Overview
The AI Disruption and Institutional Functions framework, developed by Steve Hargadon, provides a predictive model for determining where artificial intelligence will disrupt institutions versus where it will be absorbed into existing structures. The framework distinguishes between an institution's idealized narratives (the stories institutions tell about their purpose) and their actual functions (what they really do to survive and maintain their business models). According to Hargadon's rule: "AI disrupts an institution when it can deliver what the idealized narratives promise while eliminating the business model — making the actual functions unnecessary. AI gets absorbed by an institution when it improves the idealized narrative delivery but can't replace the actual functions."
Core Framework Components
Idealized Narratives vs. Actual Functions
Hargadon identifies a fundamental distinction in how institutions operate. Idealized narratives represent what institutions claim to do: schools educate children, hospitals heal the sick, law firms provide justice, banks help achieve financial security. Actual functions represent what institutions actually do to survive: schools provide childcare and credentialing, hospitals operate around billing codes, law firms bill for work requiring bar passage, banks profit from financial dependence.
Crucially, Hargadon notes that people inside institutions genuinely believe the idealized narratives—"That belief is not a lie. It's the mechanism that keeps them motivated and keeps the public cooperating." Similarly, people outside institutions value the actual functions as much as the idealized narratives, even without explicitly recognizing them.
Three Layers of Dependency
The framework identifies three layers of participants who depend on institutional continuation: the institution itself (sustaining its business model), insiders (whose income, identity, and purpose are bound to their roles), and the public (who depend on actual functions like childcare, credentialing, and guidance, whether they name them or not).
Domains of AI Disruption
Where AI Eliminates Business Models
Hargadon identifies several domains where AI can fulfill idealized narratives while destroying actual functions:
Software Development: AI enables non-programmers to produce functional software, bypassing the credential requirements that create scarcity and high salaries. Hargadon predicts the profession will bifurcate, with a smaller elite retaining high value while the "vast middle" faces severe compression within 3-5 years.
Routine Legal Services: AI performs routine legal work at lower cost with comparable accuracy, delivering the idealized narrative of accessible legal help while making the billing structure built on licensure monopoly unnecessary. High-stakes litigation remains human-dominated, but routine work migrates to AI within 5-7 years.
Content Creation: AI produces functional content at near-zero marginal cost, demolishing the scarcity that sustained the economic model. The industry collapses at the commodity level while human-created content becomes a premium category defined by provenance.
Domains of AI Absorption
Where Actual Functions Remain Protected
Several institutions will absorb AI without fundamental change because their actual functions cannot be replaced:
K-12 Education: While AI provides superior learning mechanisms, schools' actual functions—childcare, socialization, credentialing, and massive institutional employment—remain intact. Hargadon notes this mirrors what happened with YouTube: "YouTube delivered the idealized narrative — you can learn anything, from anyone, for free — better than schools ever had. Nothing changed about schools. Because schools were never really in the learning business."
Elite Higher Education: The value derives from network access and status signaling through selective admission, not educational content. Making content freely available changes nothing about the degree's value, since "the actual function is the exclusivity and the network, and AI can't replicate either."
Clinical Healthcare: Despite AI's superior diagnostic capabilities in many areas, physicians maintain their role as licensed gatekeepers to prescriptions, procedures, and referrals. The public's desire for authority figures to take responsibility for their health remains unaddressed by AI.
The Contested Middle
Hargadon identifies domains where outcomes remain uncertain, dependent on cultural and legal shifts rather than technological capabilities alone. These include mental health therapy, where AI demonstrates comparable effectiveness but faces licensure protection and therapeutic relationship sacralization; journalism, where AI threatens the revenue base that historically subsidized investigative work; and creative arts, where the outcome depends on whether consumers genuinely value human provenance.
Predictive Applications
The Simple Test
Hargadon provides a two-question diagnostic tool: First, can AI deliver what the institution's idealized narratives promise? If no, the institution is safe. If yes, does delivering the idealized narratives require the institution's actual functions to remain intact? If yes, the institution absorbs AI; if no, it faces existential disruption.
The framework suggests that "the louder an institution insists on its idealized narratives in the face of AI, the more certain you can be that its actual functions are under threat." Disruption speed depends on overcoming all three dependency layers—institutional business model, insider identities, and public preference for guidance.
Broader Implications
Hargadon concludes that the fundamental difference between previous technologies and AI lies in their targets: "YouTube attacked what institutions say they do. AI attacks what institutions actually do. That's the difference between a disruption that gets absorbed and a disruption that transforms." The framework serves as a tool for ongoing analysis, questioning "whether the transformation produces better arrangements or merely new idealized narratives layered over new actual functions."