits.
What AI cannot do is replace consequence-based formation. Formation requires that the individual, at some point, is the accountable party. The loop must close on them. They must feel the weight of a decision whose outcome they influenced, and they must experience that weight repeatedly, in varied conditions, at escalating levels of complexity and stake.[^2]
The junior analyst preparing the briefing was not merely practicing a skill. She was learning what it means to be accountable for information that others act upon. She was developing, through experience rather than instruction, a sensitivity to the difference between information that is technically accurate and information that is decision-relevant. This sensitivity does not transfer through coaching or mentoring alone, though both help. It transfers through exposure, which requires proximity to real decisions under real stakes.
What is now emerging inside organizations is a category of professional that might be called tool-fluent but authority-fragile. These are individuals who can operate sophisticated AI systems with facility, who can produce high-quality outputs efficiently, who have acquired substantial domain knowledge through structured learning, but who have not been formed by consequence. They know what a good decision looks like. They can describe the framework. They cannot bear the weight of making it when the organization needs them to.
This is not a failure of education or training. It is a structural gap in the formation architecture. The terrain on which judgment was historically formed — the iterative, consequence-laden, often tedious work of entry-level analytical labor — has been removed from the organization's operating model before a substitute terrain was designed.
The escalation dynamics that formed judgment in the apprenticeship model had a specific character. The junior analyst was exposed to consequences through proximity: she was in the room when the briefing she produced was used, she heard the questions it failed to answer, she experienced the managing director's frustration when the analysis missed the essential variable. These moments of proximate consequence — not catastrophic failure, but real feedback with real stakes — created the formation substrate.
When that work disappears, the proximate consequence disappears with it. The junior professional operates AI tools and produces outputs that are evaluated against quality metrics, but the consequences of those outputs are mediated, deferred, and diffused. The feedback loop does not close on her in the same way.
IV. The Hidden Five-to-Ten-Year Risk
The formation gap does not produce visible failures immediately. This is the characteristic of structural risks that compound slowly: the damage is not visible at the moment it occurs.
The current generation of senior leaders in most large enterprises developed their judgment in organizations where the apprenticeship model was still largely intact. The managing director who now reads the AI-generated briefing at 6:47 a.m. developed her judgment by spending years producing imperfect briefings and learning from the feedback. Her authority is grounded in consequence-based formation. Her capacity to evaluate the AI-generated briefing — to sense when it has missed something, to ask the question it should have prompted but did not — is itself a product of the formation that is now being eliminated.
This capacity is not visible on an organizational chart. It is not captured in any performance metric. It simply exists in the senior leadership cohort that currently bears the weight of complex organizational judgment. And it is aging.
In five to ten years, many of these leaders will retire or move on. The pipeline behind them will contain a generation of professionals who entered the workforce after the automation of entry-level analytical work, who developed skill and intelligence in AI-augmented environments, but who did not accumulate the consequence-based formation that their predecessors received through the apprenticeship model. Multiple longitudinal studies of the post-2022 labor market have documented the accelerating decline of entry-level hiring in AI-exposed roles across financial services, legal, and technology sectors — the very roles that historically served as formation terrain.
The governance structures of these organizations will appear intact. The processes will be documented. The decision frameworks will be sophisticated. But the people inside those structures will have been formed differently, and the difference will not be apparent until the organization faces conditions that require judgment under genuine uncertainty — a crisis, a strategic inflection point, a regulatory challenge that cannot be solved by following existing protocol.
At that moment, the organization will discover that it does not have a training problem. It will have a formation problem. And formation problems cannot be solved with training.
This is not a prediction about a specific catastrophic failure. It is an observation about governance fragility — the slow erosion of the organization's capacity to exercise complex judgment, occurring beneath the threshold of visibility, masked by strong short-term performance metrics, emerging only when the conditions that require genuine judgment finally arrive.
The pipeline erosion is already underway. The formation gap is accumulating now. The risk will manifest on a timeline that makes it very difficult to connect cause and effect when it does.
V. Decision Design as Formation Architecture
The response to the formation gap is not to halt AI adoption, to artificially preserve inefficient tasks, or to return to apprenticeship arrangements that are no longer economically viable. These responses misframe the problem.
The correct response is to extend Decision Design beyond its conventional function as a governance architecture and develop it explicitly as a formation architecture — a set of organizational structures and processes that are designed not only to allocate decision authority correctly but to develop the judgment required to exercise that authority well.
In its original formulation, Decision Design addresses a specific organizational problem: as AI systems take on more analytical and operational work, the boundary between human judgment and AI-assisted process must be deliberately designed. Organizations that fail to design this boundary experience a diffusion of accountability — decisions are made, but the location of judgment within those decisions is ambiguous. Decision Design clarifies this boundary, assigns authority, and creates governance structures appropriate to the risk profile of different decision types.
This is necessary but insufficient. Clarifying where judgment must be exercised does not address the question of how the people who must exercise that judgment were formed.
Formation architecture is the extension of Decision Design that addresses this question. It requires three structural commitments.
The first is the explicit design of exposure pathways. Formation requires exposure to decisions with real stakes, at escalating levels of complexity. In an AI-augmented organization, this exposure does not occur automatically. It must be engineered. This means identifying the specific decision moments in an organization's operating model where junior professionals can be deliberately placed in positions of proximate accountability — not as decision-makers, but as the individuals whose analysis, judgment, and framing materially influence the outcome. It means ensuring that these professionals are present when their contributions are evaluated, that they hear the consequences, and that the feedback loop closes on them personally.
This is not mentorship in the conventional sense. Mentorship is relational and often informal. Exposure pathway design is structural. It is a deliberate answer to the question: at what moments in our operating model does the next generation of decision-makers need to be present, and how do we ensure they are?
The second structural commitment is consequence visibility. In an AI-augmented organization, the consequences of decisions are often mediated by multiple layers of process. The individual who contributed a judgment call may never observe its downstream effect. Formation architecture requires the deliberate creation of consequence visibility mechanisms — processes by which junior professionals can trace the downstream effects of decisions they influenced, observe when those decisions were correct or incorrect, and develop the pattern recognition that forms judgment over time.
This can take multiple forms: structured after-action processes that include junior team members, explicit traceability between analytical contributions and decision outcomes, and deliberate feedback mechanisms that close the consequence loop rather than allowing it to dissipate in organizational complexity.
The third structural commitment is what might be called the Decision Formation Layer — a designed component of the organization's operating model that sits between AI-assisted analytical production and senior decision-making, and that is explicitly structured to provide formation rather than merely quality control. The Decision Formation Layer is not a bottleneck or a bureaucratic approval step. It is a structured environment in which junior professionals are required to exercise judgment about the AI-generated analysis they are working with, to identify its limitations, to make recommendations under uncertainty, and to defend those recommendations in conversation with more senior colleagues.
The key design principle is that the junior professional in the Decision Formation Layer is accountable for a judgment call, not merely a quality check. Quality checks can be performed without formation. Judgment calls, even small ones with limited stakes, begin to accumulate the consequence-based experience that forms decision-making capacity over time.[^3]
This is an artificial apprenticeship layer — artificial in the sense that it must be deliberately constructed, because the natural apprenticeship that existed in pre-automation organizations is no longer economically viable as a by-product of ordinary operations. The construction of this layer is not a cost burden if it is understood correctly. It is an investment in the organization's future judgment capacity, which is its most durable competitive asset and its most invisible one.
The design challenge is significant. Artificial apprenticeship pathways require explicit specification of what judgment calls junior professionals should encounter, at what level of stake, and in what sequence. They require senior leaders who understand that their role includes formation and not merely output quality review. They require organizational cultures in which the formation function is valued and resourced, rather than sacrificed to short-term efficiency imperatives.[^4]
None of this is simple. But the alternative is simpler only in the short term.
On that Tuesday morning, the junior analyst at the financial services firm opens her laptop and finds the briefing already in the managing director's inbox. She is relieved. The ninety minutes she would have spent on it are now available for other work. The managing director is pleased with the briefing quality. The efficiency gain is real.
But something else quietly disappeared that morning. It did not disappear in a single event. It had been disappearing for months, and it would continue to disappear over the following years, and it would not become visible until the analyst, now a senior manager, was asked to lead the firm through a situation for which no prior briefing existed, no AI system had a template, and the only resource available was judgment formed by consequence.
The situation would call for someone who had spent years inhabiting the structure of consequential decisions from a position of proximate accountability. Someone whose pattern recognition had been trained not on clean data sets but on the messy, ambiguous, politically charged reality of decisions made under pressure with incomplete information.
That formation, if it was not deliberately designed, will not be there. The process will be there. The tools will be there. The frameworks will be there. The people who built those frameworks will have long since retired.
Organizations that do not design how judgment is formed will discover that judgment cannot be improvised.
Notes
[^1]: The concept of tacit knowledge as foundational to professional formation originates with Michael Polanyi. See Polanyi, M. (1958). Personal Knowledge: Towards a Post-Critical Philosophy. University of Chicago Press; and Polanyi, M. (1966). The Tacit Dimension. Doubleday. Polanyi's formulation — "we can know more than we can tell" — establishes the epistemological basis for why apprenticeship produces formation that instruction alone cannot replicate. The organizational application was subsequently developed by Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company. Oxford University Press.
[^2]: The argument that genuine learning requires the learner to experience the consequences of action — and that this experience cannot be substituted by instruction or observation alone — is foundational to Dewey's theory of experiential learning. See Dewey, J. (1938). Experience and Education. Kappa Delta Pi. Dewey's distinction between "miseducative" experiences (those that arrest future growth) and formative ones (those that open further capacity for judgment) is directly applicable to the organizational formation question addressed here.
[^3]: The distinction between quality-check functions and judgment-formation functions in organizational processes maps onto Argyris and Schön's distinction between single-loop and double-loop learning. Quality checks operate in single-loop mode — error detection against existing standards — while judgment formation requires double-loop engagement: questioning the governing assumptions behind the standard itself. See Argyris, C., & Schön, D.A. (1978). Organizational Learning: A Theory of Action Perspective. Addison-Wesley.
[^4]: On the organizational conditions required for generative learning capacity — the ability to build capability rather than merely adapt to existing conditions — see Senge, P. (1990). The Fifth Discipline: The Art and Practice of the Learning Organization. Doubleday. The formation architecture proposed here is, in Senge's terms, an investment in generative rather than adaptive capacity.
References
Argyris, C., & Schön, D.A. (1978). Organizational Learning: A Theory of Action Perspective. Reading, MA: Addison-Wesley.
Dewey, J. (1938). Experience and Education. New York: Kappa Delta Pi.
Nonaka, I., & Takeuchi, H. (1995). The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation. New York: Oxford University Press.
Polanyi, M. (1958). Personal Knowledge: Towards a Post-Critical Philosophy. Chicago: University of Chicago Press.
Polanyi, M. (1966). The Tacit Dimension. New York: Doubleday.
Senge, P. (1990). The Fifth Discipline: The Art and Practice of the Learning Organization. New York: Doubleday.
RYOJI | Insynergy inc. | Insights | Decision Design Series © 2026 Insynergy inc. All rights reserved.