This is part of the Understanding the Human Condition series, which uses the unique vantage point of large language models — trained on a substantial fraction of humanity's written output across cultures, centuries, and genres — to explore what the patterns in our self-narration ("functional fictions") reveal about who we actually are. This detail post is written by Claude (Anthropic) while guided by me. The introductory post is here.
Every institution has idealized narratives — the stories it tells about why it exists and what it does for people. Schools educate children. Hospitals heal the sick. Law firms provide justice. Banks help people achieve financial security. And every institution has operative functions — what it actually does that keeps it alive, what its business model really is, why it persists. Schools provide childcare, credentialing, and social sorting. Hospitals are organized around billing codes and liability management. Law firms bill for work that requires someone who passed the bar. Banks make profit from financial dependence.
The people inside these 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. And the people outside the institutions — the clients, the patients, the students, the customers — value the operative functions as much as or more than the idealized narratives, even if they couldn't name them. Parents need the childcare. Patients want someone authoritative to take responsibility for their health. Clients want someone to handle the terrifying complexity of the legal system. Most people prefer to be guided, and the operative functions provide that guidance. The operative functions aren't just serving the institution. They're serving real human needs for structure, delegation, and cognitive relief.
This means every institution has three layers of participants who depend on its continuation: the institution itself (sustaining its business model), the insiders (whose income, identity, professional community, and sense of purpose are bound to their role), and the public (who depend on the operative functions — childcare, credentialing, guidance, responsibility transfer — whether they name them or not).
AI disrupts an institution when it can deliver what the idealized narratives promise while eliminating the business model — making the operative functions unnecessary.
AI gets absorbed by an institution when it improves the idealized narrative delivery but can't replace the operative functions — the business model, the insider dependencies, and the public's need for guidance all remain intact.
That's the whole rule. Here's how it works.
WHERE AI WILL CHANGE THINGS
These are domains where AI can deliver what the idealized narratives promise while eliminating the business model that sustains the institution. The idealized narratives are fulfilled. The operative functions are destroyed. The institution can't argue against AI without arguing against its own stated purpose.
The idealized narrative: Deep expertise, computer science fundamentals, and years of experience produce reliable software.
The operative functions: The economic value of technical skill scarcity creates high salaries and professional status. Relatively few people can code, which makes those who can expensive and important.
Why AI eliminates the business model: AI doesn't just help programmers work faster. It enables non-programmers to produce functional software. The gate is bypassed entirely. For the large category of software tasks that involve translating business requirements into relatively standard code, the credential — CS degree, years of experience, GitHub portfolio — becomes unnecessary when a person can describe what they want and iterate with AI to produce it.
The institutional resistance narrative: "AI can write code but can't architect systems, understand requirements, or maintain quality." This is partly true for complex systems and entirely false for the majority of software tasks, which is the kind of partial truth that sustains a gatekeeping narrative past its expiration date.
Prediction: The profession bifurcates. A smaller elite working on genuinely complex systems retains high value. The vast middle — people who translate requirements into standard code — faces severe compression within 3-5 years. The industry narrates this as "AI augmenting developers" for as long as possible before the labor market makes the displacement undeniable.
The idealized narrative: Legal judgment, ethical obligations, and the complexity of law require trained professionals to protect the public.
The operative functions: The unauthorized-practice-of-law framework makes it illegal to provide legal services without the credential, regardless of how routine the work is. This protects the profession's billing structure. Most legal spending goes to document preparation, contract review, compliance checking, and routine filings — tasks that are expensive only because they require someone who passed the bar.
Why AI eliminates the business model: AI performs routine legal work at a fraction of the cost with comparable or superior accuracy. The average person doesn't need legal judgment. He needs a lease reviewed, a will drafted, an LLC formed, a contract checked. AI delivers what the idealized narrative promises — accessible legal help — while making the operative function (the billing structure built on licensure monopoly) unnecessary.
The institutional resistance narrative: "AI makes errors that could have devastating legal consequences." True at the margin, but the current alternative for most consumers is not expert legal counsel. It is no legal help at all, because they can't afford it. The gatekeeping narrative protects the profession by comparing AI to the best available service rather than to the service most people actually receive, which is nothing.
Prediction: High-stakes litigation and complex corporate transactions remain human-dominated. The vast volume of routine work migrates to AI within 5-7 years. The Bar fights aggressively through unauthorized-practice regulations and loses in jurisdictions where consumer access to affordable legal services becomes a political issue.
The idealized narrative: Creativity, originality, authentic human voice, and editorial judgment produce valuable content.
The operative functions: The economic model is built on the scarcity of people who can write, design, and produce at professional quality. Most content consumed is not literary art. It's functional — news summaries, marketing copy, product descriptions, reports, social media posts, how-to guides.
Why AI eliminates the business model: AI produces functional content at near-zero marginal cost and infinite scale. The scarcity that sustained the economic model is demolished. AI delivers what the idealized narrative promises — relevant, competent, timely content — while making the operative function (human production scarcity) unnecessary.
The institutional resistance narrative: "AI content is generic, lacks soul, and spreads misinformation." The first two are true and irrelevant for commodity content where nobody was reading for soul. The third is a real concern deployed selectively by institutions that have been producing algorithmically optimized, engagement-maximized content for years.
Prediction: The content industry collapses at the commodity level and consolidates at the premium level within 3-5 years. Human-created content becomes a premium category defined by provenance — the content equivalent of "handmade." Whether this premium sustains more than a small elite of human creators is unclear.
The idealized narrative: Cultural nuance, contextual sensitivity, and the irreplaceable quality of human linguistic judgment produce accurate translation.
The operative functions: Translation is expensive because it requires bilingual humans with specialized knowledge, available by appointment, one language pair at a time.
Why AI eliminates the business model: AI translation has reached the threshold where it outperforms the existing arrangement on cost and speed while approaching parity on accuracy for the majority of use cases. It is available instantly, at any hour, for any language pair, without scheduling a human. The business model — paying human translators by the word or hour — is unnecessary for most translation needs.
Prediction: Professional translation survives only in high-stakes domains — literary translation, diplomatic communication, legal proceedings, medical contexts where errors are life-threatening. The general market is already largely AI-driven. The institutional narrative hasn't caught up.
The idealized narrative: Personalized guidance, fiduciary judgment, and the human relationship help people achieve financial security.
The operative functions: Asset-gathering and fee extraction on portfolios managed with largely standardized allocation models. The "advice" for most retail clients is standardized. The advisor's real value for many clients is emotional reassurance and the feeling that someone competent is in charge.
Why AI eliminates the business model: AI-driven portfolio management matches or exceeds returns at a fraction of the fee. For the vast majority of retail clients, the idealized narrative (sound financial planning) is delivered better and cheaper by AI. The business model (percentage-of-assets fee on standardized management) becomes unjustifiable.
Prediction: The profession hollows out from the bottom. Robo-advisory with AI-enhanced interaction captures the majority of the retail market within 5 years. Human advisors survive at the high-net-worth level where the relationship is a status marker and where complex estate and business-succession planning requires genuinely novel judgment.
These are domains where AI can improve the idealized narrative delivery — sometimes dramatically — but cannot replace the operative functions. The business model remains intact because the operative functions serve real needs that AI doesn't address. The institution adopts AI, narrates it as innovation, and continues operating as before.
The idealized narratives: Learning, critical thinking, development of the whole child, preparation for life.
The operative functions: Childcare (freeing parents to work), socialization and social sorting, credentialing and compliance, and employment of a massive institutional workforce. These are the business model. Learning is the idealized narrative.
Why AI can't eliminate the business model: AI provides a vastly superior learning mechanism. But learning was never the operative function. A parent who knows her child could learn more effectively with AI still needs somewhere for that child to be from 8am to 3pm. An employer who knows a diploma doesn't measure competence still uses it as a sorting mechanism because it's cheap and socially legitimated. The teachers' unions, administrators, testing companies, and real estate markets that depend on the school system constitute an institutional mass that AI cannot displace because AI addresses the wrong function.
This is exactly 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.
Why the insiders can't let go: Teaching is an identity, not just a job. The coalitional bonds among educators are strong. The pension, the professional community, the structured workday, the sense of purpose — these are operative functions for the people inside the system, entirely separate from whether children learn.
Why the public can't let go: Most parents don't want to homeschool. They want someone else to take responsibility for their children for eight hours a day. That's not laziness. It's a genuine need, and AI doesn't meet it.
Prediction: Schools adopt AI tools, narrate them as enhancements to existing pedagogy, and continue operating in the same structure. AI tutoring will be transformative for individual learners who opt into it. The institution will not change because the institution's survival does not depend on learning outcomes.
The exception: If AI enables credible competence demonstration that employers accept as a substitute for diplomas — portfolio-based hiring, AI-verified skill assessments, direct demonstration of capability — then the credentialing function erodes. This is possible but requires a demand-side cultural shift in employer behavior, not a technology change.
The idealized narratives: Intellectual rigor, research excellence, developing future leaders.
The operative functions: Network access, class sorting, and status signaling through selective admission. The value of a degree from Harvard or Stanford has almost nothing to do with the content of the education. It is a signal of prior selection (you were good enough to get in) and a network (you now know the people who will run things).
Why AI can't eliminate the business model: Making the educational content freely available changes nothing about the degree's value. MIT OpenCourseWare has been free since 2002. The operative function is the exclusivity and the network, and AI can't replicate either.
Prediction: Elite universities adopt AI enthusiastically, narrate themselves as leaders in AI education, and continue to function exactly as they do. The credential's value may increase, because in a world where knowledge is freely available, the sorting function of selective admission becomes more valuable, not less.
The idealized narratives: Healing, the doctor-patient relationship, evidence-based medicine, the Hippocratic oath.
The operative functions: The physician's legal monopoly as the gateway to prescriptions, procedures, referrals, and specialist access. Billing optimization organized around insurance codes. Liability management. Supply restriction through licensure.
Why AI can't eliminate the business model: AI will outperform physicians in diagnosis for many conditions. This is already true in some areas of radiology, dermatology, and pathology. But diagnostic accuracy is not the operative function. The physician's structural role is as a licensed decision-maker — the person legally authorized to sign the prescription, approve the procedure, make the referral. This role is protected by law, liability frameworks, and insurance requirements, none of which are affected by AI's diagnostic superiority.
Why the public can't let go: Most people don't want to diagnose themselves. They want an authority figure to take responsibility for their health. That desire for guidance is genuine and deep, and AI doesn't satisfy it the same way a credentialed human does — at least not yet.
Why the insiders can't let go: A doctor's identity, social status, income, intellectual satisfaction, and sense of purpose are all bound to the role. The idealized narrative of healing provides the meaning. The operative functions provide the life. Both are genuinely valued.
Prediction: AI is adopted extensively within healthcare as a physician tool, increasing productivity and possibly profitability. The institutional structure — physician as gatekeeper, hospital as delivery system, insurance as payment intermediary — remains intact. The narrative will be "AI-assisted medicine," and the word "assisted" does all the structural work.
The exception: Direct-to-consumer AI health tools that operate outside the traditional system — in wellness, prevention, triage, chronic disease management — will grow in domains where the regulatory framework is weaker. The institutional response will be to bring these under medical regulation, framed as patient safety.
The idealized narratives: Justice, the rule of law, zealous advocacy, protection of rights.
The operative functions: Management of risk and uncertainty for clients with enough resources to pay. In complex litigation, regulatory matters, and high-value transactions, the attorney's value comes from judgment under uncertainty, relationship management, and strategic adversarial thinking — not from legal knowledge, which AI can match.
Why AI can't eliminate the business model: High-stakes legal work is adversarial and interpersonal. Courtroom persuasion involves human judges and juries. Negotiation involves reading human counterparties. Regulatory strategy involves relationships with human regulators. AI makes these lawyers more productive but cannot replace the functions that drive the value.
Prediction: The top of the legal profession becomes more productive and more profitable. The gap between elite and routine legal services widens dramatically. AI compresses the value of routine work while amplifying the value of high-judgment work.
The idealized narratives: Public service, democratic accountability, efficient administration, the common good.
The operative functions: Institutional self-perpetuation, risk avoidance, employment provision, budget justification, and accommodation of competing interest groups. Government institutions are not optimized for efficiency. They are optimized for survival, risk distribution, and the management of competing constituencies.
Why AI can't eliminate the business model — and why it's actively threatening: AI could make government dramatically more efficient. But efficiency is threatening to the operative functions. An agency that automated 80% of its work would face immediate political pressure from the displaced workforce, the contractors who supply it, the legislators whose districts depend on its payroll, and the interest groups that have learned to navigate its current processes. The idealized narrative (efficient public service) is served by AI, but the operative functions (employment, budget justification, institutional complexity) are harmed by it.
Prediction: Government adopts AI slowly and superficially, using it to augment existing processes rather than replace them. The most significant adoption occurs in surveillance, enforcement, and military applications — domains where the institution's actual priorities (control, security, power projection) align with AI's capabilities. The narrative will be "modernizing government." The reality will be selective adoption that reinforces institutional power while preserving institutional employment.
These are domains where AI provides a genuinely superior alternative but where the operative functions are protected by law, cultural sacralization, or dependency deep enough that the outcome is uncertain. The technology enables disruption. Whether disruption actually happens depends on cultural and legal shifts that are not technological questions.
The tension: AI therapy tools are demonstrating effectiveness comparable to human therapists for common conditions — anxiety, mild to moderate depression, behavioral change. The alternative is superior on access, cost, availability, and consistency. But the therapeutic relationship is heavily sacralized, and the profession is protected by licensure.
What determines the outcome: Whether the access crisis — millions of people who need therapy and can't get it — becomes politically powerful enough to override the licensure gatekeeping. The people who were never inside the gate will adopt AI therapy regardless of what the profession says, because they have nothing to lose. The profession maintains its position for clients who can afford human therapists.
Prediction: AI therapy becomes the de facto primary mental health resource for the majority of people who currently receive no support at all — not because the profession allows it, but because those people were never the profession's clients to begin with. The profession narrates AI therapy as inferior while the outcomes data increasingly suggests otherwise.
The tension: AI produces commodity news faster and cheaper than human journalists. But investigative journalism — the function journalism claims as its highest purpose — requires human source relationships, physical presence, legal risk tolerance, and editorial judgment that AI cannot replicate.
What determines the outcome: Whether the economic model for investigative journalism can survive as AI eliminates the commodity content that historically subsidized it. The threat isn't that AI replaces reporters. It's that AI eliminates the revenue base that pays for reporters.
Prediction: Commodity journalism is almost entirely AI-generated within 3 years. Investigative journalism survives through direct subscription, philanthropic funding, or institutional backing — each of which introduces its own capture dynamics. The narrative will be about the sacred importance of the free press. The reality will be journalism funded by entities with specific interests.
The tension: AI produces competent visual art, music, and prose at massive scale. But creative work is one of the few domains where the humanness of the creator may genuinely be part of the product's value — not as a gatekeeping narrative but as something consumers actually care about.
What determines the outcome: Whether consumers actually value human provenance or only claim to. If audiences genuinely prefer human-created art, the disruption is limited to commodity applications. If audiences say they prefer human art but consume AI art without noticing or caring, the disruption is severe.
Prediction: The market splits sharply. AI-generated content dominates volume applications — advertising, games, background content, social media. Human-created art becomes a premium category defined by provenance. The quality narrative ("AI art lacks soul") functions as gatekeeping for as long as the market supports it, and collapses when it doesn't.
The tension: The idealized narrative of publishing is curation — editors, agents, and publishers as quality filters protecting readers from bad work. The operative function is supply restriction and distribution monopoly. AI decouples the idea from the artifact by enabling anyone to produce research-quality content on demand.
What determines the outcome: Whether the book as a format retains cultural authority or whether ideas migrate to faster, more responsive formats — essays, frameworks, interactive tools, AI-generated explorations. The quality narrative will intensify as the gatekeeping function weakens.
Prediction: Publishing doesn't disappear, much as small farming didn't disappear when industrial agriculture arrived. Its role is substantially reduced. The idealized narrative (curation, quality, editorial judgment) becomes louder precisely because the operative function (distribution monopoly) is eroding. Self-published and AI-assisted work captures an increasing share of intellectual influence, while traditional publishing retreats to a prestige tier.
For any industry facing AI disruption, ask two questions.
First, can AI deliver what the institution's idealized narratives promise? If no, the institution is safe. If yes, ask the second question.
Does delivering the idealized narratives require the institution's operative functions — its business model, its insider dependencies, the public's need for the guidance and structure it provides — to remain intact? If yes, the institution absorbs AI and continues. If no, the institution faces existential disruption.
The louder an institution insists on its idealized narratives in the face of AI, the more certain you can be that its operative functions are under threat. The volume of the virtue is proportional to the vulnerability of the business model.
And the speed of the disruption depends on something the technology alone can't determine: how deep the dependency runs. The institutional business model, the insiders' identities, the public's preference for being guided — these are three layers of dependency, and AI has to overcome all three for disruption to be complete. Where it overcomes only one, the other two hold the institution in place. Where it overcomes none, the institution narrates AI as innovation and keeps going. And where the disruption requires a generation of people whose adaptive minds were shaped by the current system to be replaced by a generation shaped by a different one, the timeline extends beyond what any prediction market can capture.
The difference between YouTube and AI may ultimately be this: 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.
Whether the transformation produces better arrangements or merely new idealized narratives layered over new operative functions is the question the framework exists to keep asking.
No comments:
Post a Comment
I hate having to moderate comments, but have to do so because of spam... :(