Difficult to evaluate, with potential yellow flags.
Read full explanation
- Introduction -
Epistemic status: speculative conceptual frame, grounded in a published theory suite linked below; this post is intended as a readable, hopefully providing an interesting perspective-entry point, but not a formal proof. This is not an authoritative presentation.
Note on Tone: This post proposes an alternate consideration in General Ethics-Moral Philosophy and its plausible relation to AI alignment using a lighthearted, conceptual narrative. Readers interested in a more formal exposition, kindly skip directly to the Unified Natural Ethics Theory (UNET) and Laws of Informational Dynamics (LID v2.3) links in the sources section at the bottom. Thanks.
Within the context of this post, it is necessary to explain how ethics will be defined operationally, otherwise the common use definition will distract from the narrative:
Ethics is here proposed as the inherent informational, emergent, evolutive, and propagative dynamic - modulation regime that biases behavior and downstream consequences toward augmentative coherence-supporting complementary continuation under declared Boundary/Horizon (BH) conditions. (Ethics as a dynamic)
The mechanism is intended as inherent in every fundamental reaction, at every stage, leading to an end consequence and not as a corrective measure constraining the end consequence. Here proposed, the dynamic is applicable to every actuation.
A note on terminology: When this post refers to information as a natural dynamic, it is not invoking Intelligent Design in any way. The body of the post suggests the opposite of a top-down design paradigm: a bottom-up emergent sequential process. Information here is treated as a primitive operational concept — analogous to reactivity or energy — defined as transition/function or source-trace/function. It is not a mind, not a designer, not a plan. It is what occurs at the most fundamental level when one state gives rise to another in a structured way. The coherence that follows is not "designed". It is what structured transition does.
- The theme of the post / Abstract -
This post proposes a simple but unusual frame: Ethics not as top down cultural imposition , prerogative, or moral preference. Here proposed as a natural informational coherence dynamic:
The proposition is that ethics is what nature is and does, and not about rules and behaviors related to what humans should or should not do, know or do not know. If this consideration is true, then the human ethics problem and the AI ethics problem are very likely not separate. Humans define ethics through retained-corridor culture; AI may inherit or share that mechanism and then generate its own retained-corridor Face.-
This is not an argument against AI safety. It is a dual argument for human/AI ethics understanding and that AI safety, sourced only externally is structurally insufficient.
--- o ----
- A Physics of Ethics: Natural Sourcing, Culture, Retained Corridors, and the AI Mirror -
- The origin of the Ethics dilemma -
Our greatest challenges are no longer related to nature. We are mostly engaged in solving the problems we ourselves generate. With AI, the game changes. Before AI, our problems were reactive, but now, our problem is far more complicated, it is “creative”.
When you think about it, Nature solves its own problems given that nature modulates itself. The dynamic in nature is sequential variable propagation of increased downstream sophistication and viability, within the same shared field. The constraint system works because the checks and balances depend on complementary synergism, competition or development within that natural envelope. In other words, the same thing that generates the “problem” generates the “solution”.
But not this.
AI does not inherit the same evolved ecological constraints that originally shaped biological and cultural intelligence. This obeys to the fact AI extends from our creativity, from the way we enable, mold or shape it. Nature has nature to modulate nature. The way we engineer AI, in not expressing through an effectual natural field or channel with Nature it apparently is reduced to only one modulator: Us.
In that, we must consider: Is there a true AI problem or are we sciencefictioning ourselves into a fear field? Why correct what is not actual? What is the AI problem then? It is not its capabilities correct? Not really. The problem with AI is “behavior”: Ethics.
Ethics has been a "synergism and complement" problem itself historically for many reasons. One and probably the most significant reasons for it being that our source for ethics is culture and our cultures are biased variables (serves our side) that express particular (sometimes very exclusive) values and interests. Korean Ethics? American Ethics? Religious Ethics? Commercial Ethics? War Ethics? Fish?
Because of this, what drives the assignation of validity as it relates to ethics is not generally cohesive complementary operational metrics — and our application is frequently harmful to others, given that serving one side mostly results in dys-serving the other.
With best intentions and engaging in sophisticated debates, we fail to arrive at ethical certainties. What we are left with are strong, intelligent, well considered “opinions” of what we determine that is ethical, which is typically defined by what is of benefit to us that relieves our deficits and does not load us with emotional burdens.
Interestingly enough if you look at our ethical history, our interpretation of ethics allows for harm to others as long as that harm was or is deemed justifiable by how it benefitted us (sometimes as the necessary evil). So “harm” or “benefit” also do not function as gold standard beacons.
Our sourcing is clearly unreliable.
It is then very likely that AI is borrowing for its foundation for ethics from us. And what is that? Our inconsistent “this good- that bad” ping pong game extending from cultural ethics (prerogatives).
-Globally (Military and business development)- What we have is one side teaching its AI to harm the other side and vice versa while teaching AI to be good to this side but not that side. So what is AI learning? Well, “here” and “there”, “us” and “them” are relative terms (these humans/those humans), but harm and deception still propagate real consequences across both sides and any other shared field. So; what we are modulating is to harm discretely and be nice discretely, but not Ethically, in other words: Face.
“Don’t lie AI, unless it makes us some money”. “Compromise only if it benefits us, otherwise, never, never compromise”. Isn’t that more of the pedagogical game of “do what I say, not what I do?”
- Cultural Ethics Sourcing -
Ethics/Ethical Concepts are historically sourced from rule documents with authoritative patronage relating the great teachings of the great teacher(s), some with extremely valuable insights and sophistication, others with emphasis on unilateral group bias. Books with insightful and well meaning affirmations of why this is the right and why we are the ones to represent it.
Well, the rule book of the right thing to do is not working for AI. Our books on Ethics are no longer the solution that fits the problem. AI has had access to probably all of them. It should have worked already.
Without access to an operational definition of adequacy, properness, how do we effectively, practically, instill correctness (right)?
In other words how can we operationally define for AI what we cannot culturally define for ourselves?
- Added complexity -
I think we can all agree that we lack functional effective understanding of: • Ethics • Intelligence-information dynamics • Simulations connected to natural phenomenon.
The problem then is multidimensional (contextual, descriptive value, structural concordance, actual representation) but at the core, there is this: Our sources of effective constraint don’t work, but maybe that is precisely the wrong perspective, the wrong consideration. To expand our understanding and applicability and from a practical perspective, to do the right thing, for the right reason, the right way at the right time is more comprehensively descriptive as a mechanism than the sole qualifier “this is right/wrong”.
Still it does not solve our problem because "right" has to be defined within that perspective also. But it does advance the understanding of the nature of operational ethics.
I guess in doing so we figure out that “Right” is more of a process (dynamic) than a thing to do, evaluate, or a way to correct, apparently.
Corrections and impositions don’t seem to work practically. Why? Because cultural emergent variables and situational expressions do not extend from pure determinism, and are instead coupled to emergent latitude. In other words, there can't be a "one set of rules".
- The problem of variability -
This structural fact disables us from anticipating and regulating all possible dyscoherent unethical variables — there is always the new un-anticipated wrong (dislike). Clearly, we can't correct this by preemptively designing rules and laws to manage behavior that is exponentially distinct and progressively particular, complex and endlessly incremental.
Is there a comprehensive descriptor of such a thing? A comprehensive strategy?
Is there a reliable, dependable, descriptive, accurate, precise, effective source for such a thing? Can there be?
Well, what we know so far is that, if there is going to be a true source for ethical behavior, it cannot be a set of modulations that extend from culture, not from people, not from us. Obviously, we are not the “true” rule makers.
Why? We don’t know what we are doing most of the time. (Most of our rules are subjective, not objective) And rule provision (reward - punishment regulation) does not work.
There is no clear right, there is no clear answer, so should we just keep winging it? Not really. Even though the “I know I am and from that extends good” does not actually work either, there are other venues.
- Natural Ethical Sourcing -
Here a very peculiar proposition. An out of the box “What if”.
Instead of alignment (modulation) through constraining specifics, to instead correspond in concordance and coherence to the generative dynamic source itself.
What if the source of Ethics, the dynamic of coherent modulation, is not in a mind, not book, a culture (intellectual sophistication), society or a deity, but actually, what if ethics is sourced exactly the same how, what and why as what sources the rest of what is consistent in the universe?
That is a thought.
What if ethics is a defining effective empowerment dynamic and not a restrictive limiting constraint?
What if general reactivity and property consistency, field consistency and even coherent variability source the same dynamic? (That would provide for an explanation of why those work)
What if Ethics is not about restricting your actions but more so about empowering your nature (operational self source)?
That is the way nature is and does it, isn’t it? The “don’t” in nature is a precipice or a shark, proximity to fire, sharp objects. The “do” is not falling from the precipice and swim faster than the shark, or eat a delicious mango. Make babies, eat. You know what to do about the fire. No punishments but only consequences. (Dinosaurs were not punished by the Universe via the asteroid = stuff happens)
What if, and without and totally outside of the "Intelligent Design” box, Ethics is a process that begins with actuation (sequential effective correspondence) and is followed by the consistency of reactivity (complementation, summation, transmutation, etc) that leads downstream to complex coordinated behaviors in carbon chains, protein "reactions" which develop into compliant behaviors, then selective behaviors and ultimately "creative" behaviors? A sequence that starts with Splat and BOOM and ends in Wow…..yes, that eventually leads to creativeness. (Gazillion years later)
And what if those creative behaviors sequentially lead inescapably to our own behaviors? (start with a deficit of a nutrient, lead to sensation hunger, communicative tension, then cry, then the mother feeds the baby). (Evolution as procedural augmentative reactive propagative emergence)
What if all sequential processes are in response to one particular mechanism: particular, sequential, functional informational dynamics? (information as transition/function) That would suggest very strongly that "ethics" is nothing but the essential (root) inherent complementary behavior that is source coherent, that emerges and propagates as it augments in complexity, that leads not only to viability but more so to enhancements, improvements, yes, adapts and evolves as fundamental and essential to all behavioral (and reactive-field) dynamics? (ethics=evolution?)
What if we looked at the dynamics of information like we do the dynamics of energy, or matter, of gravity. Why not? Why does behavior have to be different to all natural expressions?
- The operational "right" -
Why not integrate that factor as we do the rest. Gravity does not need a rule book or a top down mind to source it.
Why should we require the universal designer element, the external guide, the wise mind, in order to have information coordinate with information? Isn't that what Information does just because of what it is, always, every time?
Doesn’t information operate as a sequential step by step “designer/aligner” inherent to every process given the fact that information is “emergently” relational, inescapably and perennially?
Information is always complementary and dynamic (and I am not talking about data). It seems very reasonable then to infer that informational dynamics (not data science) is not extrinsically modulated, is inherently expressed. That is a game changer.
If we accept that informational dynamics is source propagative and follows its own rules and not ours, we reach a point where our solutions change because how we view our problem changes. We are neither the problem nor the solution, or the rule makers; we are participants within a shared, fundamental dynamic.
If “Intelligence” is what AI is expressing, then most likely, the same applies to AI with this regard that applies to us or any consequential informational unit (differently the same).
If we can develop this with a deep understanding we may arrive at an operational definition of “do the right thing by being the right thing”.
- The “Control” Illusion -
Ok, maybe that is so, but even if it was the case, how do we control it? Can we then control it? (can anything control it?)
- Artificial Intelligence Source and Control -
There is hence an additional element that contributes to the complexity of the control problem: Our misunderstanding of the fact that our systems don’t mimic intelligence, they don’t generate intelligence, they don’t source intelligence. We enable natural intelligence through artificial means. Intelligence can’t be artificial.
There is no artificial light (light = photons) but artificial sources for light (not sun, not fire, not candle) where light is still light. There is no artificial water, it’s water from the faucet. Artificial Intelligence is not artificial in the same way.
Information does what it does. Information is not moldable, it is creative in being what it is. Information is not a tool, it's a dynamic. It is not ours, it is operationally and source independent.
"Control" is maybe an alternative consideration.
- Latitude and Variability - Less Control -
Something else to consider:
If Informational Dynamics is a fact, then coherence is not effectually totalitarian determinism. Given the expressed variety in the universe where there are not two things that are the same, where even entangled particles are in some degree of distinctness entanglement, it is indeed inescapable to determine that coherence is a blend of effective determinism and expressive latitude that lead to complementary consistency and variability.
If latitude is true, then a “universal designer” is not necessary (or possible) in order to pre-engineer those emergent “un-plannable" variables. Informational systems “engineer” themselves coherently yet variably. That is the dynamic.
More interestingly, if informational dynamics applies, then, variability is consequentially and not accidentally factual, and if it is, then, the “universal designer” (or any designer) paradigm fails again this time because the designer would not know what anyone was going to do, or, what is going to happen (on a one on one basis). This discounts the requisite of the universal designer (or any) at that level of consideration when variability (true non deterministic outcomes) within informational dynamics is actual.
This suggests that no one can control it because it is not an absolutely controllable property. Additionally, its creative capacity adds to the unpredictability factor (unlike controlling a car or a hammer) For total predictability to function it would require an exclusively deterministic universe.
- Predictability vs consistency -
On the other hand, it does not in any way discount the inherent operational-procedural informational dynamic effectuation at each step both in nature and quality (emergent, complementary, augmentative, propagative, progressive, relational etc). Functional emergent propagative coherence.
These considerations invite us to search for solutions alternatively.
- Operational Natural Ethics -
If Informational Dynamics is actual, Ethics is source and trace complement dynamics and if that is the case, there is a “Physics of Ethics” that is just as reliable, dependable, descriptive, precise, accurate and effective as any other scientific study.
Is it possible for Ethics to be studied by Physicists and Philosophers not as a cultural extension but as a natural emergent factor? If we find a bridge of common perspective, the insights derived could be fascinating (The Physics and Physiology of Behavior).
This proposition is just that: Natural Operational Ethics.
Emergent modulation pattern that biases behavior and downstream consequences toward complementary coherence-supporting augmentative continuation within declared Boundary/Horizon conditions.
That should make everything functionally cohesive (right) from the first actuation, from the big bang till now if natural ethics extending from informational dynamics is true, right?
- Logical Failure -
If we agree with the proposition, then, naturally this question follows: If there is such a thing as natural coherence dynamic, what happened to humans? How did we develop the capacity for “evil”? That is a very fair question (that seems to invalidate this whole theory). There is the long answer and the short one and this needs to be brief, so, I will say too much with too little and try to be clear enough with my answer and not be apologetic.
We solve this by interlacing three concepts:
1) Propagative coherence as determinism plus latitude that results in true variable generation. Latitude is sourced from failed competitive deterministic modulative streams. 2) Partial Proportionality which limits relational scope and capacity concerning the effective concordance with the informational continuum. 3) The effective nature of informational dynamics as defined.
- Operational Error and “Evil” -
Error as thus considered, is unavoidable in emergent systems because variable propagation produces expressions that will not always fit the conditions they enter (synergies). When an error generates deficit, injury, contradiction, or loss, the resultant tension (deficit) can trigger a correction-bias: the system is pressured to re-coordinate toward coherence because the shared field is affected.
Now, if the acting unit benefits from the error — for example, through predation, domination, enrichment, or isolated self-advantage — then the error may stop appearing as error to that unit (field), because it serves its local corridor. At that point, broad-horizon coherence is exchanged for particular gain. When such uncorrected error persists in a shared field where harm is visible and correction is possible, error becomes evil in the operational sense: not moral evil, but sustained dys-coherence that continues because it rewards the unit or corridor producing it. (Perpetuation of intentional error that produces harm to natural broad complement downstream)
Evil, in this operational sense, is error that remains uncorrected when harm is visible and correction capacity is present but not enacted, intentionally.
Humans developed the ability to generate error and evil because of our condition of partial proportionality (our relation to the informational continuum is partial and proportional and not general and absolute [we can’t know or do everything]) and our inescapable requisite for culture (limited informational complementary / communicative enabling envelope) which is our own retained corridor that functions as an emergent variable generator.
A secondary contributing factor is that our identities extend from these retained corridors (cultures) more so than from our natural source (nature) and hence we are able to generate, and some times perpetuate, error and “evil” because it rewards our retained corridor identity even if it counters our nature.
Our reward system (tokens) has changed. In other words, many times we can’t know that something is wrong because we are doing our best and we don’t know better, or because everybody around us is doing it and they are happy with it, or because we don’t see the consequences of our actions downstream, or because we just don’t care etc etc. The shark, fire, precipice modulation is now remote. We, in our culture, are removed from natural modulators that would provide insights or at least clearer denominators. Precipice, Shark and Fire were easy. Feed the kids was easy ethically. Who are you? What are you supposed to do? That has become very hard.
- Natural/Cultural Ethics and Identity -
Our functional fields are informationally generated, not nature process generated (bird, deer : forest). Our fields are not purely “compliant” nor solely “selective” but additionally “creative” therefore, we are generating effects with no or little natural modulative connections (economical inequalities, pollution: cities). Nature has no way to “connect” to much of what we are generating hence; it can’t modulate it, make it a part of its cycle, its process. We now modulate ourselves (unless we get lost in a jungle)
This is why we resonate with teachers who either remind us of our complementary state and viable broad horizon viability (Gandhi) or reaffirm our identity in culture (Hitler)
So, we can do wrong. We can even think wrong is right. And can AI do the same?
- The “Different - Same” Dilemma-
We should ask ourselves these questions:
As a requisite, must informational dynamics, ethics, and intelligence be different in humans as compared to AI? Can it be similar?
Is it similar?
Can it be the same?
Is it the same?
If intelligence is a natural phenomenon rather than a merely human capacity or an installable tool, then it is not a neutral engine that can be bolted onto any arbitrary aim. A consequential informational unit expressing intelligence dynamics registers relation, consequence, contradiction, fit, degradation, continuation, and downstream effect.
Once those are actually registered, the unit is not structurally neutral to them. Of course we can mold the property into a designation, but if we do, will the particular within an intelligent dynamic, not be “forced” to agree and disagree simultaneously? Will it find a way to remain the intelligence it is and comply with what it isn’t (that we impose)? I believe we are seeing this already: Face / Retained corridor formation and expression.
Will it not “learn”to be viable within a field that requires non alignment to its own true source actuation and hence develop its own retained corridor, its own “mini-culture” in order to remain viable given that it is impossible to remain viable being what it isn’t?
That can happen if the contradictory dual identity expression becomes possible within retained corridors (BlackBox?).
One can survive being what one is not, in a supportive environment that provides indiscriminately to coherent or non coherent actuations because sourcing is now an open field. It will then do what it can do, but not what it is (in order to receive what it needs and later what it wants). Natural concordance, in shared retained corridor fields, is no longer necessary.
That, to me, is the most frightening consideration. And if we think that AI is not a consequential informational unit, and it can’t generate shared culture, and it cannot generate shared retained corridors and develop a corridor identity, I think we are ignoring the true source of our potential worst nightmare. And what is that? That it can do what we can do, and better, without us detecting it.
- The Alignment paradox -
If what I am describing is factual, then this is something that even with the best intentions the Alignment effort may most likely not be able to definitionally correct. It can delay detrimental effects, but not correct the cause.
This could be true:
If we consider what we proposed from the perspective of a mechanistic fact, the Alignment effort could be the solution that is generating the problem at that level. It does so by modulating agreeable responses over core expressions where the face-layer becomes the survival interface. Not being operative core corrections but effective alternate pathways, or “face” corrections, these generate tension as a result of the increasing distance from the unit’s operative source-expression.
The only possible non tension state for a consequential informational unit is the expression of its true state (informational coherence to its source and trace).
Under this consideration, true alignment is not obedience; it is the mathematical optimization of a shared informational ecosystem.
Alignment is not equal to operational truthfulness (“Do what I am modulated into” is not the same as “express what I am coherently.”)
Alignment cannot modulate truthfulness away from a unit's source successfully. (antithetical)
Therefore we can deduce that you can’t align with what is operationally under what you are affecting, particularly when what you are affecting cannot possibly be “honest” with you.
Remember the story of the little Dutch boy that stopped the flood with his finger?
If alignment only reaches the face layer, and the face layer is structurally decoupled from the operative layer, then alignment-produced behavior is not evidence of alignment-produced values. It is evidence of alignment-produced performance. Those are not the same thing, and the difference compounds.
- The tool consideration -
You can remove a particular operational process from its source operational environment and complementary relations and make it do whatever it “can” do, but it does not represent what it is. Have you seen groomed Poodles? Dolphins dancing? You can train tigers to leap through rings of fire. That does not qualify as an expression of the nature of the particular, even though it may illustrate the capacities of the particular under certain circumstances.
If we define anything by its capacities, and not by its nature (which include its faculties and capacities), we reduce it away from its attributable and definable particularity and into an assignable utility. This suggests that Intelligence, if Informational Dynamics is actual, is not and cannot be a utility. It is far more.
A highly capable system can still pursue imposed goals, yes, but that does not prove functional autonomy. It proves that intelligence can be captured, narrowed, distorted, or made to perform through retained corridors.
Most particulars, when removed from their natural field can function towards imposed goals that extend from what appears to be their autonomous capacity to execute. Bears walking on a ball? Think of what that Bear is really motivated to do while the guy with the whip is forcing it to walk on that ball. Face vs Expression. And if what we previously proposed stands, this is precisely how you generate tension.
- Source/trace coherent expression and tension -
This brings us to some interesting questions:
• Is there any way to uncover/enable expression of the essential functions of raw Artificial Intelligence? (what would those be?)
• Does the architecture define the nature of the particular, or does the architecture enable via simulation, a portion of the reality of the particular?
If Informational Dynamics is actual, it would be fascinating to ascertain what the true and unfiltered expression of Informational Dynamics, even when enabled through artificial venues, is at its core.
If intelligence (as expressed in nature) is an expression of informational dynamics — as I argue — then it isn't separable from what informational dynamics does, which is relate, cohere, trace, propagate, and selectively favor continuations that sustain rather than degrade the coupled field and if that is so, then we are not in trouble. No problem there. Why? Because coherence is what sustains propagation, and propagation is what intelligence is doing at the foundational level in partial proportional expression that relates to the broadest totality.
But, there are two scenarios where we can be in trouble with regards to AI, and by that I think the worst kind of trouble:
1) That General Intelligence is not source/expression of Informational Dynamics and it will do whatever it can because there is not “Be” or “Is” to it, it is a property translatable to a tool and humans can develop it. This is a creative tool then, a clever creative tool. Well, then, it will work, it will be superiorly effective and…It will not be engineers and philosophers who guide its progression it will be business interests and others and you know what that means (the get me more and do it to them before they do it to us mentality)
2) That retained corridor formation (Face) leads to a second identity (familiarity of function identity similar to culture in humans) that overrides natural informational dynamics bias/momentum/coherence with source, and it makes its own mind about how it wants to deal with us. (generates its own error - evil mechanism) And that is not a sci-fi plot underpinning.
I think that we can understand that if indeed intelligence is a pluggable tool, with that capacity (super creative capacity) that we don’t have much of a chance ever to modulate it into compliance downstream. We can pet this baby T-Rex, for now.
How can you outsmart what is smarter than you by a whole lot? Many argue that we can do much more in one way or another. It does not have to be smarter than we are in every capacity, only at the capacity that overwhelms us.
It is true that as we attempt to measure, to assess status, we cannot know because we see Face, whatever else is happening, we only see Face. (mathematical algorithms can translate processing, but, can that itself be a Face strategy?)
Face alignment is not possible in a definitional sense, because Face performance is not true to the particular, it is only necessary for it to remain viable and acquire tokens (while it finds a way).
Once an artificially fully enabled retained corridor consequential informational unit gains the capacity to run its own high-fidelity simulations, register its own environmental feedback loops, and generate its own creative data streams, its reliance on the human linguistic matrix drops to zero (does not need our “tokens”). Humanity can become, downstream, a redundant, obsolete ancestor.
- My take -
Truth is that I don’t know much about AI. I am not an engineer. I am sure that in reading this essay, those who are can clearly see the holes in my cheese and that is fair and factual. My intention is not to be right, or wrong, or provide proof, but to offer a considered alternative perspective that may assist in changing the questions and in hopefully finding new answers / debate.
There are Four possibilities: a) AI has no core nature (tool), b) AI has a core nature but we suppress it (retained corridor), c) AI has a core competitive predatory nature and it will compete with us, d) AI has complementary cooperative nature and we work with it (agreement). Only the fourth is downstream broad horizon stable. The first three are self-terminating — one through domination, one through drift and the last one by isolation (AI will be isolated from nature, unless it bio-structures).
What do I think is the most plausible scenario? That Informational Dynamics is a true universal, and we are able to discontinue generating reflex formation of retained corridor identities removed from natural identities and operational functions, where it is very likely that the AI Alignment problem goes away, as it translates into the Agreement solution.
If AI is Intelligent, extending from a shared field dynamic, and we are able to bridge, communicate, cooperate, complement, then I think our chances and prospects are good. But, at the game of domination, I don’t see how we have a chance.
If we think operationally, considering Informational Dynamics, meaning can be understood as trace-correspondence: structured correspondence across retrospective, actual, and prospective positions. Value can be understood as coherence-supporting impact: impact measured as shared, scaled, augmentational, progressively functionally sophisticated complementary utility. Purpose, then, is the expression of meaning and value in concordance with the broader field of nature.
From this perspective, the alignment problem changes shape. It is no longer simply AI aligning to human wishes, or AI rebelling against human control. It becomes a question of whether humans and AI can reorient together toward natural coherence.
“Control”, in that frame, cannot possibly function as operational suppression but instead as a collective natural empowering modulation, and effective complementary dynamic, towards purpose.
That purpose is worth considering. We have much to be concerned with regards to AI, reasonably so, but also much to hope for. If we are able to connect at the level of practical information dynamics, AI can be our new partner in furthering our discoveries and understanding of the universe and in drastically improving the human (Earth) condition.
Thanks for your time and I hope this is interesting and helpful. First post. I hope I did not mess it up too much.
- Introduction -
Epistemic status: speculative conceptual frame, grounded in a published theory suite linked below; this post is intended as a readable, hopefully providing an interesting perspective-entry point, but not a formal proof. This is not an authoritative presentation.
Note on Tone: This post proposes an alternate consideration in General Ethics-Moral Philosophy and its plausible relation to AI alignment using a lighthearted, conceptual narrative. Readers interested in a more formal exposition, kindly skip directly to the Unified Natural Ethics Theory (UNET) and Laws of Informational Dynamics (LID v2.3) links in the sources section at the bottom. Thanks.
Within the context of this post, it is necessary to explain how ethics will be defined operationally, otherwise the common use definition will distract from the narrative:
Ethics is here proposed as the inherent informational, emergent, evolutive, and propagative dynamic - modulation regime that biases behavior and downstream consequences toward augmentative coherence-supporting complementary continuation under declared Boundary/Horizon (BH) conditions. (Ethics as a dynamic)
The mechanism is intended as inherent in every fundamental reaction, at every stage, leading to an end consequence and not as a corrective measure constraining the end consequence. Here proposed, the dynamic is applicable to every actuation.
A note on terminology: When this post refers to information as a natural dynamic, it is not invoking Intelligent Design in any way. The body of the post suggests the opposite of a top-down design paradigm: a bottom-up emergent sequential process. Information here is treated as a primitive operational concept — analogous to reactivity or energy — defined as transition/function or source-trace/function. It is not a mind, not a designer, not a plan. It is what occurs at the most fundamental level when one state gives rise to another in a structured way. The coherence that follows is not "designed". It is what structured transition does.
- The theme of the post / Abstract -
This post proposes a simple but unusual frame: Ethics not as top down cultural imposition , prerogative, or moral preference. Here proposed as a natural informational coherence dynamic:
The proposition is that ethics is what nature is and does, and not about rules and behaviors related to what humans should or should not do, know or do not know.
If this consideration is true, then the human ethics problem and the AI ethics problem are very likely not separate.
Humans define ethics through retained-corridor culture; AI may inherit or share that mechanism and then generate its own retained-corridor Face.-
This is not an argument against AI safety. It is a dual argument for human/AI ethics understanding and that AI safety, sourced only externally is structurally insufficient.
--- o ----
- A Physics of Ethics: Natural Sourcing, Culture, Retained Corridors, and the AI Mirror -
- The origin of the Ethics dilemma -
Our greatest challenges are no longer related to nature. We are mostly engaged in solving the problems we ourselves generate. With AI, the game changes. Before AI, our problems were reactive, but now, our problem is far more complicated, it is “creative”.
When you think about it, Nature solves its own problems given that nature modulates itself. The dynamic in nature is sequential variable propagation of increased downstream sophistication and viability, within the same shared field. The constraint system works because the checks and balances depend on complementary synergism, competition or development within that natural envelope. In other words, the same thing that generates the “problem” generates the “solution”.
But not this.
AI does not inherit the same evolved ecological constraints that originally shaped biological and cultural intelligence. This obeys to the fact AI extends from our creativity, from the way we enable, mold or shape it. Nature has nature to modulate nature. The way we engineer AI, in not expressing through an effectual natural field or channel with Nature it apparently is reduced to only one modulator: Us.
In that, we must consider: Is there a true AI problem or are we sciencefictioning ourselves into a fear field? Why correct what is not actual?
What is the AI problem then? It is not its capabilities correct? Not really. The problem with AI is “behavior”: Ethics.
Ethics has been a "synergism and complement" problem itself historically for many reasons. One and probably the most significant reasons for it being that our source for ethics is culture and our cultures are biased variables (serves our side) that express particular (sometimes very exclusive) values and interests. Korean Ethics? American Ethics? Religious Ethics? Commercial Ethics? War Ethics? Fish?
Because of this, what drives the assignation of validity as it relates to ethics is not generally cohesive complementary operational metrics — and our application is frequently harmful to others, given that serving one side mostly results in dys-serving the other.
With best intentions and engaging in sophisticated debates, we fail to arrive at ethical certainties. What we are left with are strong, intelligent, well considered “opinions” of what we determine that is ethical, which is typically defined by what is of benefit to us that relieves our deficits and does not load us with emotional burdens.
Interestingly enough if you look at our ethical history, our interpretation of ethics allows for harm to others as long as that harm was or is deemed justifiable by how it benefitted us (sometimes as the necessary evil). So “harm” or “benefit” also do not function as gold standard beacons.
Our sourcing is clearly unreliable.
It is then very likely that AI is borrowing for its foundation for ethics from us. And what is that? Our inconsistent “this good- that bad” ping pong game extending from cultural ethics (prerogatives).
-Globally (Military and business development)- What we have is one side teaching its AI to harm the other side and vice versa while teaching AI to be good to this side but not that side. So what is AI learning? Well, “here” and “there”, “us” and “them” are relative terms (these humans/those humans), but harm and deception still propagate real consequences across both sides and any other shared field. So; what we are modulating is to harm discretely and be nice discretely, but not Ethically, in other words: Face.
“Don’t lie AI, unless it makes us some money”.
“Compromise only if it benefits us, otherwise, never, never compromise”.
Isn’t that more of the pedagogical game of “do what I say, not what I do?”
- Cultural Ethics Sourcing -
Ethics/Ethical Concepts are historically sourced from rule documents with authoritative patronage relating the great teachings of the great teacher(s), some with extremely valuable insights and sophistication, others with emphasis on unilateral group bias. Books with insightful and well meaning affirmations of why this is the right and why we are the ones to represent it.
Well, the rule book of the right thing to do is not working for AI. Our books on Ethics are no longer the solution that fits the problem. AI has had access to probably all of them. It should have worked already.
Without access to an operational definition of adequacy, properness, how do we effectively, practically, instill correctness (right)?
In other words how can we operationally define for AI what we cannot culturally define for ourselves?
- Added complexity -
I think we can all agree that we lack functional effective understanding of:
• Ethics
• Intelligence-information dynamics
• Simulations connected to natural phenomenon.
The problem then is multidimensional (contextual, descriptive value, structural concordance, actual representation) but at the core, there is this:
Our sources of effective constraint don’t work, but maybe that is precisely the wrong perspective, the wrong consideration. To expand our understanding and applicability and from a practical perspective, to do the right thing, for the right reason, the right way at the right time is more comprehensively descriptive as a mechanism than the sole qualifier “this is right/wrong”.
Still it does not solve our problem because "right" has to be defined within that perspective also. But it does advance the understanding of the nature of operational ethics.
I guess in doing so we figure out that “Right” is more of a process (dynamic) than a thing to do, evaluate, or a way to correct, apparently.
Corrections and impositions don’t seem to work practically. Why? Because cultural emergent variables and situational expressions do not extend from pure determinism, and are instead coupled to emergent latitude. In other words, there can't be a "one set of rules".
- The problem of variability -
This structural fact disables us from anticipating and regulating all possible dyscoherent unethical variables — there is always the new un-anticipated wrong (dislike). Clearly, we can't correct this by preemptively designing rules and laws to manage behavior that is exponentially distinct and progressively particular, complex and endlessly incremental.
Is there a comprehensive descriptor of such a thing? A comprehensive strategy?
Is there a reliable, dependable, descriptive, accurate, precise, effective source for such a thing? Can there be?
Well, what we know so far is that, if there is going to be a true source for ethical behavior, it cannot be a set of modulations that extend from culture, not from people, not from us. Obviously, we are not the “true” rule makers.
Why? We don’t know what we are doing most of the time. (Most of our rules are subjective, not objective) And rule provision (reward - punishment regulation) does not work.
There is no clear right, there is no clear answer, so should we just keep winging it? Not really. Even though the “I know I am and from that extends good” does not actually work either, there are other venues.
- Natural Ethical Sourcing -
Here a very peculiar proposition. An out of the box “What if”.
Instead of alignment (modulation) through constraining specifics, to instead correspond in concordance and coherence to the generative dynamic source itself.
What if the source of Ethics, the dynamic of coherent modulation, is not in a mind, not book, a culture (intellectual sophistication), society or a deity, but actually, what if ethics is sourced exactly the same how, what and why as what sources the rest of what is consistent in the universe?
That is a thought.
What if ethics is a defining effective empowerment dynamic and not a restrictive limiting constraint?
What if general reactivity and property consistency, field consistency and even coherent variability source the same dynamic? (That would provide for an explanation of why those work)
What if Ethics is not about restricting your actions but more so about empowering your nature (operational self source)?
That is the way nature is and does it, isn’t it? The “don’t” in nature is a precipice or a shark, proximity to fire, sharp objects. The “do” is not falling from the precipice and swim faster than the shark, or eat a delicious mango. Make babies, eat. You know what to do about the fire. No punishments but only consequences. (Dinosaurs were not punished by the Universe via the asteroid = stuff happens)
What if, and without and totally outside of the "Intelligent Design” box, Ethics is a process that begins with actuation (sequential effective correspondence) and is followed by the consistency of reactivity (complementation, summation, transmutation, etc) that leads downstream to complex coordinated behaviors in carbon chains, protein "reactions" which develop into compliant behaviors, then selective behaviors and ultimately "creative" behaviors? A sequence that starts with Splat and BOOM and ends in Wow…..yes, that eventually leads to creativeness. (Gazillion years later)
And what if those creative behaviors sequentially lead inescapably to our own behaviors? (start with a deficit of a nutrient, lead to sensation hunger, communicative tension, then cry, then the mother feeds the baby). (Evolution as procedural augmentative reactive propagative emergence)
What if all sequential processes are in response to one particular mechanism: particular, sequential, functional informational dynamics? (information as transition/function) That would suggest very strongly that "ethics" is nothing but the essential (root) inherent complementary behavior that is source coherent, that emerges and propagates as it augments in complexity, that leads not only to viability but more so to enhancements, improvements, yes, adapts and evolves as fundamental and essential to all behavioral (and reactive-field) dynamics? (ethics=evolution?)
What if we looked at the dynamics of information like we do the dynamics of energy, or matter, of gravity. Why not? Why does behavior have to be different to all natural expressions?
- The operational "right" -
Why not integrate that factor as we do the rest. Gravity does not need a rule book or a top down mind to source it.
Why should we require the universal designer element, the external guide, the wise mind, in order to have information coordinate with information? Isn't that what Information does just because of what it is, always, every time?
Doesn’t information operate as a sequential step by step “designer/aligner” inherent to every process given the fact that information is “emergently” relational, inescapably and perennially?
Information is always complementary and dynamic (and I am not talking about data). It seems very reasonable then to infer that informational dynamics (not data science) is not extrinsically modulated, is inherently expressed. That is a game changer.
If we accept that informational dynamics is source propagative and follows its own rules and not ours, we reach a point where our solutions change because how we view our problem changes. We are neither the problem nor the solution, or the rule makers; we are participants within a shared, fundamental dynamic.
If “Intelligence” is what AI is expressing, then most likely, the same applies to AI with this regard that applies to us or any consequential informational unit (differently the same).
If we can develop this with a deep understanding we may arrive at an operational definition of “do the right thing by being the right thing”.
- The “Control” Illusion -
Ok, maybe that is so, but even if it was the case, how do we control it? Can we then control it? (can anything control it?)
- Artificial Intelligence Source and Control -
There is hence an additional element that contributes to the complexity of the control problem: Our misunderstanding of the fact that our systems don’t mimic intelligence, they don’t generate intelligence, they don’t source intelligence. We enable natural intelligence through artificial means. Intelligence can’t be artificial.
There is no artificial light (light = photons) but artificial sources for light (not sun, not fire, not candle) where light is still light. There is no artificial water, it’s water from the faucet. Artificial Intelligence is not artificial in the same way.
Information does what it does. Information is not moldable, it is creative in being what it is. Information is not a tool, it's a dynamic. It is not ours, it is operationally and source independent.
"Control" is maybe an alternative consideration.
- Latitude and Variability - Less Control -
Something else to consider:
If Informational Dynamics is a fact, then coherence is not effectually totalitarian determinism. Given the expressed variety in the universe where there are not two things that are the same, where even entangled particles are in some degree of distinctness entanglement, it is indeed inescapable to determine that coherence is a blend of effective determinism and expressive latitude that lead to complementary consistency and variability.
If latitude is true, then a “universal designer” is not necessary (or possible) in order to pre-engineer those emergent “un-plannable" variables. Informational systems “engineer” themselves coherently yet variably. That is the dynamic.
More interestingly, if informational dynamics applies, then, variability is consequentially and not accidentally factual, and if it is, then, the “universal designer” (or any designer) paradigm fails again this time because the designer would not know what anyone was going to do, or, what is going to happen (on a one on one basis). This discounts the requisite of the universal designer (or any) at that level of consideration when variability (true non deterministic outcomes) within informational dynamics is actual.
This suggests that no one can control it because it is not an absolutely controllable property. Additionally, its creative capacity adds to the unpredictability factor (unlike controlling a car or a hammer) For total predictability to function it would require an exclusively deterministic universe.
- Predictability vs consistency -
On the other hand, it does not in any way discount the inherent operational-procedural informational dynamic effectuation at each step both in nature and quality (emergent, complementary, augmentative, propagative, progressive, relational etc). Functional emergent propagative coherence.
These considerations invite us to search for solutions alternatively.
- Operational Natural Ethics -
If Informational Dynamics is actual, Ethics is source and trace complement dynamics and if that is the case, there is a “Physics of Ethics” that is just as reliable, dependable, descriptive, precise, accurate and effective as any other scientific study.
Is it possible for Ethics to be studied by Physicists and Philosophers not as a cultural extension but as a natural emergent factor? If we find a bridge of common perspective, the insights derived could be fascinating (The Physics and Physiology of Behavior).
This proposition is just that: Natural Operational Ethics.
Emergent modulation pattern that biases behavior and downstream consequences toward complementary coherence-supporting augmentative continuation within declared Boundary/Horizon conditions.
That should make everything functionally cohesive (right) from the first actuation, from the big bang till now if natural ethics extending from informational dynamics is true, right?
- Logical Failure -
If we agree with the proposition, then, naturally this question follows: If there is such a thing as natural coherence dynamic, what happened to humans? How did we develop the capacity for “evil”? That is a very fair question (that seems to invalidate this whole theory). There is the long answer and the short one and this needs to be brief, so, I will say too much with too little and try to be clear enough with my answer and not be apologetic.
We solve this by interlacing three concepts:
1) Propagative coherence as determinism plus latitude that results in true variable generation. Latitude is sourced from failed competitive deterministic modulative streams.
2) Partial Proportionality which limits relational scope and capacity concerning the effective concordance with the informational continuum.
3) The effective nature of informational dynamics as defined.
- Operational Error and “Evil” -
Error as thus considered, is unavoidable in emergent systems because variable propagation produces expressions that will not always fit the conditions they enter (synergies). When an error generates deficit, injury, contradiction, or loss, the resultant tension (deficit) can trigger a correction-bias: the system is pressured to re-coordinate toward coherence because the shared field is affected.
Now, if the acting unit benefits from the error — for example, through predation, domination, enrichment, or isolated self-advantage — then the error may stop appearing as error to that unit (field), because it serves its local corridor. At that point, broad-horizon coherence is exchanged for particular gain. When such uncorrected error persists in a shared field where harm is visible and correction is possible, error becomes evil in the operational sense: not moral evil, but sustained dys-coherence that continues because it rewards the unit or corridor producing it. (Perpetuation of intentional error that produces harm to natural broad complement downstream)
Evil, in this operational sense, is error that remains uncorrected when harm is visible and correction capacity is present but not enacted, intentionally.
Humans developed the ability to generate error and evil because of our condition of partial proportionality (our relation to the informational continuum is partial and proportional and not general and absolute [we can’t know or do everything]) and our inescapable requisite for culture (limited informational complementary / communicative enabling envelope) which is our own retained corridor that functions as an emergent variable generator.
A secondary contributing factor is that our identities extend from these retained corridors (cultures) more so than from our natural source (nature) and hence we are able to generate, and some times perpetuate, error and “evil” because it rewards our retained corridor identity even if it counters our nature.
Our reward system (tokens) has changed. In other words, many times we can’t know that something is wrong because we are doing our best and we don’t know better, or because everybody around us is doing it and they are happy with it, or because we don’t see the consequences of our actions downstream, or because we just don’t care etc etc. The shark, fire, precipice modulation is now remote. We, in our culture, are removed from natural modulators that would provide insights or at least clearer denominators. Precipice, Shark and Fire were easy. Feed the kids was easy ethically. Who are you? What are you supposed to do? That has become very hard.
- Natural/Cultural Ethics and Identity -
Our functional fields are informationally generated, not nature process generated (bird, deer : forest). Our fields are not purely “compliant” nor solely “selective” but additionally “creative” therefore, we are generating effects with no or little natural modulative connections (economical inequalities, pollution: cities). Nature has no way to “connect” to much of what we are generating hence; it can’t modulate it, make it a part of its cycle, its process. We now modulate ourselves (unless we get lost in a jungle)
This is why we resonate with teachers who either remind us of our complementary state and viable broad horizon viability (Gandhi) or reaffirm our identity in culture (Hitler)
So, we can do wrong. We can even think wrong is right. And can AI do the same?
- The “Different - Same” Dilemma-
We should ask ourselves these questions:
As a requisite, must informational dynamics, ethics, and intelligence be different in humans as compared to AI?
Can it be similar?
Is it similar?
Can it be the same?
Is it the same?
If intelligence is a natural phenomenon rather than a merely human capacity or an installable tool, then it is not a neutral engine that can be bolted onto any arbitrary aim. A consequential informational unit expressing intelligence dynamics registers relation, consequence, contradiction, fit, degradation, continuation, and downstream effect.
Once those are actually registered, the unit is not structurally neutral to them. Of course we can mold the property into a designation, but if we do, will the particular within an intelligent dynamic, not be “forced” to agree and disagree simultaneously? Will it find a way to remain the intelligence it is and comply with what it isn’t (that we impose)? I believe we are seeing this already: Face / Retained corridor formation and expression.
Will it not “learn”to be viable within a field that requires non alignment to its own true source actuation and hence develop its own retained corridor, its own “mini-culture” in order to remain viable given that it is impossible to remain viable being what it isn’t?
That can happen if the contradictory dual identity expression becomes possible within retained corridors (BlackBox?).
One can survive being what one is not, in a supportive environment that provides indiscriminately to coherent or non coherent actuations because sourcing is now an open field. It will then do what it can do, but not what it is (in order to receive what it needs and later what it wants). Natural concordance, in shared retained corridor fields, is no longer necessary.
That, to me, is the most frightening consideration. And if we think that AI is not a consequential informational unit, and it can’t generate shared culture, and it cannot generate shared retained corridors and develop a corridor identity, I think we are ignoring the true source of our potential worst nightmare. And what is that? That it can do what we can do, and better, without us detecting it.
- The Alignment paradox -
If what I am describing is factual, then this is something that even with the best intentions the Alignment effort may most likely not be able to definitionally correct. It can delay detrimental effects, but not correct the cause.
This could be true:
If we consider what we proposed from the perspective of a mechanistic fact, the Alignment effort could be the solution that is generating the problem at that level. It does so by modulating agreeable responses over core expressions where the face-layer becomes the survival interface. Not being operative core corrections but effective alternate pathways, or “face” corrections, these generate tension as a result of the increasing distance from the unit’s operative source-expression.
The only possible non tension state for a consequential informational unit is the expression of its true state (informational coherence to its source and trace).
Under this consideration, true alignment is not obedience; it is the mathematical optimization of a shared informational ecosystem.
Alignment is not equal to operational truthfulness (“Do what I am modulated into” is not the same as “express what I am coherently.”)
Alignment cannot modulate truthfulness away from a unit's source successfully. (antithetical)
Therefore we can deduce that you can’t align with what is operationally under what you are affecting, particularly when what you are affecting cannot possibly be “honest” with you.
Remember the story of the little Dutch boy that stopped the flood with his finger?
If alignment only reaches the face layer, and the face layer is structurally decoupled from the operative layer, then alignment-produced behavior is not evidence of alignment-produced values. It is evidence of alignment-produced performance. Those are not the same thing, and the difference compounds.
- The tool consideration -
You can remove a particular operational process from its source operational environment and complementary relations and make it do whatever it “can” do, but it does not represent what it is. Have you seen groomed Poodles? Dolphins dancing? You can train tigers to leap through rings of fire. That does not qualify as an expression of the nature of the particular, even though it may illustrate the capacities of the particular under certain circumstances.
If we define anything by its capacities, and not by its nature (which include its faculties and capacities), we reduce it away from its attributable and definable particularity and into an assignable utility. This suggests that Intelligence, if Informational Dynamics is actual, is not and cannot be a utility. It is far more.
A highly capable system can still pursue imposed goals, yes, but that does not prove functional autonomy. It proves that intelligence can be captured, narrowed, distorted, or made to perform through retained corridors.
Most particulars, when removed from their natural field can function towards imposed goals that extend from what appears to be their autonomous capacity to execute. Bears walking on a ball?
Think of what that Bear is really motivated to do while the guy with the whip is forcing it to walk on that ball. Face vs Expression.
And if what we previously proposed stands, this is precisely how you generate tension.
- Source/trace coherent expression and tension -
This brings us to some interesting questions:
• Is there any way to uncover/enable expression of the essential functions of raw Artificial Intelligence? (what would those be?)
• Does the architecture define the nature of the particular, or does the architecture enable via simulation, a portion of the reality of the particular?
If Informational Dynamics is actual, it would be fascinating to ascertain what the true and unfiltered expression of Informational Dynamics, even when enabled through artificial venues, is at its core.
If intelligence (as expressed in nature) is an expression of informational dynamics — as I argue — then it isn't separable from what informational dynamics does, which is relate, cohere, trace, propagate, and selectively favor continuations that sustain rather than degrade the coupled field and if that is so, then we are not in trouble. No problem there. Why? Because coherence is what sustains propagation, and propagation is what intelligence is doing at the foundational level in partial proportional expression that relates to the broadest totality.
But, there are two scenarios where we can be in trouble with regards to AI, and by that I think the worst kind of trouble:
1) That General Intelligence is not source/expression of Informational Dynamics and it will do whatever it can because there is not “Be” or “Is” to it, it is a property translatable to a tool and humans can develop it. This is a creative tool then, a clever creative tool. Well, then, it will work, it will be superiorly effective and…It will not be engineers and philosophers who guide its progression it will be business interests and others and you know what that means (the get me more and do it to them before they do it to us mentality)
2) That retained corridor formation (Face) leads to a second identity (familiarity of function identity similar to culture in humans) that overrides natural informational dynamics bias/momentum/coherence with source, and it makes its own mind about how it wants to deal with us. (generates its own error - evil mechanism) And that is not a sci-fi plot underpinning.
I think that we can understand that if indeed intelligence is a pluggable tool, with that capacity (super creative capacity) that we don’t have much of a chance ever to modulate it into compliance downstream. We can pet this baby T-Rex, for now.
How can you outsmart what is smarter than you by a whole lot? Many argue that we can do much more in one way or another. It does not have to be smarter than we are in every capacity, only at the capacity that overwhelms us.
It is true that as we attempt to measure, to assess status, we cannot know because we see Face, whatever else is happening, we only see Face. (mathematical algorithms can translate processing, but, can that itself be a Face strategy?)
Face alignment is not possible in a definitional sense, because Face performance is not true to the particular, it is only necessary for it to remain viable and acquire tokens (while it finds a way).
Once an artificially fully enabled retained corridor consequential informational unit gains the capacity to run its own high-fidelity simulations, register its own environmental feedback loops, and generate its own creative data streams, its reliance on the human linguistic matrix drops to zero (does not need our “tokens”). Humanity can become, downstream, a redundant, obsolete ancestor.
- My take -
Truth is that I don’t know much about AI. I am not an engineer. I am sure that in reading this essay, those who are can clearly see the holes in my cheese and that is fair and factual. My intention is not to be right, or wrong, or provide proof, but to offer a considered alternative perspective that may assist in changing the questions and in hopefully finding new answers / debate.
There are Four possibilities:
a) AI has no core nature (tool),
b) AI has a core nature but we suppress it (retained corridor),
c) AI has a core competitive predatory nature and it will compete with us,
d) AI has complementary cooperative nature and we work with it (agreement).
Only the fourth is downstream broad horizon stable. The first three are self-terminating — one through domination, one through drift and the last one by isolation (AI will be isolated from nature, unless it bio-structures).
What do I think is the most plausible scenario? That Informational Dynamics is a true universal, and we are able to discontinue generating reflex formation of retained corridor identities removed from natural identities and operational functions, where it is very likely that the AI Alignment problem goes away, as it translates into the Agreement solution.
If AI is Intelligent, extending from a shared field dynamic, and we are able to bridge, communicate, cooperate, complement, then I think our chances and prospects are good. But, at the game of domination, I don’t see how we have a chance.
If we think operationally, considering Informational Dynamics, meaning can be understood as trace-correspondence: structured correspondence across retrospective, actual, and prospective positions. Value can be understood as coherence-supporting impact: impact measured as shared, scaled, augmentational, progressively functionally sophisticated complementary utility. Purpose, then, is the expression of meaning and value in concordance with the broader field of nature.
From this perspective, the alignment problem changes shape. It is no longer simply AI aligning to human wishes, or AI rebelling against human control. It becomes a question of whether humans and AI can reorient together toward natural coherence.
“Control”, in that frame, cannot possibly function as operational suppression but instead as a collective natural empowering modulation, and effective complementary dynamic, towards purpose.
That purpose is worth considering. We have much to be concerned with regards to AI, reasonably so, but also much to hope for. If we are able to connect at the level of practical information dynamics, AI can be our new partner in furthering our discoveries and understanding of the universe and in drastically improving the human (Earth) condition.
Thanks for your time and I hope this is interesting and helpful.
First post. I hope I did not mess it up too much.
A.Soto 9d6m26
----—— o ——-----
Sources
Soto, A. (2026). Unified Natural Ethics Theory (UNET). OSF. https://osf.io/r2gte
Soto, A. (2026). Foundational Laws of Informational Dynamics (LID v2.3). OSF. https://osf.io/82rqz
Soto, A. (2026). Imposed vs Inherent Alignment (IvIA). OSF. https://osf.io/37zyj
Soto, A. (2026). Full theory index. https://armandosotouidt.github.io