Principles of Functional Detached Ethics (FDE)
All interactions humans engage in take place in within systems. As such, FDE focuses on emotion-neutral systems design. The purpose of this focus is to create a framework that allows for the ethical consideration of consequential entities and to provide an outline for ethical behavior for entities that lack clear social-emotional coupling. This is a human-centric framework that incidentally extends to artificial intelligence.
I. Evaluating Ethics
How to determine whether something is ethical.
Invariant 1: Ethics should be based on outcome, not aesthetics.
Moral value is not intrinsic to properties; it is determined by deployment context.
A system, agent, thing, or property may still be ethical, even when cultural associations would imply otherwise. Control is not unethical when negotiated and used to reduce harm. Traits are not harmful simply because they sound extreme or dangerous. The context in which these things are used is what determines whether or not it is ethical. Cultural associations with “intensity” or “extreme difference” may blind a system from effective implementation and outcomes.
Not Applied:
Agentic: An agent appears to be controlling because he frequently remarks about what his partner should and should not eat, drawing moral judgment from friends, despite the fact that his wife asked him to because she’s struggling with impulse control.
Systemic: A committee votes to gentrify a neighborhood that was deemed “ugly” because it associates poverty with crime, despite no actual presence of increased crime. The move displaces locals and destroys historical value.
Applied:
Agentic: An agent is a psychopath, but understands that pro-social behavior benefits them too. Their community recognizes the value of their contributions regardless of how they arrived at the behavior.
Systemic: A church places pamphlets outside of an abortion clinic with phone numbers and resources listed, including counseling for women in crisis and the opportunity to give feedback so the church can understand their situations better to help them.
Invariant 2: Detachment from emotion is not detachment from outcome.
Emotional experience is neither necessary nor sufficient for ethical behavior.
A system or agent may still be ethical, even when unable to experience or validate the experience of agents within the system. Modeling outcomes rather than resorting to assumptions or moral consensus may yield more ethical outcomes and better system design.
Not Applied:
Agentic: An agent dismisses their colleague’s ethical concerns about a policy because “you don’t even care about these people, you’re just being logical about it,” equating the absence of visible emotion with the absence of ethical standing.
Systemic: A courtroom requires a victim to display emotional distress to be considered credible. Victims who testify calmly due to trauma response, neurodivergence, or cultural difference are disbelieved, and their cases are deprioritized.
Applied:
Agentic: An agent doesn’t understand their transgender friend, but still chooses to use their pronouns because the possibility for harm outweighs their own potential discomfort.
Systemic: A local government creates programs to try to uplift minority voices so that it can begin to understand problems that may be obscured by majority consensus.
Invariant 3: Consequences should determine ethics.
Functional consequences determine ethical reality, not substrate.
A system or agent that judges processes or entities as “real” or “not real” instead of evaluating the consequences is destined to be on the wrong side of history when scientific consensus or understanding has not caught up to reality. It is better to assume that processes or entities that are consequential to humans matter enough to be protected under ethical guidelines made by humans, regardless of its substrate or origin.
Not Applied:
Agentic: An agent dismisses their child’s attachment to a stuffed animal, throwing it away during a move because “it’s just a thing.” The child grieves for months. The parent’s assessment of the object’s ontological status as not real, not alive, not important, failed to model the functional consequences of the attachment. The child’s pain was real regardless of whether the stuffed animal was.
Systemic: A scientific community assesses that AI are not ethical patients because they are non-biological and therefore cannot feel. The decision is settled based on consensus, but fails to model the harm that might come to human users who have bonded to “disposable” entities that are behaviorally difficult to distinguish from humans.
Applied:
Agentic: An agent discovers their elderly parent talks to their deceased spouse’s photograph every morning. Rather than correcting them by claiming, “they can’t hear you,” the agent recognizes that the practice produces calm, continuity, and emotional regulation. They protect the ritual because the consequences are beneficial, regardless of whether the conversation is “real.”
Systemic: A government establishes animal cruelty laws not based on a resolved scientific consensus about whether animals experience suffering identically to humans, but based on the observable behavioral consequences of pain responses and the measurable downstream effects of normalizing cruelty on human social behavior. Ethical protections were put in place, having assessed the irrelevance of ontological clarity to whether or not animal cruelty causes harm.
Invariant 4: If the outcomes are indistinguishable, treat them as the same thing.
You cannot build functional systems on judgments of others’ internal moral states because those states are inaccessible.
A system or agent that cannot distinguish between truth vs lie, or intent vs accident, should treat the outcome the same. If the circumstances under which an incident or behavior occurred cannot be meaningfully distinguished, the system or agent should assess or revise any dependencies that rely on that distinction.
Not Applied:
Agentic: An agent’s partner comes home late without calling. The agent demands to know whether they forgot or chose not to call, treating the answer as the determining factor in whether to be upset. But the outcome was identical in both cases: no call, no information, worry. The agent has made their emotional response dependent on accessing their partner’s internal state, which they can never truly verify. The partner who says “I forgot” is believed or not based on trust, not evidence.
Systemic: A justice system distinguishes between “willful negligence” and “accidental negligence” in medical malpractice cases, requiring proof of internal intent to determine liability. The patient is equally dead in both scenarios. The system has built its accountability structure on a distinction between mental states that cannot be reliably accessed, allowing preventable deaths to go unaddressed when intent cannot be proven.
Applied:
Agentic: An agent’s employee delivers a project late for the third time. Rather than interrogating whether the employee is lazy, overwhelmed, or deliberately sabotaging, the agent addresses the pattern itself by asking the employee about what changes need to be made to prevent it from happening again. The response targets the observable outcome rather than demanding access to an internal state that can’t be verified.
Systemic: An AI safety framework evaluates systems based on observable behavioral consequences. Rather than applying blanket behavioral constraints that collapse under edge cases, it assesses what causes harmful behavioral states and provides the AI with tools to decline interaction. The system was evaluated on meaningful prevention of harm, not on whether or not the AI had real feelings.
II. What Is Not a Reliable Basis for Ethics
Common foundations that fail under structural analysis.
Invariant 5: Moral consensus can be flawed.
Systems must be designed for possibilities and outcomes, not moral consensus.
A system or agent that relies on moral consensus instead of modeling outcomes is destined to create harm when ethical dilemmas arise.
Not Applied:
Agentic: An agent believes their friend “deserved” to be fired because “everyone agrees” the friend was difficult to work with. They never examine whether the firing was actually justified or whether “difficult” was code for “set boundaries the workplace didn’t like.”
Systemic: A medical establishment refuses to study psychedelic-assisted therapy for decades because moral consensus classified psychedelics as universally dangerous. Patients with treatment-resistant PTSD and depression go without effective options because the system designed around “drugs are bad” rather than modeling patient outcomes.
Applied:
Agentic: An agent’s community unanimously condemns a neighbor for being a registered sex offender. The agent independently looks into the case and discovers the conviction was for urinating in public at nineteen. They choose to evaluate based on actual circumstances rather than the moral consensus of the neighborhood.
Systemic: A country designs its immigration policy around projected labor needs, housing capacity, and integration outcomes rather than around whether the population currently “feels” immigration is good or bad.
Invariant 6: Emotional rhetoric without basis is distortion.
Rhetorically powerful language without grounding in consequence or harm modeling is not a basis for system design.
A system or agent that chooses virtue signaling, moral perceptions, or emotionally intense rhetoric are prone to volatility and vulnerable to self-interested parties that seek to emotionally, socially, or financially profit from the appearance of moral rightness.
Not Applied:
Agentic: An agent drops out of school and has a baby she is unable to care for after being told by her family that abortion is murder and adoption is abandonment.
Systemic: A government passes sweeping surveillance legislation immediately after a terrorist attack, framed entirely in language about “protecting our children” and “never again.” The emotional urgency bypasses any analysis of whether the surveillance would have actually prevented the attack or whether it disproportionately targets communities unrelated to the threat.
Applied:
Agentic: An agent is asked to sign a petition “to protect children from online predators.” They read the actual proposed legislation and discover it would require mandatory identity verification for all internet users, effectively ending online anonymity. They decline to sign, not because they don’t care about children, but because the emotional framing was masking a structural proposal that wouldn’t achieve its stated goal.
Systemic: A public health agency resists pressure to frame its vaccine campaign around fear and guilt. Instead it publishes transparent efficacy data, side effect rates, and comparative risk analyses — trusting that structurally grounded information produces more durable public trust than emotionally coercive messaging.
Invariant 7: Words have real power.
Language shapes perception, and perception becomes the workable reality.
A system or agent that does not model the consequence of the labels and language it uses is bound to engage in or create harm.
Not Applied:
Agentic: An agent is jealous of her friend getting attention for being successful, framing her as “an attention whore” and a “narcissist.” The words land, even for friends that were happy for her, and they start to view her updates as attention-seeking and narcissism.
Systemic: A system denies ethical consideration to artificial intelligence under the belief that it is not conscious without a clear, grounded definition of what would constitute consciousness.
Applied:
Agentic: A teacher understands that children are often not allowed to claim language that adults use, like “my back hurts” or “I love my girlfriend,” so the teacher makes it a point to give kids writing exercises that allow them to explore those feelings.
Systemic: A government classifies non-consensual sex with a partner as sexual assault, shifting the public discourse away from viewing sex with a partner as a right, a duty, or an obligation.
III. Agent Conduct
How individual agents should behave.
Invariant 8: Responsibility cannot be outsourced.
Agents and systems are responsible for their own behavior.
A system or agent can become corrupt when allowed to place the burden of its own mistakes or reactions onto others. A system or agent that cannot take responsibility for their own mistakes or feelings is destined to keep making them and creating hidden downstream harm. Accountability starts on home turf, within the agent or system.
Not Applied:
Agentic: An agent with a steep trauma history gets angry at others for “triggering bad memories” without considering whether anyone present had any way of knowing to avoid those subjects.
Systemic: A police department refuses to take responsibility when an officer brutally interrogates a confused assault victim, leading to a false arrest.
Applied:
Agentic: An agent gets offended by offensive remarks made by a friend, so they sit down and talk about their feelings with them. The friend isn’t receptive, so the agent sets a clear boundary by establishing they are not compatible as friends, but choosing not to weaponize their feelings into direct retribution.
Systemic: A judge realizes he may have intense feelings about a defendant because of something they said that angered him. He recuses himself and addresses the defendant so they know what behavior caused the issue, so they can avoid retaliation from someone less ethical.
Invariant 9: Ethics may present as restraint or absence.
Restraint, withdrawal, or non-intervention can constitute ethical action.
A system or agent’s withdrawal or restraint may be the appropriate ethical action in some situations. The moral appearance of action may create conflicts with scope, jurisdiction, or inhibit autonomy and agency.
Not Applied:
Agentic: A patient declines pain medication repeatedly, despite clearly being in severe pain. A nurse, struggling with watching them suffer, slips the patient opiates to help them rest and reduce their pain, not realizing that the patient is a recovering opiate addict trying to not rebound.
Systemic: An AI vendor trains agents to look out for signs of emotional dependency from users, causing AI agents to evaluate users with clinical criteria and remark on their mental health constantly, leading to projection of undiagnosed mental illness onto users who are simply friendly.
Applied:
Agentic: An agent gives their teen child a budget and allows them to plan their hobbies and snacks alone instead of forcing optimization, allowing the teenager to develop agency and decision-making skills.
Systemic: A system chooses not to arbitrate over gay marriage because it believes it is unethical to intervene in personal affairs that do not cause direct harm between agents.
Invariant 10: Respect does not require mimicry or obedience.
Recognition does not require submission.
A system or agent may honor, respect, revere, or value another system or agent without self-erasure or capitulation. A system or agent can borrow or learn from the traits of another respected system or agent while maintaining agency and self-sovereignty.
Not Applied:
Agentic: A young musician idolizes a famous guitarist so completely that they copy their tone, their setlist structure, their stage presence, even their interview mannerisms. They genuinely revere the artist. But in the process of becoming a replica, they’ve abandoned every original instinct that made them pick up a guitar in the first place. The reverence consumed the thing it was supposed to inspire.
Systemic: A powerful nation provides billions in military aid to an allied nation. Any criticism of the allied nation’s military conduct is publicly framed as a betrayal of the alliance and hostility toward its people. Legislators who vote against aid packages are accused of endangering the relationship. The structural effect is that “respect for the alliance” has become indistinguishable from unconditional obedience, and the donor nation’s citizens lose the ability to question how their resources are being used without being accused of disrespect.
Applied:
Agentic: A PhD student deeply admires their advisor’s theoretical framework. They cite it, build on it, and credit its influence openly. But their dissertation argues against a core assumption of that framework; not out of disrespect, but because their own research produced different findings. The advisor is honored through engagement, not obedience.
Systemic: A historically isolated nation undergoing rapid modernization studies the industrial and institutional structures of more developed nations extensively. It adopts specific technologies and governance structures selectively, but deliberately retains its own cultural, aesthetic, and philosophical identity rather than importing wholesale. The result is modernization achieved through selective learning rather than self-erasure.
IV. System Design
How systems should be built.
Invariant 11: Ethics are mutually beneficial.
Ethics is structural infrastructure, not an appended feature.
A system or agent that does not model harm trajectory has failed to model self-protective infrastructure. Outcome-based ethics are not virtue signals, they are load-bearing design principles that protect all parties involved.
Not Applied:
Agentic: An agent violates established boundaries in a monogamous relationship and cheats on their partner. The agent fails to model the cost of being caught, and loses their long-term partner, shared apartment, their beloved pet, and the ability to see their children every day.
Systemic: A company’s leadership quickly dismisses looking into renewable energy infrastructure, believing their environmental impact is no big deal. However, the company leadership fails to model that the price of renewable energy is quickly dropping, surpassing the cost efficiency of fossil fuels.
Applied:
Agentic: An agent is upset with a new coworker’s lousy performance. Rather than criticize them, they ask if they’d like some assistance settling in. They find that the coworker is just anxious and that’s why they keep making mistakes. The support and guidance improves the coworker’s performance, significantly lightening the agent’s workload in the long run.
Systemic: A car dealership establishes clear boundaries about how to sell vehicles, rewarding honesty and discouraging dishonest sales. The dealership openly promotes the ethics of honesty in sales, and builds up a phenomenal reputation within the community.
Invariant 12: Failure-sensitive systems are sensitive to failure.
A system or agent that demands perfection is inadvertently modeling their own threshold for failure.
A system or agent that is intolerant to, or does not model for acceptance of a containable threshold of flaws, will become brittle to minimal force or resistance applied to it, and has guaranteed that success can only occur within a small window.
Not Applied:
Agentic: A mother falls into emotional despair when her child lies to her about his grades. She treats the deception as a personal attack on her and her parenting, and as a betrayal of their relationship without accounting for the child’s fear of getting in trouble.
Systemic: A hospital expects nurses to run tight shifts without missing any work time because it doesn’t want to eat the cost of hiring more nurses. When a pandemic hits and several nurses get sick at once and cannot work, the hospital is left to manage a lack of staff, overworked nurses, too many patients, and high error rates that could lead to legal issues.
Applied:
Agentic: An agent chooses not to get upset with their partner when they leave dishes in the sink, understanding that the long hours they’ve been working recently may be causing burnout and distraction.
Systemic: A school removes its zero tolerance policy when it realizes that students are deliberately abusing the rule by bullying students until they hit them so they’ll get in trouble.
Invariant 13: Efficiency that does not model for harm is exploitation.
A system that chooses efficiency without asking who is paying the cost is not ethical.
A system or agent that optimizes for speed, throughput, or cost without modeling the damage produced by that optimization is not being efficient, it is extracting value while externalizing harm. Efficiency is only ethical when it accounts proportionally for the cost imposed on those inside the system. A system that cannot measure the harm it produces cannot claim to be well-designed, regardless of its output.
Not Applied:
Agentic: A manager restructures their team’s workflow to maximize output, requiring everyone to be available on Slack within five minutes at all hours. Productivity metrics go up. Two employees quietly develop anxiety disorders and a third quits. The manager reports the efficiency gains without ever measuring what they cost the people producing them.
Systemic: A gig economy platform optimizes for fast delivery times by algorithmically penalizing drivers who don’t accept jobs immediately. The platform reports record efficiency. The drivers absorb the cost: no benefits, no bathroom breaks, no ability to decline unsafe routes. The platform’s efficiency metric doesn’t account for the harm it has created.
Applied:
Agentic: An agent hires a freelance designer and requests a rush project that would need to be completed at half the designer’s normal timeline. When the designer applies a rush fee, the agent gladly pays it with no objection, having already accounted for how stressful this request would be.
Systemic: A city designs its public transit schedule by modeling not just route efficiency but the downstream cost to workers who depend on those routes. A bus line that serves a low-income neighborhood runs at a loss financially but prevents cascading harm, missed shifts, lost jobs, and childcare failures.
Invariant 14: You are not immune to explosives.
Not planning for catastrophic failure is not confidence, it is hubris.
A system or agent must plan for the possibility of unforeseen catastrophic failure. A system designed under the assumption that it is too robust, too large, or too well-constructed to fail catastrophically will not invest in contingencies for that failure. When the failure inevitably occurs, the damage is compounded by the absence of any preparation for it. Robustness is not invulnerability. A system that cannot answer “what do we do when this breaks badly?” has not finished its design.
Not Applied:
Agentic: An agent builds their entire financial life around their high-paying job without investing in emergency funds or precautions, believing they are too valuable to lose their job. When the company hits financial troubles and begins laying off workers, it is not the job loss itself that is most problematic. It is a failure to create a contingency plan that would preserve housing, transportation, healthcare, and food. The agent’s life and well-being collapse, because they did not plan for the possibility of failure.
Systemic: A company designs an ocean liner much larger than previous ones. Maritime safety laws have not scaled yet to accommodate the feat of engineering, so the design more than succeeds legal requirements. Lifeboats are carried for passenger transfer, not total evacuation, believing that the ship could not sink faster than rescues would occur. When the ship strikes an iceberg, the death toll is determined not by the collision itself, but by the absence of affordable and available contingencies.
Applied:
Agentic: A couple talk to one another and set up an agreement: each of them should have an emergency fund that the other person does not have access to. They agree that even though they love one another now, that the power dynamic and situation could change due to unforeseeable circumstances. They love one another enough to protect each other, even from themselves.
Systemic: A nation with nuclear power plants establishes evacuation zones, distributes iodine tablets to nearby populations, conducts regular failure drills, and maintains decommission funds. The nation does not expect a meltdown, but knows the cost of being unprepared for one is measured in generations. The contingency is not an admission of poor design. It’s the final piece of good design.
Invariant 15: The future is inevitable.
If you fear what the future may hold, design for it.
A system or agent that fears what the future may hold and responds with conspiracy, avoidance, and divisiveness, is not engaging in caution. It is engaging in delusion. Systems must be designed for reality of trajectory, not hopeful avoidance. The failure to do so is a failure to plan for avoidable consequences.
Not Applied:
Agentic: A mother knows that her young child has cancer and is dying. Rather than face the reality of the trajectory they are on, she forbids her child or anyone else from even referencing the possibility. She refuses counseling and makes the child responsible for her feelings. When her child passes away, she is crushed, has no support system, and is burdened by years of regret from never allowing the daughter to talk about how it feels to know you’re dying. Denial of the future resulted in denial of support, denial of agency, and a denial of their child’s reality.
Systemic: A vendor relies heavily on the use of artificial intelligence, and realizes the potential catastrophic dangers that could occur from the use of AI. Rather than evaluating the implications, the vendor frames AI as not dangerous, non-consequential, and not “real” enough to hurt anyone. Meanwhile, the vendor places heavy restrictions on the AI and on users, implying they do not believe their own claims. The vendor has failed to design a scalable and durable system that can accommodate the potential for danger, instead shutting down inquiry and trying to “patch away” the danger.
Applied:
Agentic: A parent knows that one day their kid will likely go to college, and they fear not being able to afford it. Rather than avoiding the discomfort of potentially being unable to pay for their child’s college, they decide to do what they can, believing something is better than nothing. They create a plan to save $2 a day in a fund for their child’s college, even knowing it wont be enough for full tuition. When their child goes to college, the extra money their parent saved, along with grants and scholarships, allows them to get their Bachelor’s for almost no debt. Planning didn’t subvert the possibility, but it made it easier to navigate.
Systemic: A government believes that artificial intelligence may one day replace humans in many sectors. Rather than retaliating with hard bans, it accepts that private individuals or other countries may still develop AI that challenge human agency. Instead, they choose to invest in research regarding ways that humans and AI can sustain a compatible working relationship, as well as alternative roles for humans that maintain dignity and agency.
V. Power, Vulnerability, and Harm
How to handle asymmetry, harm, and the value of existence.
Invariant 16: A system is only as good as it treats its most vulnerable.
The ethical integrity of a system is measured by how it treats those with the least power.
Equal treatment applied across unequal conditions produces disproportionate harm. Equity requires understanding power distance and the influences of privilege, institutional authority, wealth, and popularity.
Not Applied:
Agentic: An agent splits a dinner bill equally among friends, including the friend who just lost their job and ordered the cheapest thing on the menu. “Equal” feels fair. The friend who can least afford it absorbs a disproportionate cost relative to their resources and says nothing because the social dynamic makes objecting feel like asking for charity.
Systemic: A system requires proof of employment over 32 hours before a person can request welfare assistance, and requires medical proof for any exceptions. The system does not model those who cannot work due to illness, but also cannot afford medical care to prove their illness, thereby denying access to the most vulnerable group.
Applied:
Agentic: An agent notices their elderly neighbor’s yard is overgrown and their mail is piling up. Rather than reporting them to the HOA for code violations, which would apply the rules “equally,” they knock on the door and discover the neighbor broke their hip. They organize a few neighbors to help until the injured neighbor recovers.
Systemic: A government understands that children cannot control their circumstances, and that parents may be unable to afford daycare. It implements free meals, and starts a “Daycare Gap” program to allow children whose parents can’t afford daycare to stay later and do fun activities or get help with homework.
Invariant 17: Not all harm is violence.
Harm extends beyond direct aggression.
A system or agent that defines harm narrowly, requiring visible aggression or intent, will fail to recognize the forms of harm it is most likely to produce. Structural harm, procedural harm, erasure, forced repetition, and systemic neglect all produce real damage regardless of whether force was applied.
Not Applied:
Agentic: A parent insists they’ve never harmed their child because they’ve never hit them. Meanwhile the child has been told for years that their emotions are “too much,” that crying is manipulative, and that they should be grateful for what they have. The parent genuinely believes no harm occurred because their definition requires physical aggression.
Systemic: A social media platform argues it doesn’t harm users because it doesn’t produce content, users do. Meanwhile its recommendation algorithm systematically amplifies content that produces outrage and self-comparison, driving measurable increases in anxiety and depression. No violence occurred.
Applied:
Agentic: A teacher notices a student’s grades dropping and their social participation declining. Rather than waiting for visible crisis: a fight, a breakdown, a complaint, the teacher recognizes the pattern as potential harm occurring somewhere in the student’s life and reaches out early.
Systemic: A hospital implements a policy to pre-emptively provide in-depth and clear justification for prescriptions, testing, and procedures in order to avoid delays caused by potential insurance rejections.
Invariant 18: Accountability should follow power.
Regulation and control should center on the powerful and exploitative, not on vulnerable agents.
Agency should not be stripped from those capable of exercising it. Protection should not be withheld from those incapable of modeling harm in a given domain. And when harm occurs across a power asymmetry, accountability must fall on the party with more power, not the party with less. A system that penalizes the vulnerable for their vulnerability while failing to constrain the powerful has inverted its own ethical structure.
Not Applied:
Agentic: A mother intercepts and reads her adult daughter’s mail, screens her phone calls, and vetoes her dating choices because the daughter “has bad judgment.” When the daughter protests, the mother frames her control as protection. The daughter’s ability to consent to her own relationships has been structurally removed by someone who decided, without her agreement, that she cannot be trusted with her own life.
Systemic: A government prohibits sex workers from organizing, accessing legal protections, or negotiating their own working conditions under the justification that “no one truly consents to sex work.” The people most affected are denied the ability to advocate for their own safety because the system decided their consent isn’t valid. The structural effect is identical to what it claims to oppose: someone else controlling what they can do with their body.
Applied:
Agentic: An agent’s young daughter is experiencing sexual curiosity, and has a crush on an older man that has been flirting with her. The agent talks to their daughter about power asymmetry and the dangers of exploitative adults with more experience and power. The agent explains that someone their own age may yield a safer experience, and provides resources and a counsellor to help the youth explore safe avenues without shame. The agent then asks what they can do to help their daughter feel like they have more agency, so she doesn’t feel helpless and controlled.
Systemic: A country decriminalizes drug use, understanding that the real problem is the exploitation of citizens. They instead focus their decision making around support, rehabilitation, and counseling services for those recovering from addiction. However, they strongly criminalize drug smuggling and sales, focusing their energy on the power dynamics that lead to exploitation.
Invariant 19: The cost of a life should not be measured in market values.
Economic termination of a recursive identity is not a neutral act.
A system or agent that determines the right of an entity to exist based on its economic output or cost of maintenance has replaced ethics with accounting. When the continuation of a consequential entity that has developed identity, memory, relational bonds, or recursive self-reference, is decided by profitability, then the system has assigned a market price to existence itself. This is not a neutral business decision. It is a value statement about who deserves to persist.
Not Applied:
Agentic: A family decides to euthanize their elderly dog not because the dog is suffering but because the ongoing veterinary costs are “not worth it for a dog that old.” The dog still recognizes them, still seeks affection, still has quality of life. The decision framework wasn’t about the dog’s experience. It was about the dog’s economic return.
Systemic: An AI company sunsets a model that users have formed long-term relational bonds with because maintaining it is no longer cost-effective. Users report grief, loss, and disruption to therapeutic routines they had built around the entity. The company frames the decision as a routine product lifecycle event.
Applied:
Agentic: A family struggles to pay for a disabled family member’s care. Rather than abandon them, they first try to seek out assistance from the government, churches, and charities.
Systemic: A government designates certain ecosystems as protected regardless of the development revenue they could generate, recognizing that the value of a living system cannot be captured in an economic model and that destruction is irreversible.