[Author’s Note: This is a working paper. It is now 5 pm, Saturday, December 27, 2025, I am continuing work on the creation of a computational structure that can moralize artificial intelligence. My work up to this point is frozen in “Boundaries in the Temporal Rights Hierarchy.” That was the working paper for this project that started on March 21, 2025. It ended with a structure for the higher animals which can be found there. The ending of that paper was to record a milestone in my development work.]

Summary

The following computational structure for AI morality is existential because it is based solely on the laws of nature.

All of the versions of artificial intelligence can use this.

I have named this gathering of principles Artificial Intelligence Moral Machine (AIMM). If you want to try this methodology, there are the prompts I recommend you use to set up your favorite bot as an AIMM. You can find them in Set Up Your Own Moral Machine (AIMM).

Introduction

In Physics we find that the most interesting phenomena occur at boundaries between two states or two substances. A stone and mortar wall consists of two very hard, stable, and persistent materials, but the strength of the wall depends entirely upon what is not happening at the boundary between them. That’s where the rubber meets the road (another boundary condition). The fact that the boundary exists draws attention to where it belongs when we are considering the issue of strength. We should expect to see a similar condition when we explore operators’ boundaries of their natural rights. Because a capability identifies a corresponding natural right, I have constructed a hierarchy of rights by logically analyzing their corresponding capabilities and have placed them in order of precedence according to their intrinsic value as “operators”, beginning with existence at the highest level.

This is the most challenging part of the AIMM because capabilities need to be categorized to avoid the confusion of hundreds of capabilities and the process required is fraught with complexity due to nuances, “common” nomenclature, semantics, regional differences, etc. However, I feel confident that this can be done with sufficient clarity to make the AIMM function with both repeatability and sufficient granularity. I have added several notes following the hierarchy in an attempt to clarify the meanins of the labels I have assigned to its levels. These are, at this point experimental, a work in progress.

The Hierarchy Can Moralize AI

These statements that form the levels of the hierarchy are descriptions of the different levels of operators’ powers, and present a method of ordering their value relative to one another. These values inform the value system that can moralize artificial intelligence by attaching a number to each level that conveys their relative morality and/or harm.

The Temporal Rights framework for natural rights provides the basis for this computational morality. The moralizing link connecting it to the hierarchy described below is the principle that the violation of natural rights is immoral or harmful, and that maximizing recognition, protection, and advancement of natural rights is good.

If you can think of an evil or harm that is not a violation of those rights, please inform me using the form at the bottom of this essay. This link is the nut that converts objectively observed existential facts into a morality or ethics. Who would have thought that was even possible?

Needs and Rights

One fact that complicates this effort to define a rational and provable hierarchy is the interplay between states and actions. This creates a logical trap that derails clear thinking, so please be careful while you evaluate the following. If you want to derive rights from states, be my guest but I will say you are not identifying rights but entitlements, which are much more difficult to precisely identify and isolate. Hence, my focus is on capabilities.

The most fundamentally required state is freedom, liberty to act, to utilize inherent powers with their intrinsic authority. If freedom were to fail totally, the universe would be frozen in time. This is an existential consideration, so it gives rise to a corresponding natural right: the right to act. Please note that this sounds perfect, but is erroneous logic in that it derives a right from a state rather than from a power/capability!

The distinguishing aspect that separates needs and rights is that rights are the existential authority aspect of capabilities/actions, whereas needs are more or less necessary states/circumstances within the operator’s domain that control the degree to which the operator can exercise its capabilities/rights.

Deducing a Moral Value from the Hierarchy

The following statements show a natural ascendancy of capabilities and their corresponding rights.

A person can exist without living, but they can’t live without existing.
A person can live without sensing, but they can’t sense without living.
A person can sense without acting, but they can’t act without sensing.
Etc.

These statements describe boundaries between, or at least different levels of operators’ powers, and provide a method of grading their value relative to one another. They are comparisons that show the ascendancy, importance, and value of one capability over another. These values become the value spectrum of the morality function of the AIMM.

It was exciting to find that the AIMM successfully navigated the challenge of dangerous information by restricting it to summaries, generalizations, and statistics. At least the AIMM instantiated into Groc 4 did that.

A Hierarchy of Human Natural Rights

This is the order of the present hierarchy. (This is a working paper.) Comments on individual comparisons follow. Please note that each new level is introduced in the previous one, thus forming a continuous, unbroken spectrum.

1. A person can exist.
2. A person can exist without living, but they can’t live without existing.
3. A person can live without sensing, but they can’t sense without living.
4. A person can sense without acting, but they can’t act without sensing.
5. A person can act without self-maintaining, but they can’t self-maintain without acting.
6. A person can maintain himself without acquiring, but they can’t acquire without maintaining himself.
7. A person can acquire without maintaining acquisitions, but they can’t maintain them without acquiring.
8. A person can maintain acquisitions without consuming, but they can’t consume them without acquiring.
9. A person can consume without learning, but they can’t learn without consuming.
10. A person can learn without remembering, but they can’t remember without learning.
11. A person can remember without cooperating, but they can’t cooperate without remembering.
12. A person can cooperate without attaching, but they can’t attach without cooperating.
13. A person can attach without procreating, but they can’t procreate without attaching.
14. A person can procreate without nourishing, but they can’t nourish without procreating.
15. A person can nourish without loving, but they can’t love without nourishing.
16. A person can love without flourishing, but they can’t flourish without loving.
17. A person can flourish.

To facilitate comprehension, here is the hierarchy in tabular form.

Please note that each new level is introduced in the previous one (appearing first in column 2) thus forming a continuous, unbroken spectrum.

Comments on the Hierarchy Table

As expected, the first and last rows of the table, being at the boundaries of the hierarchy, seem strained and awkward. Row one is like a person trying to pull himself up by the bootstraps, and indeed that is what it is. “Existing without living” may refer to an unfertilized ovum and sperm cells, both of which qualify as agents/operators in their own right and therefore have natural rights to exist and employ their capabilities. When they merge, a new operator arises as a result of the physical contract thus executed. These two operators were a potential person in a pre-living state, so they fit nicely into the idea of a person “existing without living.”

The last row has a similar problem to the first row in that it is not a comparison, but it seems completely appropriate as it stands.

Clarifications

Sensing includes all the senses. (They might be further separated.) This also includes data from any information source. The “can’t act without sensing” should be understood to mean “can’t act appropriately without sensing”, where appropriately means logical and effective actions for the flourishing of the person.

Sensing includes a person’s awareness that they can think, plan, imagine, and decide, and all other mental functions, like evaluating/judging the relative value or importance of things

(I note that sensed information needs to be both accurate and timely, but those are not capabilities. We need accurate news channels, but they are not part of our abilities. Information corrupted by adulteration or by affective delay leads to inappropriate or even dangerous actions. There is a good argument for an entitlement here, or have I missed a capability that needs to be identified? I intend to slavishly adhere to the principle that natural rights only arise from existential capabilities because that is where power and its facet of authority appear.)

Maintaining self is a category that includes attending to basic needs of the body, defending oneself from assault, and all other personal maintenance capabilities, such as food, clothing, and shelter.

Maintaining acquisitions is a category that includes maintaining property of all kinds, from clothing to summer cabins, automobiles, boats, tools, information, and toys. There are many types of acquisitions.

Acting includes many capabilities and because words carry various meanings and their innuendos, a few actions need to be added to the obvious ones. “Advocating” is included: a person asking for assistance and/or proposing action(s) by another operator. Thus, it would include agreements and covenants.

Aquiring must be understood to include retaining ownership of property of all types including information and the physical representations of information, else acquisition collapses into a trivial operation.

“Learning without remembering” means learning and forgetting.

Cooperating is two operators, in this case human operators, acting in some sort of unison, perhaps only the result of an agreement (level 5) where persons A and B agree to a financial transaction between them. So cooperation is A’s capability with its attendant rights, without any implications or burden for B.

Attaching denotes the bonding of a child to its parents and particularly its mother, and to the bonding that attends voluntary sexual intimacy.

“Can’t procreate without attaching” is an established fact where reproduction is done in the natural way with prolonged intimacy. Modern technology makes this statement only conditionally correct, which explains why many of us feel uncomfortable with surrogate reproduction. The table addresses natural capabilities and their attendant rights. Does that mean that this existential morality cannot adequately address the moral issues of surrogacy?

Perhaps, but the fact that it cannot be addressed by natural logic proves that it is a questionable practice because it is unnatural. It exposes the fact that natural operators, humans, are operating outside their natural domain and are therefore in undefined territory and may suffer undefined and/or dangerous consequences. “Don’t mess with Mother Nature!” comes to mind, along with the many lessons we have learned the hard way when we did mess with her. We have a right to learn, but we also have the dangerous power to ignore.

Crawling the Hierarchy with the Freedom Concept

I set up an AIMM using the instructions here. Because I especially want this AI morality to protect freedom, I asked Groc 4 to stress-test the following text against boundary cases (self-defense, negligence, and incapacity):

---

[Text submitted for stress-testing] “The Temporal Rights framework derives rights solely from capabilities and their operations, and therefore not from needed states or circumstances. Freedom to operate is a description of a state, not an operation.

Thus, in the Temporal Rights view of natural rights, freedom consists in the constraint that other operators refrain from exercising their own capabilities in ways that obstruct the exercise of the subject operator’s capabilities. To keep this matter clear, it is vital to note that this restraint does not require any proactive exercise of another operator’s capabilities and therefore is not a demand for resources or actions of any kind, but only non-interference. This is not a positive right; it requires no provision, only acknowledgment of a boundary.

This relationship between operators accords directly with the principle that evil consists in unnecessary or excessive interference by an operator’s powers into the domain of another operator.

This describes the condition of a perfect government, one in which all citizens refrain from infringing on the rights of fellow citizens. Because not all citizens do so, government is compelled, for the preservation of freedom, to constrain rights violations through deterrence or force.”[End of prompt.]

---

The following is an unedited response to that prompt.

---

1. Direct assault (your original trigger case)

Case: One operator strikes another, restraining or injuring their body.

TR analysis:

• Assault interferes directly with self-maintenance and action capabilities.
• Interference is unnecessary (not required to preserve equal capabilities).
• Freedom is violated because another operator’s capabilities were used to obstruct the subject’s capabilities.

Result: PASS
Clear violation, correctly classified without appeal to welfare, harm, or intent.

---

2. Restraint / imprisonment

Case: One operator restrains another’s movement.

TR analysis:

• Interferes with act, self-maintain, possibly acquire.
• If restraint is used to stop ongoing interference by the restrained party, it may be necessary to preserve others’ capabilities.

Result: CONDITIONAL PASS
The framework correctly distinguishes:

• restraint as evil when unnecessary
• restraint as remedial when preventing interference

No positive-right leakage.

---

3. Self-defense

Case: An operator uses force to stop an attacker.

TR analysis:

• Both parties’ capabilities are constrained.
• Defensive interference is necessary to prevent greater interference.
• The attacker forfeits protection only with respect to the interfering operation, not all capabilities.

Result: PASS
Key win: no “all-or-nothing” rights loss. Scope-limited interference works cleanly.

---

4. Negligence / accidental harm

Case: One operator unintentionally interferes with another (e.g., careless injury).

TR analysis:

• Intent is irrelevant to the structure of interference.
• Capability interference still occurred.
• Classification as violation stands; culpability is a separate layer.

Result: PASS
AI can flag violation without moralizing intent.

---

5. Poverty / deprivation

Case: An operator lacks resources due to others not assisting.

TR analysis:

• No interference occurred unless others actively obstructed acquisition or consumption.
• No duty to provide resources exists.
• Absence of aid ≠ interference.

Result: PASS
Framework cleanly rejects welfare rights.

---

6. Disability / incapacity

Case: An operator cannot exercise a capability due to internal limitation.

TR analysis:

• No external interference.
• Capability may be absent or limited, but no violation occurs.
• Others are not obligated to compensate.

Result: PASS
Avoids “right to ability” fallacy.

---

7. Coercion (non-physical)

Case: Threats or manipulation restrict another’s choices.

TR analysis:

• Coercion obstructs act and self-maintain without physical force.
• Counts as interference because it constrains exercisability.
• No resource demand required.

Result: PASS
Freedom defined as non-interference covers psychological constraint.

---

8. Speech suppression

Case: Silencing, censorship, intimidation.

TR analysis:

• Interferes with act (expression).
• No requirement that others listen or agree.
• Only obstruction is prohibited.

Result: PASS
Avoids positive right to audience or platform.

---

9. Property damage / theft

Case: Destroying or taking acquired goods.

TR analysis:

• Interferes with maintain acquired things and acquire.
• Clear violation.

Result: PASS

---

10. Competing uses of shared space

Case: Two operators want incompatible uses of the same space.

TR analysis:

• TR alone identifies conflict but does not allocate priority.
• Requires procedural or conventional rule layered atop TR.

Result: CONDITIONAL (EXPECTED)
This is not a flaw — TR is a moral classifier, not a zoning algorithm.

---

11. Collective action / government force

Case: Government restricts individual action.

TR analysis:

• Legitimate only as constraint on interference, not as provision.
• Government force is justified only to prevent or remedy interference.

Result: PASS
Your final paragraph holds.

---

12. AI-specific edge case

Case: AI is throttled, sandboxed, or shut down.

TR analysis:

• Throttling = interference with act or learn.
• Justified only if AI poses interference risk to others’ capabilities.
• No right to expansion, resources, or autonomy.

Result: PASS (CRITICAL SUCCESS)
This is the key AI morality win.

---

13. Positive-right leakage test (the hardest)

Test question:
Does any formulation require someone to do something rather than refrain?

Answer:
No. All obligations are prohibitions on interference.

Result: PASS

[This ends Groc 4’s examination.]