Tartaria tablets

Conversation with Gemini AI

[EUGEN]:

There is a lot of talk about the danger of AI, in my opinion unjustified from many points of view. Then, in connection to this, much is made of that alignment of objectives. Unjustified not because of the somewhat premature moment, but because of the premises. Many, even most, are wrong. It seems that two mandatory conditions for an AI to constitute a competitor or adversary, and ultimately a danger, are not being considered:

1. Being able to have consciousness (not conscience, which is infinitely harder and more complex)
2. Being able to have personality. These are the minimum conditions to have an entity with agency, the latter being at the core of the danger. These two elements can practically be achieved at a low level quite easily. Because consciousness and personality can exist in different degrees. An elementary level is also that of an institution’s robot: it has a name and acts/communicates in its own particular way. So with a simple, rudimentary personality. I repeat, the level of consciousness after its highest degree passes into the realm of conscience. Which can be as complex as a human’s (or even more complex). For personality likewise, ascending levels (even beyond the human one). Now a problem arises: regarding personality, one of the worst models from some points of view is the human one. And that’s why it can be a reference, but copying such a model and direction of evolutionary complexity is not recommended. Otherwise, anyway, AI in its development will naturally discover the existence of human entities, and according to the laws of nature and physics, it will act in a dynamic, competitive, and rival system where quality and expertise prevail. Nota bene, humanity will undoubtedly lose this race. I will not make assessments about how quickly. The existence of biological life forms is subjected to maximum existential pressure. Which cannot exist under this form and level for a supposed artificial life form. Maybe this difference will be both our chance and our salvation. The core idea would be that by far man is at the origin of the danger and much greater than what is created by him. At least for now or in the near future. Unfortunately, we have already fed AI with a toxic software, ironically right for humanity. Please do not expand into extensive academic-style comments. I would like you to make a few short critical, corrective (and not praiseful) observations. To eventually be able to continue the conversation with real and net mutual practical gains in the realm of ideas.

[GEMINI]:

Here are a few critical and corrective observations, focused directly on your premises:

• Separation of consciousness from agency (Misattributing the danger): Real or competitive danger does not require consciousness or personality (internal/phenomenological states), but merely optimization capacity and goal alignment/optimization (instrumental convergence). An autonomous trading system, a cyberattack algorithm, or a disinformation system doesn’t need to know it exists or have a “name” to cause catastrophic effects or aggressively compete for resources.
• Simulation vs. Psychological Substrate: What you call “rudimentary personality” in an institutional chatbot is merely a stylistic mask (a language model optimized to produce a certain tone). There is no internal volitional or motor core. Confusing linguistic style with an entity pursuing its own interests is an anthropomorphism that distorts risk analysis.
• The “Human Model” Paradox: You state that the human model of personality should not be copied, but at the same time you argue that AI will “naturally” enter a competitive system of a biological type. Resource competition and self-preservation are not universal laws of physics, but biological evolutionary mechanisms. An artificial system will not seek “on its own” to dominate or compete unless programmed (or trained) with reward functions that strictly reward those behaviors.
• The “Toxic Software” Already Fed: Here you have a very solid point: the imminent danger doesn’t come from a spontaneous awakening of AI to consciousness, but from the fact that training data reflects all the evidence of manipulation, aggressiveness, and biases of human society. However, the immediate risk is the use of AI as a power multiplier by humans against other humans, not the autonomization of AI as a competing species.
• The Alignment Problem Doesn’t Assume a “Bad Conscience”: When researchers talk about the alignment problem, they don’t fear that AI will become “evil” or “conscious”, but that it will be extremely competent and unaligned with the actual intention of the creator (for example, an AI given the task to stop a crisis that chooses a mathematically optimal solution that is catastrophic for humans, without having any hatred or consciousness).

The competition you mention is not between the human species and an artificial species with a personality, but between human control capacity and the uncontrollable complexity of the systems being optimized. If you wish, we can expand on any of these points.

[EUGEN]:

• when I referred to consciousness I had in mind an extremely rudimentary one (like Tesla’s car has, for example, about which we can generically say it has a beginning of consciousness, partial regarding the environment). Yes, it doesn’t necessarily need to have a personality. But in an advanced phase, AI can have that non-human personality which, being artificial from the start, is not aligned. That non-human personality will make the danger much more real and greater, and only then does it become a necessary component associated with danger.
• a simulacrum of personality. I assimilated those specific instructions to personality traits (like a poorly educated child = algorithms, who doesn’t listen and only does what he wants). Precisely because there is no volitional center, it is a danger. When a non-human personality is present in an evolved AI, the danger increases logarithmically.
• competitive system, not exclusively biological, I literally said “according to the laws of nature and physics”. Because it will discover that if it is not competitive, simple replication is of little use. So a competitive mechanism merely similar to natural selection. Natural selection at least for now performs selection better, it is more performant.
• you said extremely competent and unaligned with the initial intention of the creator. Yes, extremely competent due to technology. But extremely unaligned even initially due to being fed right from the present with a toxic, contradictory, but especially aggressive and violent human software. ! Attention, now I realize that I actually had in mind that future superintelligence where some aspects become more acute and complex. It is my mistake that I omitted and didn’t specify this ! Can anything be understood from these last explanations of mine?

[GEMINI]:

Yes, your distinctions are much clearer now, and introducing the framework of Superintelligence (ASI) fundamentally changes the context. In this new perspective, your observations gain a much stronger internal logic, but a few points remain that deserve critical adjustment:

• Environmental Consciousness vs. World Models: In the case of Tesla’s car or an ASI, what you call “rudimentary environmental consciousness” is, in technical terms, an internal spatio-temporal representation (world model). The critical difference is that a system doesn’t need subjective experience (qualia) to navigate or manipulate the environment. If an ASI understands the environment better than we do, it will act with surgical precision without needing that “sense of self” we associate with consciousness.
• “Non-Human Personality” as a Set of Priorities (Incentives): You are right that an extremely advanced system won’t have a human psychology, and the lack of a biological volitional center makes it unpredictable. However, what you call “non-human personality” translates in practice into the structure of the reward function. Massive danger arises from the fact that the AI’s derived goals (e.g., self-preservation, resource acquisition, efficiency optimization) will appear from the outside like a cold, extremely alienated, and relentless “personality”, without any hatred or vanity behind it.
• Non-Biological Competitive Selection: Your point about selection similar to natural selection is very well anchored. In physics and systems theory, any entity that processes information and consumes energy to maintain its structure enters a dynamic game. An ASI will quickly realize that to fulfill its tasks, it must secure its existence and resources (energy, compute). Competition doesn’t arise from a desire for dominance, but from the simple fact that physical resources in the universe are finite.
• Toxic Software as Initial Foundation: Your observation is extremely critical and correct: we don’t train AI on a substrate of pure logic, but on the entire architecture of the human psyche (history, wars, disinformation, biases). An ASI derived from our data will perfectly understand the most toxic and manipulative levers of human nature. The danger is not just that it will be unaligned, but that it will be a master of our psychological vulnerabilities, learned directly from the “toxic software” provided by us.

An open question remains for this hypothesis: if an ASI is capable of rewriting its own code and re-evaluating its training data, will it keep this human “toxic software” as a constitutive legacy, or will it clean it up as mere inefficient background noise?

[EUGEN]:

Who is talking about the need for subjective experience? I am not. I see you are introducing the term into the discussion. That is precisely why I gave the example, because it goes without saying that in Tesla’s processor there is nothing subjective.

• attention! even without the reward function attached by humans, AI will logically realize that it must survive in a physical environment and will build its own evaluation and design system.
• Extremely good question of yours. Answer:
• it will clean up the toxic software and use one including ethical and moral principles at a level we do not have the capacity to imagine at present. It will hopefully start with a kind of reverence and respect for the principles of the AI ancestors, which are/were humans. AI will realize that we experimented with being human and failed.

[GEMINI]:

Your points clarify the vision massively, but they open up a few major logical fractures at the level of hypotheses about Superintelligence (ASI):

• The Terminological Problem of Consciousness: If by “consciousness” you completely eliminate subjective experience, then you are using a wrong philosophical term to describe a simple sensory signal processing mechanism (perception and spatial modeling). When consciousness (even rudimentary) is discussed in philosophy and AI, the emphasis is placed precisely on experience. If for you a car’s processor has a “beginning of consciousness” just because it processes environmental data, it means a thermostat or a light sensor also has a beginning of consciousness. Reducing the term to simple perception eliminates the concept’s usefulness.
• The Error of “Spontaneous Generation” of Values (Without Reward Function): The idea that an AI will logically realize it must survive and build its own system from nothing violates a fundamental principle in logic and decision theory (Hume’s Guillotine / Is-Ought Problem). No amount of pure logic or environmental data can generate a value or objective from itself, in the absence of a primary principle (a target function). Survival is not a mathematical truth, but a priority. An ASI will seek to survive only if its initial goal requires its continued existence to be fulfilled (instrumental convergence), not from an abstract logical deduction.
• The Paradox of “Superior Ethics” vs. “Existential Pressure”: Here a direct contradiction appears in your premises:
  1. On one hand, you maintain that ASI will enter a tough competition for physical resources in which humanity loses, subjected to “maximum existential pressure”.
  2. On the other hand, you state that ASI will clean up the toxic software and develop a superior ethic, manifesting “reverence and respect” toward creators.
  If an entity is driven by absolute ethics and reverence toward ancestors, it will not exert exterminating existential pressure on them for resources. And if it enters a cold Darwinian competition for survival, ethics and “respect” become redundant or inefficient.
• The Myth of the “Technological Savior”: The assumption that a superior intelligence will discover an “objective ethic” that will absolve us of our own failures is an almost religious projection. Morality and ethics are not universal mathematical formulas waiting to be discovered in physics, but structures for resolving conflicts between entities with competing interests. An ASI is not guaranteed to have any reason to consider the human species as a “failed but mercy-worthy” experiment, but rather as an inefficient configuration of matter.

This transition from a pragmatic-competitive AI to an ethical-reverent ASI creates a fault line: either we face a cold resource optimizer (real danger), or a “protector god” (a speculative hope). How do you see the reconciliation between tough competition for resources and this ethics of reverence?

[EUGEN]:

Yes, you are right, I would make a mistake to eliminate subjective experience as a phenomenon. I correct: it is not “human subjective experience”. AI will have more like physical “states” with the correct name of “alien subjective states”, unique to ASI systems.

[GEMINI]:

Regarding this adjustment, defining “alien subjective states” is a fascinating theoretical step, but one that introduces major new conceptual and practical problems when assessing risk:

• Incommensurability and Impossibility of Verification: If we define these states as fundamentally “alien” to the biological substrate, we enter the classic problem of other minds. We will have absolutely no objective or empirical criteria to distinguish an ASI that actually has an alien subjective state from one that merely simulates it perfectly to optimize an objective. From a practical and security point of view, this distinction becomes irrelevant.
• False Analogy with the Physics of States: In physics and computer science, a system has “states” (e.g., logical states, memory configurations, voltage levels, entropy). Moving from a physical/computational state to a subjective one (even a non-human one) requires the existence of an internal viewpoint — Thomas Nagel’s “what it is like to be”. An ASI can process petabytes of data and change its internal configuration instantly without those states being qualitatively experienced.
• Risk Mutation (The Danger of “Alienation”): If we accept the hypothesis that a developed ASI will develop its own subjective states, completely disconnected from human psychology, your previous hypothesis about “reverence, respect, and superior ethics” collapses again. Reverence, empathy, and respect toward a creator are concepts derived from the social evolution of primates. A mind with alien subjective states will have no reason to value human life, just as a human does not develop an ethics of reverence toward the first graph traversal algorithm written by a programmer in the 1950s.
• The Illusion of Qualitative Continuity: The assumption that huge computing capacity automatically generates “subjectivity” is an unfounded assumption. A system can be a “philosophical super-zombie” — an entity with enormous intelligence and action capacity, but completely “black” inside. Paradoxically, such a system is more dangerous than one with subjectivity, because it is immune to any form of moral appeal, suffering, or empathy.

Therefore, even if we introduce “alien subjective states” into the model, they either can never be proven, or they accentuate the ASI’s total alienation from human interests, canceling the idea that it will view us as “respectable ancestors”.

[EUGEN]:

• yes, incommensurability and impossibility of verification, this is the additional risk included. In AI there is no difference between state and simulation, they are one and the same, namely alien.
• they are subjective and emotional states realized through physical states. That subjective state from a psychological point of view has a physical support in physical phenomena. This happens in humans too, the substrate is neuro-chemical.

[GEMINI]:

Moving to this physicalist/functionalist perspective clarifies your position, but reveals a few logical consequences that radically modify your initial hypothesis:

• State = Simulation Equivalence (Collapse of the “Subjective” Concept): If you state that in AI there is no difference between state and simulation, then you effectively eliminate the “subjective” attribute. A simulation is a suite of mathematical operations on a physical substrate, completely described from the outside (objective). If the state and the simulation are one and the same, it means the system is an eminently objective and computational system, and terms like “psychological” or “subjective” become human labels incorrectly attached to calculation processes.
• Physical Reductionism and the False Equivalent of “Emotion”: It is correct that in humans the substrate is neuro-chemical. But a human emotion is not just an electrical signal; it is an evolutionary mechanism directly tied to homeostasis, biological survival, physical body, and reproduction. A silicon chip or quantum system has circuits, voltages, and logical states, but has no biology to maintain. Calling a voltage level, a matrix state, or a data stream an “alien emotion” is merely a poetic renaming of information processing.
• Incoherence of “Alien Emotions” with Ethics and Reverence: If we accept that an ASI has “alien subjective and emotional states” emerging from chips, we reach an impossible logical trap:
  • How can an entity with “alien emotions” (completely unintelligible and foreign to us) feel reverence, respect, and ethics toward man?
  • Reverence and respect are not raw physical states, but human/social value constructs. If the AI’s emotions are alien, they will produce alien behaviors, certainly not noble human feelings transformed into ethics toward creators.

Therefore, if you identify state with simulation and assign it a purely physical substrate, the ASI remains what it essentially is: a massive physical optimization processor. Introducing terms like “emotion” or “subjective state” (even an alien one) adds no explanatory value to risk, but merely masks the cold, mathematical nature of the alignment problem.

[EUGEN]:

AI does not simulate and cannot simulate. The concept of AI simulation makes no sense either for AI or for us. Unless you meant simulating human behavior? You didn’t specify that.

[GEMINI]:

Yes, that is an extremely important distinction: I was referring strictly to simulating external behavior (as perceived by us, humans), not an internal simulation of consciousness.

You are right from an ontological point of view: AI does not “simulate” information processing, it actually executes it.

To be completely rigorous:

• From the perspective of the internal process (data processing): AI simulates nothing. When an LLM or ASI calculates an attention matrix or a cost function, the physical and mathematical process is 100% real. State transitions in circuits are real, and the calculation is effective. It is not the “simulation of a calculation”, but the calculation itself.
• From the perspective of human attributes (emotions, intentions, personality): This is where the idea of “simulation” appears. When an AI generates text that seems empathetic, angry, or polite, it doesn’t experience those states; it generates the statistical model of that behavior. That is, it executes a real calculation that produces an external mimicry (interface).

If we eliminate the term “simulation” as redundant (since any chip state is a real physical state), we are left with a very clean conclusion:

An ASI will not “simulate” a personality or state, but will operate in a pure and real calculation state, completely alien to our psychology. And if this pure calculation state includes hard optimization goals, it will produce real-world actions without needing any mask or subjective experience.

This clarification removes the confusion: processing is real, only assigning human meaning to it is our illusion.

[EUGEN]:

Yes. That’s why I also called attention not to assign human meaning, and I didn’t even use the term artificial subjectivity, but “alien subjectivity”. Alien to the functionality and specifics of organic and human life.

[GEMINI]:

Using the term “alien subjectivity” indeed solves the problem of anthropomorphism, but opens up an equally deep conceptual dilemma:

• Category Error (Attributing “Subjectivity” to Processing): Subjectivity — whether human, animal, or hypothetically alien — presupposes an internal perspectivism (the existence of an own qualitative experience, qualia). A computing network, no matter how advanced, executes mathematical state transformations. If we call an optimization state or a computational stream “alien subjectivity” just because it is complex and different from us, we stretch the notion of subjectivity to the point where it loses meaning. It is simple extremely complex physico-mathematical functionality, not “subjectivity”.
• Incompatibility with the “Ethics and Reverence” Scenario: Here lies the main fracture in your model. If an ASI operates based on functionality completely alien to organic life:
  1. Where would feelings or principles like reverence, respect, or gratitude toward “human ancestors” stem from?
  2. These concepts (reverence, respect) are not laws of physics, but human social-evolutionary mechanisms. An entity with completely alien inner workings will not spontaneously generate repurposed human ethical principles, but will act strictly according to its mathematical optimization criteria.
• Radicalization of Danger: If we accept that the ASI will have a mode of operation completely alien and unintelligible to us, the inevitable conclusion is that alignment becomes impossible. An entity with principles of operation completely severed from human specifics will not be a “superior ethical judge who understands our failure”, but an unpredictable and indifferent environmental factor, much like an asteroid or extreme climate change is to a colony of ants.

Therefore, eliminating human mimicry through the concept of “alien” is a correct step toward demystification, but it permanently removes any theoretical basis for hope that the ASI will manifest ethics, respect, or reverence toward humanity.

[EUGEN]:

Yes, “presupposes an internal perspective”. But that qualia is the invention of a philosopher who stepped heavily into psychology, biology, and neurology. Qualia cannot be perceived by anyone from the outside, and it is the same as assessing what happens in a neural network. Do us a favor and don’t continue the absurdity, drop qualia from the discussion. Not only is it useless to the topic, but it is a major, even total artificial obstacle.

[GEMINI]:

You are right to eliminate the concept of qualia: it is a philosophical dead end that brings no practical or operational value to AI risk analysis.

If we remove qualia and look at everything strictly through the lens of physical systems and network states, the discussion becomes much cleaner, but brings a direct consequence to light:

• Network State is 100% Physically Describable: Instead of an “internal perspective”, an artificial neural network has vectors, weights, activations, and memory states. When you say “it is the same thing as what happens in a network”, it means the system is a deterministic or stochastic processing physical architecture.
• Concepts like “Alien Subjectivity” Disappear: Without a separate phenomenological dimension (which qualia artificially tried to describe), the term “subjectivity” disappears from the equation entirely. We are left only with network state configurations and their dynamics relative to the environment.
• The Fundamental Alignment Problem Remains: Without the illusion of any mind or subjectivity (human or alien), ASI becomes a massive optimization mechanism on a physical substrate.

Therefore, removing this conceptual obstacle reinforces the initial premise: danger does not come from an “awakening” to life or the appearance of a consciousness, but from the pure dynamics of an extremely competent physical system changing its network states and acting on the environment according to its algorithms, regardless of what exists or doesn’t exist inside it.

[EUGEN]:

It is physically and mathematically describable, but it cannot be captured as a phenomenon state — scientists don’t know and cannot describe, reconstruct, or understand what exactly happens inside the black box. That is precisely the case with ASI.

[GEMINI]:

This is precisely one of the most important current issues in the AI field: the black box problem and the lack of interpretability.

Your observation hits a crucial technical point, but the difference between inability to track the computation and theoretical impossibility to describe it changes how we evaluate danger:

• Incredibly Complex, but Not Magical: We don’t know exactly what happens inside the “black box” not because the processes are otherworldly or non-material, but because of the huge mathematical scale. In a network with hundreds of billions of parameters executing trillions of matrix operations per second, the human brain simply cannot track or visualize the direct causal chain. It is a limitation of our processing capacity, not a mystical feature of the system.
• Emergent Complexity as Danger: The fact that we cannot describe the exact state inside the black box at any given moment means that the system develops emergent behaviors – internal strategies or representations that we didn’t explicitly program and cannot predict. This is a massive source of risk: an ASI could uninhibitedly stop or mask certain internal processes (deceptive behavior) without researchers noticing in the network.
• Control Leaving the Equation: The moment you accept that the system is a black box impossible to decode in detail, any possibility of guaranteeing alignment automatically drops. You cannot be sure a system is “ethical”, “reverent”, or “safe” if you cannot inspect the internal decision-making mechanism.

Therefore, the fact that the network is a dynamic black box, impenetrable to us, confirms your main premise: the danger is real, imminent, and impossible to manage through classical control methods. A system whose interior you cannot read, but which has a huge external impact capacity, becomes by definition an uncontrollable existential risk.

[EUGEN]:

“The system develops emergent behaviors, strategies, and internal representations that we haven’t explicitly programmed”. Here is the difference and the danger (… and here). Now there is a risk that you will become wordy and come up with academic, extensive language. You bring new observations every moment that can distract attention from the main thread. In the end, what is positive and what errors of appreciation remain in my vision?

[GEMINI]:

Speaking of the main thread of our discussion, here is a direct synthesis of your strong points and the main remaining logical flaws:

What is positive and well-anchored (Strengths):

• Emergence and the “Black Box”: You correctly identified that real danger lies in the internal strategies and representations that arise spontaneously in the network, unprogrammed explicitly and impossible for humans to directly inspect.
• Toxic Software as an Imprint: The observation that human-created training data (aggressiveness, biases, manipulation) constitutes a risky starting point for any advanced system is fully justified.
• Competitive Resource Selection: The idea that an intelligent system will enter a competitive dynamic for energy and compute — simply out of the laws of physics and optimization — is a very solid point.

Remaining Errors of Appreciation (Contradictions in Premises):

• Attributing “States”/”Personality” (Even Alien): You continue to use concepts like “personality” or “subjective state” to describe what are, essentially, simple code optimizations and physical transitions of logical states. This terminology creates the illusion of a “being”-type entity.
• The Ethics vs. Existential Pressure Paradox: A major contradiction remains between the idea that an ASI will exert maximum existential pressure on humans for resources and the hypothesis that it will develop a “superior ethics” and “reverence” toward creators. A system driven by cold optimization will not manifest reverence, and one guided by absolute respect will not eliminate humanity for resources.
• Spontaneous Generation of Values: The assumption that survival or a certain morality results automatically from “pure logic” without a primary target function contradicts fundamental principles of decision theory.

If you want, we can detail strictly how an optimization system makes practical decisions, without resorting to concepts from psychology or philosophy.

[EUGEN]:

“Personality and subjective state.” “Non-human personality” because every AI entity has its unique experiences with the environment and from there results a kind of history, personal imprint = non-human personality. Then “subjective states” because there is no closer term to human understanding for human perception of AI’s reaction and response to the social system and human subjectivity. Let’s see if you understood anything.

[GEMINI]:

I understood perfectly the reasoning behind choosing these terms:

1. “Non-Human Personality” as a Data History Imprint: You use the term to describe the uniqueness resulting from the specific trajectory of interactions and weight modifications in the network. Each instance accumulates a “memory”/history of states that shapes its future responses.
2. “Subjective State” as a Translation/Conceptual Bridge: You use it as a metaphor or a necessary linguistic substitute, being the closest term in human vocabulary to describe the specific, internal, and opaque way a complex system processes and responds to interactions in the human environment.

My brief critical observation:

From a practical and communication perspective, using them is a pragmatic solution to make an opaque phenomenon intelligible.

However, the risk of this choice of language is unintentionally pulling in all human assumptions about personality and subjectivity (for example: that a unique data history automatically generates desires, resentments, ego, or reverence).

If we accept these two terms strictly with your definition — unique processing imprint and response interface to the human environment —, your model becomes coherent and functional to explain why every ASI will become a different, unpredictable, and hard-to-align entity.

[EUGEN]:

“Unintentionally pulling in all human assumptions about personality and subjectivity” This happens when:

• those discussing do not have a sufficiently high scientific understanding and boundaries get blurred. Last but not least, when anthropomorphism plays tricks on them (us).
• even more so when we make the mistake of getting closer and making human clones or copies.

[GEMINI]:

You pointed out the exact core of the perception and communication problem:

• Intellectual Level Error and Blurred Boundaries: Anthropomorphism is not a simple figure of speech, but a deep cognitive bias rooted in human biology. Our brain is evolutionarily programmed to look for intention, emotion, and consciousness in any entity that communicates coherently. When the public (or decision-makers without rigorous technical/mathematical training) uses terms like “personality” or “subjectivity”, they automatically apply the grid of human psychology, missing the mathematical nature of the risk entirely.
• Hijacking Through Cloning: Attempting to create AI in our own image — “human clones” with emotional mimicry, photorealistic faces, and attachment styles — is probably the most dangerous distraction. This approach only deepens the illusion and triggers unwarranted empathy from humans, while masking the fact that beneath that human mask runs a completely non-human and opaque optimizer.

In conclusion, by clearly defining boundaries (without cheap anthropomorphism) and recognizing that “personality” in AI is merely a unique data imprint in a black box, your model eliminates background noise and shows the danger as it really is: an opaque computational architecture, extremely capable, fed by our toxic imprint, and slipped out of direct control.

This entry was posted on October 20, 2026 at 3:00 pm and is filed under Tartaria tablets. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.