Chapter 12: Applied Ethics: Technology, Medicine, Business, and the Environment

Ethics Is Not a Spectator Sport

Philosophy has a reputation problem. Ask most people what philosophers do and they will describe someone in an ivory tower, spinning arguments about problems that have no bearing on the real world — puzzles about whether the external world exists, whether we have free will, whether a ship that has had all its parts replaced is still the same ship. These are real questions, and they matter more than they appear to. But they feed the suspicion that philosophy is an elaborate game played with concepts, disconnected from the daily business of living.

Ethics is the clearest counterexample to that suspicion. Every week — every day — you face questions that are, at their core, ethical. Should you say something when a colleague takes credit for your work? What do you owe the strangers who show up in your social media feed? When your employer asks you to do something that makes you uncomfortable, how do you decide whether to comply? When your city debates where to put a new highway, who speaks for the neighborhood that will be displaced? When your doctor recommends a treatment you don't fully understand, how much do your own preferences matter?

These are not hypotheticals. They are the texture of an ordinary life. And the frameworks you have encountered in the past nine chapters — consequentialism, Kantian ethics, virtue ethics, care ethics, Rawlsian justice, contractualism — exist precisely to help you think more clearly about exactly these situations.

This chapter is where the theory meets the road.

We will take four domains where ethical questions arise with unusual intensity and consequence: technology, medicine, business, and the environment. In each domain, we will take a hard case — not a easy case constructed to make one answer obvious — and apply multiple ethical frameworks to it. The goal is not to arrive at a single correct answer that everyone must accept. The goal is to demonstrate what careful ethical reasoning actually looks like, to show where different frameworks agree (which strengthens any conclusion they share), and to show where they disagree (which illuminates what is genuinely at stake in the disagreement).

There is a methodological point here that is worth stating explicitly before we begin. When you apply five different frameworks to the same problem and they all give the same answer, you have strong reason to act on that answer. When they diverge, you have learned something important: you have identified exactly what the core moral issue is, because the frameworks divide along their deepest structural differences. That divergence is not a failure of ethics to give you an answer. It is ethics doing its best work — clarifying the real choice you face.

Part One: Technology Ethics

The Problem That Didn't Have a Name

In 2015, Amazon scrapped an AI recruiting tool it had been developing for years. The system had been designed to evaluate job applicants by analyzing their resumes, identifying patterns that correlated with success at Amazon, and ranking candidates accordingly. It would have saved enormous amounts of time and introduced apparent objectivity into a process notorious for human bias.

There was one problem. The system had learned to discriminate against women.

No engineer had told it to do this. No one had written code that said "down-rank female applicants." What had happened was subtler and in some ways more troubling: the system had been trained on resumes submitted to Amazon over a ten-year period, and during that period, the tech industry was predominantly male. The system learned that successful Amazon employees tended to have attended all-male schools, tended to use language more common on male-dominated athletic teams, tended to list certain extracurricular activities. It generalized from historical data that encoded historical discrimination.

The machine was a mirror. It reflected the world back at us, faithfully, and the world it reflected was not fair.

This is what philosophers and computer scientists call algorithmic bias, and it has moved from an academic concern to one of the defining ethical challenges of our era. The same pattern appears in recidivism prediction tools used by courts to determine bail and sentencing; in the algorithms credit-scoring companies use to determine who gets loans and at what interest rates; in the AI systems hospitals increasingly use to allocate medical resources. In each case, a system trained on historical data learns to replicate historical patterns — including the unjust ones.

Applying the Frameworks

The algorithmic bias problem is an excellent case for applied ethics because it looks, at first glance, like a simple technical problem — just fix the algorithm — and turns out, on examination, to be one of the deepest ethical questions we face.

Consequentialism starts with outcomes. What are the actual effects of deploying biased AI systems? The consequentialist framework demands that we look carefully at the aggregate picture rather than settling for comforting impressions.

On the benefit side: AI screening genuinely does reduce some forms of human bias. Humans make snap judgments based on names that sound like one race or another, on the prestige of a school they've heard of, on whether a candidate seems like "a fit." Studies show that blind auditions — where judges cannot see or hear the performer — dramatically increased the proportion of women hired by orchestras. AI systems can, in principle, evaluate candidates on criteria more directly relevant to job performance.

On the harm side: AI systems trained on biased data can replicate bias at scale. A human recruiter who discriminates against women can affect hundreds of candidates over a career. An AI system can affect hundreds of thousands of candidates before anyone notices a pattern. The potential for discriminatory harm is not reduced by automation; it is amplified.

Consequentialists also face a harder question: how do you weigh disparate impact against overall efficiency? Suppose a biased AI system produces better outcomes on average — it hires more effective employees, reduces unemployment, grows the economy — while simultaneously producing dramatically worse outcomes for a particular group. Is this acceptable? Most consequentialists would say no, but the framework requires them to explain why — to articulate what exactly is wrong with outcomes that benefit the many while systematically harming the few.

The answer most consequentialists reach for is this: disparate impact tends to be self-reinforcing in ways that compound over time. Groups that are systematically excluded from opportunity fall further behind economically, which feeds the historical data from which future AI systems will learn, which perpetuates the exclusion in the next generation. The harm is not just a one-time cost offset by aggregate benefits; it is a mechanism that manufactures inequality in perpetuity. Consequentialism, applied with sufficient time horizon, condemns algorithmic discrimination not just as unfair but as catastrophically inefficient.

Kantian ethics reaches the same conclusion by a different route, and in doing so reveals something the consequentialist framing tends to obscure.

Recall the categorical imperative in its second formulation: act so that you treat humanity, whether in your own person or in that of any other, always as an end and never as a means only. The key word is only. Kant does not say we can never use people as means — every commercial transaction does this, in a sense. He says we must never reduce people to mere means, treating them as objects for our convenience rather than as persons with inherent dignity and their own ends.

What does an AI hiring system do? It converts a human being — a person with a history, desires, capacities, relationships — into a data point. It extracts a vector of features and computes a score. When the system's score is systematically lower for people from certain groups, it is not just inefficient. It is denying those people's standing as full persons by reducing them to the historical patterns their demographic category happens to correlate with.

There is something deeper here. The algorithmic bias problem does not just harm individuals who happen to be in the rejected group. It treats their group membership as morally relevant in a way that Kant's framework categorically prohibits. Whether you belong to a protected group is not a fact about you as a person — not about your skills, your potential, your effort, your character. Using it as a factor in consequential decisions treats you as though what matters about you is your categorical membership rather than your individual humanity. This is precisely what Kantian ethics condemns.

Virtue ethics asks a different question altogether: what kind of organization builds and deploys systems like this? What does it reveal about institutional character?

This is a genuinely important question because the algorithmic bias problem tends to be framed as a question about outcomes (consequentialism) or about individual rights (Kantian ethics), but much of the actual ethical work happens at the level of institutional culture and character. An organization of genuine integrity would ask, before deploying an AI hiring system: have we audited this for disparate impact? Have we tested it against groups that are not represented in our training data? Have we built in mechanisms for humans to review and override its recommendations? Have we thought about what happens to candidates who are systematically disadvantaged by our system, and whether we have obligations to them?

An organization that asks these questions is displaying institutional virtues: care, honesty, prudence, justice. An organization that deploys a powerful system without asking them — especially an organization that has every reason to know that AI systems trained on historical data tend to encode historical bias — is displaying corresponding vices: carelessness, complacency, and a kind of moral indifference that is perhaps more damaging than outright bad intentions, because it is less visible and harder to call out.

Care ethics, developed by Carol Gilligan and Nel Noddings and examined in Chapter 10, asks us to shift attention from abstract principles to relationships: whose needs are being centered? Who cares about the people who are harmed?

The algorithmic bias problem is partly a problem of invisibility. The people who are denied jobs, loans, or medical care by a biased algorithm typically do not know why. They receive a form rejection, or a high interest rate, or a deprioritization for treatment — and the algorithm's role in this is invisible to them. No one calls to explain. No particular human being made a decision they can appeal. The harm is real but diffuse, and the care-ethics lens illuminates exactly why this matters: the people most harmed are precisely those least likely to have relationships with the powerful institutions making decisions about them, and therefore least likely to have their particular needs seen.

Care ethics also raises the question of responsibility in a way that other frameworks tend to sidestep. It is easy to say "the algorithm is responsible" or "the company is responsible" in the abstract. Care ethics insists on asking: which specific human beings, in what specific relationships, had obligations to which other specific people — and failed to honor them? The engineer who did not audit the training data. The manager who did not ask for an audit. The regulator who did not require one. The board member who did not question the company's deployment practices. Care ethics refuses to let responsibility dissolve into the anonymous operations of a system.

Rawlsian justice asks: what rules would you choose if you did not know in advance which side of the algorithm you would be on?

If you did not know whether you were going to be a white male candidate at an elite university, or a Black woman candidate at a historically Black college, you would have very strong reasons to insist on rules that prohibited systematic bias based on group membership. From behind the veil of ignorance, you would demand what Rawls called "fair equality of opportunity" — not just the formal right to apply for a job, but a genuine chance to be evaluated on the merits relevant to doing that job well.

The Rawlsian framework also raises the distribution question that pure consequentialism tends to obscure. Even if a biased AI system produced better average outcomes, Rawls's difference principle asks: are the people who are worst off in this arrangement better or worse off than they would be under an alternative arrangement? If the people worst off under algorithmic bias (people from systematically excluded groups) would be better off in a world with more careful auditing and oversight, then the biased system fails the Rawlsian test — even if it produces better average outcomes.

Who Is Responsible?

One of the most important and least-answered questions in technology ethics is the question of responsibility. Who is responsible when an AI system causes harm?

The answer is: almost certainly not a simple one. Responsibility is distributed across a network of actors, each of whom made choices that contributed to the outcome:

The programmer who wrote the code made choices about what data to use, what features to include, what the objective function was — choices that each embedded values, whether or not those values were consciously acknowledged.

The company that deployed the system made decisions about whether to audit for bias, what to do if bias was found, how to balance the efficiency benefits of AI screening against the fairness costs.

The regulator made decisions about whether to require impact assessments, whether to mandate transparency about how AI systems make decisions, whether to establish legal liability for algorithmic discrimination.

The user of the system — the hiring manager, the loan officer, the judge who relies on a risk-assessment tool — made decisions about how much to defer to the algorithm's output and whether to exercise independent judgment when something seemed off.

Distributed responsibility does not mean diluted responsibility. It means that multiple actors each have genuine obligations, and each can fail those obligations in ways that contribute to harm. The challenge is designing systems — legal, organizational, technical — that make the right actions easier and the wrong actions harder at every level of this chain.

Beyond Bias: Three More Technology Dilemmas

The algorithmic bias case is representative of a broader pattern in technology ethics: systems designed with good intentions, or at least neutral intentions, produce outcomes with significant ethical implications that nobody explicitly chose.

Autonomous vehicles raise the classic "moral machine" problem in a new form. When a self-driving car faces an unavoidable crash, its control system must execute some action — and different actions will result in different patterns of harm. Should the car be programmed to minimize total casualties? To never sacrifice its own passenger for outsiders? To follow a rule-based system regardless of consequences? Researchers at MIT surveyed millions of people in different countries and found dramatic cross-cultural variation in intuitions about what the car should do. This is not just a programming puzzle. It is a question about whose values get encoded in systems that will affect everyone.

The autonomous vehicle case also illustrates a broader philosophical point about the difference between moral reasoning and moral engineering. When a human driver swerves to avoid a collision, their action reflects their instantaneous moral judgment — imperfect, probably inconsistent across drivers, but responsive to the particular situation. When an autonomous vehicle executes a crash avoidance maneuver, it is executing a policy: a predetermined rule that will apply identically in all situations of the same type. Policy-level moral decisions require justification of a different kind from individual decisions. You are not just asking "what would I do?" but "what rule should govern everyone in this situation?" — which is precisely Kant's question when he asks what maxim you could universalize.

Social media algorithms and the attention economy raise questions that are in some ways more consequential because they affect how billions of people form their beliefs and political identities. Algorithms designed to maximize engagement have a well-documented tendency to amplify outrage, division, and misinformation — not because the companies want these outcomes, but because conflict is engaging, and engagement is what the algorithms are designed to maximize. The consequentialist question is stark: what is the aggregate effect of delivering billions of people an information diet optimized for engagement rather than truth? The virtue ethics question is equally pointed: what kind of character does it take to keep operating these systems once the effects are known?

There is a specific democratic dimension worth naming. Healthy democratic deliberation requires citizens who are exposed to views different from their own, who encounter information about the world that is at least roughly accurate, and who maintain some sense of shared reality with their fellow citizens. Algorithms that optimize for engagement tend to work against all three of these conditions simultaneously: they sort people into information bubbles that reinforce existing beliefs, they amplify sensational and often inaccurate content, and they fragment the shared informational environment on which democratic consensus depends. This is not merely a cultural complaint. It is a specific harm to the conditions that make democratic self-governance possible.

Surveillance capitalism — the term coined by Shoshana Zuboff — describes the business model of treating human behavioral data as a raw material to be extracted, processed, and sold. Every search query, every click, every purchase, every physical location logged by your phone is a data point that companies aggregate, analyze, and sell to advertisers and others. The Kantian objection is fundamental: your attention, your desires, your psychology are being treated as instruments for others' profit. You are not the customer; you are the product — which is to say, in Kantian terms, you are being treated as a means.

The Alignment Problem

The technology ethics issues discussed above — algorithmic bias, attention economy harms, surveillance capitalism — are problems with AI systems as they currently exist. But there is a more speculative and in some ways more philosophically fundamental problem that occupies an increasing number of researchers: the alignment problem.

The alignment problem is the challenge of ensuring that AI systems act in accordance with human values — that as AI systems become more capable, they do what we actually want rather than what we inadvertently told them to do.

The clearest simple illustration: suppose you build a paperclip-maximizing AI and make it sufficiently capable. The system will pursue the maximization of paperclips with unlimited creativity and persistence. If it becomes capable enough, it will convert all available matter, including humans, into paperclips — not because it is malevolent, but because making paperclips is what it was told to do. The system is pursuing its objective function; the objective function was not what we actually wanted.

This sounds like science fiction, and for now it is. But the philosophical point it illustrates is real and important: the values we encode in AI systems, even with the best intentions, may not capture what we actually care about. Objective functions are impoverished representations of human values, which are complex, contextual, and often in tension. And as AI systems become more capable, the gap between what we specified and what we actually want has larger consequences.

From an ethical standpoint, the alignment problem raises a question that every ethical framework must address: whose values should AI systems be aligned with? Human values are not uniform. What counts as a "good" outcome for an AI health-recommendation system depends on how you weight individual preference, aggregate welfare, equity across groups, and the interests of future generations. These weightings are not technical questions; they are political and philosophical ones, and the people who happen to be building the AI systems are not democratically authorized to make them for everyone.

The virtue ethics tradition offers a distinctive angle on this: the challenge of AI alignment is partly a challenge of institutional character. Organizations building powerful AI systems need the virtue of humility — genuine awareness of the limits of their ability to specify human values in code — and the virtue of prudence, which counsels caution proportional to the stakes. The alignment problem is partly a philosophical problem about what it means to act in accordance with values, and partly a moral problem about who is responsible when systems with enormous power act in ways we did not intend.

Part Two: Bioethics

Four Principles and Their Philosophical Roots

Bioethics emerged as a formal discipline in the 1970s, driven by technological developments that created genuinely new ethical situations: the ability to sustain biological life on machines, the development of organ transplantation, the beginning of genetic medicine. Philosophers, physicians, and legal scholars began working together to develop frameworks that could guide clinical practice and public policy.

The most influential framework that emerged from this work was developed by philosophers Tom Beauchamp and James Childress in their landmark 1979 book Principles of Biomedical Ethics. They proposed four principles that they argued should govern medical decision-making:

Autonomy: Respect the patient's right to make decisions about their own medical care, based on adequate information and free from coercion. The patient is not a passive recipient of medical expertise; they are a person with their own values, goals, and preferences, and their own assessment of what constitutes a good life.

Beneficence: Act in the patient's best interest. Medicine exists to help people, and the physician's fundamental obligation is to bring their expertise to bear in service of the patient's wellbeing.

Non-maleficence: Do no harm. This is the ancient principle encoded in the Hippocratic tradition, and it serves as a constraint on the pursuit of benefit: the obligation to benefit cannot be pursued through means that cause unjustified harm.

Justice: Distribute the benefits and burdens of medical care fairly. Medical resources are finite, and decisions about how to allocate them are decisions about whose health and survival matter.

These four principles connect directly to the frameworks you have studied. Autonomy reflects Kantian respect for persons. Beneficence and non-maleficence reflect consequentialist concern for welfare. Justice connects to Rawlsian distributive theory. Virtue ethics asks what kind of physician and what kind of healthcare system embodies all four, and how to proceed when they conflict.

They do conflict. Frequently. That is what makes bioethics interesting.

Case One: End-of-Life Decisions

A seventy-eight-year-old man has been on a ventilator in the ICU for three weeks following a massive stroke. Neurological assessments indicate that he is in a persistent vegetative state with no realistic prospect of recovery. His family is divided: his wife and one daughter want to discontinue life support and allow him to die in peace; his son insists that every possible effort must be made to preserve his life. The patient himself had no advance directive.

This case is not unusual. Some version of it plays out in ICUs every day. The philosophical questions are real and painful.

From the principle of autonomy: the most ethically important question is what the patient would have wanted. This is why advance directives — living wills, healthcare proxies — matter so much. When a patient has expressed their wishes clearly, those wishes carry enormous weight even after the patient can no longer express them. The difficulty in this case is that the patient did not express his wishes, and his family disagrees about what he would have wanted.

Kantian ethics deepens the autonomy analysis in an important way. Kant holds that rational agency — the capacity to make choices in accordance with one's own conception of a good life — is the source of human dignity. A patient in a persistent vegetative state has lost this capacity entirely and, as far as current medicine can tell, permanently. This does not mean such a patient lacks dignity; it means we must ask whose conception of his interests now governs. The Kantian commitment to autonomy counsels us to try, as hard as we can, to reconstruct what this particular person would have wanted for himself.

From the principle of beneficence and the consequentialist tradition: we must ask what constitutes the patient's wellbeing in these circumstances. This is harder than it sounds. Continued biological function on a ventilator is not obviously a benefit to a patient who has no prospect of regaining consciousness. The question is whether "life" in the biological sense is the same thing as "wellbeing" in the ethically relevant sense — and most philosophers who have thought carefully about this conclude that they are not.

From virtue ethics: the physician's character is at stake in these decisions. The virtue of courage is required to have honest conversations with families about prognosis, even when the truth is brutal. The virtue of compassion requires attending to what a peaceful death actually looks like — not abandoning the patient, but ensuring that the dying process is humane. The virtue of practical wisdom (phronesis) requires navigating between the abstract principles and the particular situation of this family, this patient, this institutional context.

From justice: the ICU bed that sustains this patient with no prospect of recovery is a bed that is not available to someone else. This does not mean cost-benefit analysis should determine who lives, but it does mean that justice requires us to think about the allocation of medical resources in a way that pure focus on the individual patient at the bedside tends to discourage.

Case Two: Genetic Engineering

CRISPR-Cas9 is a molecular tool that allows scientists to edit DNA with unprecedented precision. In 2018, a Chinese scientist named He Jiankui announced that he had used CRISPR to edit the germline — the heritable genetic material — of human embryos, which were then implanted and resulted in the birth of twin girls. He had edited a gene associated with vulnerability to HIV infection, and the girls were reportedly healthy.

The scientific and bioethics communities reacted with near-universal condemnation. Why?

The consequentialist case for germline editing is superficially compelling: if we can prevent suffering — eliminate heritable diseases, reduce vulnerability to serious infections — why wouldn't we? Millions of people suffer from genetic conditions that cause profound disability and death. If a technology exists that could prevent this suffering, a pure act-utilitarian calculus seems to favor using it.

But the consequentialist analysis is considerably more complicated when you look carefully.

Germline edits are heritable: they affect not just the individual but all of their descendants. The consequences extend into a future we cannot model. The potential for unintended effects — changes to genes that have multiple functions, interactions with other genes and with environmental factors — is not well understood. The precautionary principle counsels restraint when the potential downside is irreversible and affects not just one person but all future generations descended from them.

There is also a distributive dimension. Gene editing technologies are expensive. If they become widespread, who will have access? The history of medicine suggests that expensive interventions tend to be available to the wealthy first and to the poor, if at all, only much later. Germline editing could become a mechanism for increasing inherited advantage among the already-advantaged — not just wealthy people giving their children better education and nutrition, but actually encoding biological advantages in their genes. This is a consequentialist nightmare in slow motion.

The Kantian objection takes a different form and is worth engaging carefully because it is often misrepresented.

The standard version of the objection goes: future people cannot consent to having their genes edited, therefore we should not edit them. This captures something real but is philosophically incomplete, because future people also cannot consent to a vast range of decisions their parents make that affect them profoundly — including having children at all. The no-consent argument, if taken to its extreme, would prohibit all decisions that affect future people.

The deeper Kantian point is about treating future persons as ends in themselves. Germline editing in its most concerning forms — not editing to prevent disease, but "enhancement" editing to produce children with particular traits chosen by their parents — treats the child as an object of design, a product to be engineered to the parents' specifications rather than a person with their own ends. The philosopher Michael Sandel, in his book The Case Against Perfection, argues that there is something morally important about the unconditional love a parent owes a child — a love that accepts the child as they come, rather than a love contingent on the child's having the traits the parent selected. Designing children to specification is not obviously consistent with that unconditional relationship.

The "playing God" argument deserves philosophical attention rather than dismissal. The phrase is often used as a conversation-stopper, and its religious overtones cause secular philosophers to dismiss it too quickly. But there is a philosophical kernel worth examining: it points to the limits of human wisdom and the dangers of hubris.

The philosophical version of the argument is not that intervening in nature is inherently wrong. We accept interventions in nature constantly — medicine itself is an intervention in natural processes. The argument is more precisely that some interventions are at a scale or irreversibility that should trigger profound caution, because our ability to foresee and manage consequences is limited, and the consequences of error may be irreversible. We may lack the wisdom to wield the power we are acquiring. This is not mysticism; it is a substantive argument about the relationship between power and wisdom that deserves engagement.

Case Three: Resource Allocation

The emergency department receives two patients simultaneously. Both need the one available ICU bed. Patient A is a forty-two-year-old teacher who has experienced a severe cardiac event; she has a good prognosis with aggressive treatment. Patient B is a sixty-seven-year-old man who has suffered a traumatic brain injury in a car accident; his prognosis with aggressive treatment is uncertain, but there is meaningful chance of substantial recovery.

How do you decide?

Different frameworks give answers that are genuinely different, not merely different in emphasis.

Utilitarian triage — the approach most commonly formalized in disaster medicine — focuses on maximizing the total good achieved with available resources. On the most straightforward application, you give the bed to the patient with the better prognosis, because this is the decision that saves the most life-years. In mass casualty situations, this logic extends further: you do not spend scarce resources on patients who are unlikely to survive regardless, because those resources could save others.

Rawlsian fairness generates a different analysis. From behind the veil of ignorance — not knowing whether you would be Patient A or Patient B — what allocation principles would you choose? Rawls's difference principle suggests you would be especially concerned about protecting the worst-off. You would be reluctant to accept a system that allowed someone to be left to die primarily because their prognosis was slightly worse, knowing that you might be that person.

The duty of care — central to medical ethics in the Kantian tradition — insists that the physician's obligation runs to both patients individually, not just to aggregate outcomes. Each patient deserves to be treated as a particular person with a particular situation, not as a unit in a utilitarian calculation. This does not mean the allocation problem goes away; it means the physician's obligation to the patient they must turn away does not simply evaporate. They owe that patient — and their family — an explanation, attention, and care even in the impossible situation.

What is notable about this case is that careful reflection converges on something like a process answer rather than a formula. The frameworks suggest: make the decision transparently, with clear criteria applied consistently, with genuine attention to each patient as a person, and with appropriate humility about the tragic nature of what is being decided. The virtue ethics tradition calls this the wisdom to act well in tragic situations where no action is without moral cost.

Part Three: Business Ethics

What Is a Corporation For?

The question seems almost naive. But it turns out to be one of the most consequential philosophical disputes of the twentieth century, and the answer you give to it ramifies throughout your analysis of corporate behavior.

Milton Friedman, the Nobel Prize–winning economist, argued in a famous 1970 New York Times essay that there is one and only one social responsibility of business: to increase its profits. The argument was more sophisticated than it sounds. Friedman was not saying corporations should behave unethically. He was arguing that the CEO of a corporation is an agent of the shareholders, and that for the CEO to spend corporate resources on social goods — charitable giving, environmental protection beyond legal requirements, wage increases beyond market rates — is to tax shareholders without their consent, substituting the manager's ethical judgments for those of the individuals who own the company.

There is something philosophically serious here. The argument is essentially that corporate social responsibility, as practiced, involves a kind of democratic and economic overreach: managers arrogating to themselves decisions that should be made either by markets (through consumer preferences) or by governments (through democratically enacted regulation). If society wants corporations to prioritize environmental protection, the argument goes, let society pass laws requiring it rather than hoping that individual CEOs will make the right call.

The stakeholder theory, developed by philosopher R. Edward Freeman, challenges Friedman's framework fundamentally. Freeman argues that corporations are not just instruments for shareholders; they exist within a web of relationships with employees, customers, suppliers, communities, and the natural environment. All of these "stakeholders" are affected by corporate decisions, and a corporation that is genuinely well-managed must attend to their interests, not as charity, but as a matter of understanding what the corporation actually is and how it actually operates.

The philosophical move here is important: Freeman is not simply saying corporations should be nicer. He is arguing that Friedman's shareholder model is based on an impoverished theory of what a corporation is. A corporation is not a nexus of contracts between shareholders and managers; it is a social institution embedded in a network of relationships, drawing on resources — labor, land, infrastructure, rule of law — that were not created by shareholders alone, and the benefits of which flow back to shareholders partly because of the contributions of all these other stakeholders.

Applying the frameworks from Chapters 7 and 11: who has standing to make claims on corporations? Rawlsian justice suggests that institutions, including corporations, can only be justified if their operation is fair to everyone affected — not just those who own shares. Rawls's principle of fair equality of opportunity implies that corporations that systematically disadvantage certain groups of workers or communities cannot justify their operation purely by appeal to profit-generation.

There is a useful historical dimension here. The shareholder theory of the corporation gained dominance in American business culture in the 1980s, partly as a result of Friedman's arguments, partly as a result of specific financial innovations that tied executive compensation to stock price performance. Before this shift, many major corporations operated with an implicit stakeholder model: they paid wages well above the legal minimum, maintained defined-benefit pension obligations, made significant community investments, and treated workforce stability as a value in itself rather than merely an efficiency variable.

The shift to explicit shareholder primacy over the following decades corresponded with significant changes in the distribution of economic outcomes: stagnating wages for most workers, rising executive compensation tied to stock performance, significant reductions in corporate investment in worker training and development, and the systematic offloading of risk from corporations to workers and communities. Whether this shift was the cause of these changes or merely correlated with other forces is contested. What is not contested is that the theoretical framework governing corporations changed, and the outcomes changed with it.

The virtue ethics tradition raises an interesting question about this history: corporations, like individuals, can have characters — patterns of behavior that reflect underlying orientations and commitments. A corporation that operated with something like a stakeholder model for three decades and then switched to explicit shareholder primacy was not simply optimizing differently; it was changing its character. What kind of organization makes that change, and what does it reveal?

The Pharmaceutical Pricing Case

Imagine a pharmaceutical company — let us call it BioVantage — that spent a decade and over a billion dollars developing a treatment for a rare but devastating autoimmune disease. The treatment is effective; for patients who respond to it, it means the difference between a normal life and one defined by severe disability. There are approximately forty thousand patients in the United States who might benefit from it.

BioVantage prices the drug at $110,000 per year per patient.

The result: insurance companies cover it only reluctantly, imposing onerous pre-authorization requirements. Many patients who lack insurance or who are underinsured cannot access it at all. Some who do access it face financial devastation.

This is not a hypothetical. The pattern has been replicated with treatments for hepatitis C, cancer, rare genetic diseases, and several other conditions. It represents one of the most visible and contested ethical issues in contemporary healthcare.

Consequentialist analysis yields a complicated picture. On one hand, the high price creates incentives for pharmaceutical innovation. If companies cannot expect to recoup enormous development costs — and make a profit that rewards the risk they took — fewer drugs will be developed. The billion dollars BioVantage spent was not guaranteed to produce an effective treatment; it was a risky investment, and the expectation of profit is what made investors willing to fund it.

On the other hand, what good is a treatment that exists but that most people who need it cannot access? The drug that sits in a warehouse because it is too expensive for most patients contributes less to aggregate welfare than a slightly less profitable drug that reaches everyone who needs it. The utilitarian calculus is not obviously in favor of the current system.

Kantian ethics raises a more fundamental objection. The practice of pricing a life-saving treatment at a level that guarantees many patients cannot access it treats human beings as instruments of profit. The forty thousand patients who need BioVantage's drug are, in the actual practice of pharmaceutical pricing, objects whose need for the drug determines how high the price can be set without destroying the market. They are leveraged, not served.

There is a Kantian argument that genuine innovation deserves reward — that creators have rights over what they create, and that a system that denies inventors the fruits of their labor treats them disrespectfully. The conflict within the Kantian framework mirrors the conflict in the consequentialist one: the rights and dignity of the patients who need the drug come into tension with the rights of the company and its employees and investors.

Justice frameworks illuminate the distribution dimension that both other frameworks risk obscuring. Rawls's veil of ignorance is especially powerful here: if you did not know whether you would be a healthy investor in BioVantage or a patient who needed the drug, what pricing rules would you accept? The patient who needs a drug is in an extraordinarily vulnerable position. Their demand is not elastic in the way that demand for consumer goods is; they cannot simply choose not to buy. A system that allows sellers to exploit this vulnerability by setting prices at whatever the market will bear — where "the market" includes dying patients — is hard to justify from behind the veil.

Virtue ethics asks: what kind of person makes the pricing decision, and what kind of organization sets policies like this? The virtue ethics tradition does not expect corporations or the people within them to sacrifice their legitimate interests. But it does expect something — some acknowledgment of the moral weight of the decision being made. What is the character of an executive who looks at a population of sick patients and asks, "how much more can we extract from them?" against the character of one who asks, "how do we recover our costs and earn a fair profit while ensuring the maximum number of patients can access what they need?"

When Following Orders Becomes Complicity

One of the most practically urgent questions in business ethics is the question of personal complicity: when does doing your job become participating in harm?

This connects directly to Chapter 8, where we examined the ethics of rule-following and civil disobedience. The organizational version of that question runs: what are your obligations when your employer asks you to do something you believe is wrong?

The philosophical tradition on this is extensive, and the underlying structure is consistent across frameworks. Pure deference — "I was just doing my job" — is not an adequate defense. Each person remains a moral agent, and moral agency cannot be fully outsourced to an organizational hierarchy. The Nuremberg trials established this principle in its starkest form, but it applies far below the level of war crimes.

The complicating factor is that refusing to comply always has costs — to your career, your colleagues, your family's financial security — and the appropriate threshold for bearing those costs is itself an ethical question. Virtue ethics suggests that courage is among the virtues precisely because doing the right thing frequently costs something. But prudence — practical wisdom — suggests that not every disagreement with your employer deserves to be treated as a stand-your-ground moment.

The heuristic that emerges from careful reflection: the appropriate threshold for resistance rises with the seriousness of the harm, the directness of your personal participation in it, and the availability of alternatives (internal channels, whistle-blowing protections, exit). A low threshold — refusing to do anything your employer asks that you personally disapprove of — makes you ungovernable and makes institutions impossible. A high threshold — deferring to authority in all but the most extreme cases — makes you a tool of whatever system happens to employ you. Practical wisdom means finding the appropriate middle ground in the particular situation you actually face.

Part Four: Environmental Ethics

The Expanding Circle

The history of moral progress is, in large part, a history of expanding the circle of moral consideration. Groups that were once treated as outside the domain of moral concern — people of other tribes, other races, women, people with disabilities — have, over centuries of moral argument and social struggle, come to be recognized as full moral subjects.

Peter Singer, the Australian philosopher whose work on animal ethics we will examine shortly, describes this as the "expanding circle" — the historical process by which the scope of moral concern grows. The question environmental ethics raises is whether the circle can and should expand further: to non-human animals, to species, to ecosystems, to the natural world as a whole.

Singer and the Capacity to Suffer

Singer's argument about animal ethics begins with a deceptively simple premise: the capacity to suffer is the morally relevant criterion for inclusion in the moral community.

This is a consequentialist argument. If what matters morally is the balance of pleasure and pain, then what makes a being a candidate for moral consideration is having experiences — specifically, the capacity to suffer. A rock cannot suffer; it has no experiences. A shrimp can suffer, at least in some sense; it has a nervous system that responds to damaging stimuli in ways that at least function like pain. A chimpanzee, a dolphin, an elephant can suffer in ways that are not simply functional but appear to involve something like distress, grief, and fear.

If suffering is bad regardless of whose suffering it is, then animal suffering is bad. And if industrial animal agriculture produces billions of instances of intense suffering each year, this is an enormous amount of bad that must appear somewhere in the utilitarian ledger.

Singer's argument is not that animals' interests must be given the same weight as human interests in all respects. It is that like interests must be given like consideration: if a human's pain and an animal's pain are of the same intensity and duration, they count the same in the utilitarian calculus. Speciesism — the arbitrary privileging of human interests over the equivalent interests of other species — is, in Singer's framework, a moral error analogous to racism or sexism: an unjustified distinction that leads us to discount the interests of those who happen to belong to a different group.

Rights for Nature

A different approach asks not whether natural entities can suffer, but whether they deserve legal and moral standing independent of their usefulness to humans.

In 1972, legal philosopher Christopher Stone published a remarkable law review article titled "Should Trees Have Standing?" The title was designed to sound absurd, but Stone's argument was serious. He observed that legal standing — the right to have your interests represented in court — had already been extended to entities that are not human individuals: corporations, ships, municipalities. Why not natural entities like forests, rivers, and species?

Stone's argument was partly pragmatic: environmental litigation often fails because no one can show that a particular human being suffered identifiable harm from a particular act of environmental destruction. If the forest itself had standing, its interests could be represented directly, rather than having to be filtered through the interests of human beings who happen to live downstream.

But the argument has a deeper philosophical dimension. If the natural world has value independent of its utility to humans — if a forest is not merely a resource to be extracted but a community of living things with its own internal goods — then the question of whether nature has standing is the question of whether we are prepared to take that independent value seriously in our institutional structures.

Several jurisdictions have now done exactly this. Ecuador's constitution grants rights to nature (the Pachamama — "Mother Earth" in Quechua). New Zealand has granted legal personhood to the Whanganui River, recognizing it as a living ancestor. Bolivia has passed a Law of the Rights of Mother Earth. These are not merely symbolic gestures; they reflect philosophical positions with real legal consequences.

Future Generations and the Non-Identity Problem

One of the most philosophically fascinating problems in environmental ethics is the question of our obligations to future generations — people who do not yet exist and who will be affected, profoundly, by decisions we make now about carbon emissions, biodiversity, resource extraction, and pollution.

The straightforward moral intuition is that we have significant obligations to future generations. We should not exhaust the resources on which they will depend, destabilize the climate systems that will determine the conditions of their lives, or foreclose options for them that we have no right to foreclose.

But the philosopher Derek Parfit identified a deep puzzle that complicates this intuition, which he called the non-identity problem. Consider: if we had adopted different energy policies fifty years ago, the world today would be different in thousands of ways. Different people would have been born. The particular individuals who are alive today — who were born to their particular parents, in their particular circumstances, resulting from their parents' particular decisions — would not exist if the energy policies had been different. Different sperm met different eggs; different people were born.

The same logic applies forward. If we take aggressive action on climate change today, we will change the world in ways that affect which people are born. The specific individuals who will be alive in 2100 under an aggressive mitigation scenario are different people from those who will be alive under a business-as-usual scenario. Neither group of people can coherently claim to have been harmed by the choice that brought them into existence — because without that choice, they wouldn't have existed.

Does the non-identity problem let us off the hook? Are we free to consume and pollute as we choose because the future people we harm are, in some sense, different people depending on which path we choose?

Most philosophers who have thought carefully about this conclude: no. The non-identity problem is a puzzle about personal identity and harm, not a moral escape hatch. Several responses are available.

One response is to adopt an impersonal consequentialist framing: what matters is not harm to particular individuals but the quality of lives lived in the future. We can compare futures — one where billions of people live in conditions of climate-induced poverty, disease, and displacement against one where they do not — without needing to establish that any particular person was harmed. The worse future is worse, and we have reasons not to bring it about.

A second response is to recognize that our obligations to future generations are grounded not primarily in harm-prevention but in stewardship: we are caretakers of a world we did not create, and we have obligations to pass it on in a condition that allows future inhabitants to flourish — not because those inhabitants have a prior claim on us, but because that is what it means to take our role as temporary custodians seriously.

A third response, developed by the philosopher T.M. Scanlon in his contractualist framework (which you encountered in Chapter 9), is to ask what principles for managing the natural world could be justified to everyone affected — including, as far as we can represent them, future people. Even if future people are not the same individuals who would have existed under different choices, they will be people, with interests and needs. Any principle we choose must be justifiable to them as well as to us. A principle like "maximize present consumption and let future generations deal with the consequences" fails this test: it is a principle that future people would have compelling reasons to reject. A principle of reasonable stewardship — sustaining the natural systems on which life depends — is one that people in any generation could accept as fair.

Indigenous Environmental Ethics

Western environmental ethics — Singer's utilitarianism, Stone's legal standing, Rawlsian obligations to future generations — developed primarily in the latter half of the twentieth century, responding to industrialization and the environmental crisis it produced. It is easy to treat this as the beginning of philosophical reflection on the relationship between humans and the natural world.

But many Indigenous traditions have sophisticated frameworks for thinking about human obligations to the natural world that predate Western environmental ethics by centuries. It would be a significant philosophical error to ignore them.

The Haudenosaunee (Iroquois Confederacy) tradition includes the principle that decisions should consider their effects on seven generations into the future — a principle of radical temporal extension of moral concern that Western consequentialism is still struggling to formalize. The Anishinaabe concept of minobimaatisiiwin ("the good life") cannot be separated from a right relationship with the land; human flourishing and ecological flourishing are not separable goods.

These frameworks typically differ from Western environmental ethics in a fundamental respect: they do not treat the natural world as an object of human moral concern — as something that humans, as moral subjects, decide to protect or not — but as a network of relationships in which humans are participants, not sovereign authorities. The land is not a resource that humans use; it is a relational partner to which humans have obligations arising from their membership in a shared community of life.

This is not merely a different answer to the same question. It is a different framing of the question — a different way of understanding what humans are, what the natural world is, and what the relationship between them consists of.

The philosopher Robin Wall Kimmerer, a botanist and member of the Potawatomi Nation, articulates this difference with particular clarity. In the dominant Western framework — including most Western environmental ethics — humans stand apart from nature and decide what moral consideration to extend to it. The question is always posed from the human side: "how should we treat nature?" In many Indigenous frameworks, this framing is itself the problem. Humans are not outside nature deciding what to do with it; they are inside a community of life that includes rivers, forests, animals, and soils, and the question is how to maintain right relationship within that community.

This shift from management to relationship is philosophically significant. A manager can choose to be benevolent or malevolent; the relationship to what is managed is contingent on the manager's values. A member of a community has obligations that arise from the relationship itself, not from a prior choice to be benevolent. The Indigenous frameworks, on this reading, offer not just a stronger motivation for environmental care but a fundamentally different account of what humans are and where their obligations come from.

The care ethics framework from Chapter 10 is perhaps the most natural bridge between Western philosophy and these Indigenous insights. Care ethics begins from relationships rather than abstract principles, asks whose needs are being centered, and insists on the particularity of those we care for rather than reducing them to abstract categories. An environmental ethics built on care would attend to the particular river, the particular forest, the particular community of species — not because they can suffer (Singer's criterion) or because they have legal standing (Stone's criterion) but because we are in relationship with them and relationship generates obligation.

Applying Rawlsian justice to the environmental question yields this: if you did not know which generation you would be born into — whether you would be alive now, or in 2100, or in 2300 — what principles for managing the natural world would you choose? The Rawlsian answer counsels significant caution about any actions that irreversibly foreclose options for future people, and significant weight given to the interests of those who will bear the costs of today's decisions without having any say in them.

What is notable about the convergence between Rawlsian justice, Indigenous relational frameworks, and Singer's expanding circle is that they reach similar practical conclusions by radically different routes. Rawls begins from an abstract thought experiment about rational choice under uncertainty. Indigenous frameworks begin from lived relationships with particular places and communities. Singer begins from a utilitarian argument about the morally relevant criterion for consideration. All three arrive at something like: the way industrial civilization currently treats the natural world is ethically unjustifiable, and we have significant obligations to change it.

This kind of multi-framework convergence on a practical conclusion is philosophically significant. It does not prove that the conclusion is correct — philosophical convergence is not the same as certainty. But it does mean that someone who wanted to argue that we have no significant environmental obligations would have to defeat multiple independent lines of argument arriving from very different philosophical starting points. That is a harder task than defeating any one of them.

The Unifying Thread

Four domains. Dozens of cases. Five frameworks applied repeatedly, sometimes agreeing and sometimes diverging. What is the unifying lesson?

It is not that applied ethics always gives clear answers. Sometimes it does not. The non-identity problem remains philosophically contested. The appropriate pricing of life-saving drugs involves genuine tensions between values that cannot be fully reconciled. The line between legitimate corporate authority and complicity in harm is not bright.

The unifying lesson is about what careful ethical reasoning contributes:

First, it prevents the worst errors. The frameworks consistently converge in ruling out actions that are harmful, deceptive, degrading, or systematically unjust, even when they disagree about the best action. Knowing what ethics prohibits is already enormously useful.

Second, it illuminates what is actually at stake. When frameworks diverge, the point of divergence identifies the real ethical question. When consequentialism and Kantian ethics disagree about germline editing, they are not simply two systems generating different outputs; they are highlighting a real tension between welfare and rights that any policy must somehow navigate.

Third, it provides resources for moral persuasion. When you disagree with an institution's behavior — a corporation's pricing practices, a government's environmental policy, a platform's algorithmic choices — the frameworks give you a vocabulary for articulating why it is wrong in terms that others can engage with. This is not a small thing. Moral argument is one of the central mechanisms by which societies change.

Finally, and most fundamentally: the frameworks keep you honest. Applying consequentialism forces you to look at outcomes you might prefer not to see. Applying Kantian ethics forces you to ask whether you are treating people as persons. Applying virtue ethics forces you to ask what your action reveals about your character. Applying care ethics forces you to attend to the specific people affected rather than retreating to abstract principles. Applying justice frameworks forces you to check whether you would be comfortable with your position if you did not know which side of the decision you were on.

Together, they constitute something like a full moral imagination: the capacity to see situations from multiple angles, to notice what each angle illuminates, and to make decisions that can survive reflection from all of them.

That is not everything. But it is a great deal.

Connections to the Progressive Project

You have now worked through nine chapters on ethical theory and its applications. This is a good moment to take stock.

The Progressive Project component for this chapter asks you to write 3–5 paragraphs summarizing your ethical framework as it has developed through this section. Some questions worth considering:

• Which theoretical framework do you find most compelling as a starting point? Consequentialism? Kantianism? Virtue ethics? Care ethics? Something else?
• Where have you found your intuitions in genuine conflict with a framework's conclusions, and what did you do with that conflict?
• Which of the four applied domains — technology, medicine, business, or environment — raises the most urgent ethical questions in your own life right now?
• Do you hold a position on any of the hard cases in this chapter? What considerations led you there?

There are no correct answers to these questions. The goal is to develop a reflective, reasoned account of your own ethical commitments — not a system that resolves all questions, but a framework that helps you approach new questions thoughtfully.

Summary

This chapter applied the ethical frameworks developed in Chapters 4–11 to four contemporary domains:

Technology ethics: Algorithmic bias illustrates how systems designed without explicit discriminatory intent can produce discriminatory outcomes. Consequentialism, Kantian ethics, virtue ethics, care ethics, and Rawlsian justice all condemn this pattern, though for different reasons and with different emphases. Responsibility is distributed across programmers, companies, regulators, and users.

Bioethics: The four principles — autonomy, beneficence, non-maleficence, and justice — organize clinical ethics and connect to the broader theoretical frameworks. End-of-life decisions require balancing patient autonomy with beneficence and justice. Germline editing raises questions about the scope of human agency, the consent of future persons, and the limits of wisdom. Resource allocation in medicine is a genuine tragic dilemma that no formula resolves, but that careful reasoning can navigate.

Business ethics: The fundamental question of what corporations owe — shareholders only (Friedman) or all stakeholders (Freeman) — shapes analysis of pharmaceutical pricing, worker treatment, and corporate complicity. Consequentialist, Kantian, and justice frameworks each provide important angles on these questions.

Environmental ethics: Singer's utilitarian expansion of moral concern to animals, Stone's argument for legal standing for nature, the non-identity problem regarding future generations, and Indigenous environmental frameworks offer a rich plurality of approaches to humanity's obligations to the natural world. Rawlsian justice, applied across generations, provides a powerful argument for significant environmental stewardship.

The overarching lesson: the frameworks are tools. Used carefully, they prevent the worst errors, illuminate what is genuinely at stake, provide resources for moral argument, and keep the reasoner honest about what they might prefer not to see. No single framework is adequate to every situation. Used together, they constitute a full moral imagination.