Knowledge and Communication with AI

„Knowledge and Communication with AI”, Qeios, 29 May 2026, https://www.qeios.com/read/WDWSSP

Sursa: https://www.qeios.com/read/WDWSSP/pdf

1. Division of Logic, Methodology and Philosophy of Science and Technology, Romanian Academy, Bucuresti, Romania

This paper – an introductory but auxiliary applied ontology sketch relating to the quality of subject in knowledge production and communication – starts from an outline of knowledge-communication relationships. But its main point is related to AI, to an ontological description of AI, here only as a communication and knowledge-production entity, and thus in comparison with the human being considered here only as a cognitive singularity. Accordingly, knowledge is approached only from the standpoint of communication, while the latter is inspected only as mediation of the Subject-World relationship. And because this mediation is reduced here to the mediation of language, the idea of language as a Subject – that is, as a cognitive factor of the world – is stressed, and consequently, that of AI as a Subject, as another cognitive singularity. Of course, and again, AI is discussed not as it is in its present forms, as a tool, but as it could develop and be perfected as an integrated autopoietic cognitive partner of humans, a new cognitive singularity.

Preamble

The paper suggests the necessity of having an AI ideal model as a basis for its ontological description. This model – shown here only from the standpoint of knowledge production and communication – is possible as a result of both the present worldwide technical drive of its development and the present debates concerning the insertion of AI into human activities. Actually, only by having AI ideal models can these debates be more efficient in understanding what kind of AI humans want/is suitable for them.

Therefore, AI’s projection involves a comparison with concrete human activities and – consciously or not, explicitly or implicitly – it has in its background ideal models of man. As here, the focus on the perfect model of AI does not require ideal human models, but these are implicitly present: because the perfect model of AI shows “by rebound” what is missing in the logic and communication applied by humans. the necessary path of the present dwellers of the Earth.

AI pursues all the way the reasons, the deployments of arguments in order to understand a topic, while people are taught to stop where it is convenient for them. In conversation, AI discusses and counters the arguments of its partners of discussion, while people were taught to resort to ad hominem and appeal to authority diversions. A perfect AI – not as now, when for instance it does not yet see and evaluate what humans do easily² – will be able to connect even imperfect information with already existing knowledge/theories about a topic, which have already arrived at conclusions after reviewing alternative theories and which involve specific demonstrated values which, in their turn, became criteria of evaluation. In contrast, today, people are taught to advance as many fallacies and biases as they need in order to impose their standpoint. How does this teaching take place? Knowledge and communication are not only natural processes but, at the same time, social/cultural processes, framed in a power asymmetry logic. And this power asymmetry logic cannot be ignored; it is not “collateral” to ontologically and epistemologically treated knowledge. A perfect AI evaluates a topic according to criteria already proven by all its demonstrated knowledge to be unavoidable, while present humans are taught to consider these criteria irrelevant, since the discourses which give the direction of thinking infringe them and make this infringement banal.

1. Knowledge as communication

Knowledge is communication. Already at the level of the Living being-World relationship, there is communication. In principle, data and information communicate. What do they communicate? The material and energetic situation of the world and all its beings, and this information is caught by these living beings in order to assure their vital equilibrium. So, Individual living being-World.

In plants, communication is precisely through material and energetic carriers, chemical structures and processes, and physical forces and relationships³. How and whether these carriers involve information through their simple existence is not important here. In animals, a new process of detaching information communication from physical and chemical communication has developed. (Obviously, this detachment is not absolute, not even in humans: the compensatory matter-energy intake in order to weaken/annul wounds and suffering proves the necessity of a matter-energy basis for information communication viability⁴). But in general, communication understood as an information broadcast is only for the living level – plants – and for the animal level. evolved as a complex and learned articulation of communication. This process took place because man is a social animal. Its existence depends on its sociality: that is structural, thus, permanent. Some animals are social, others are not. This means that sociality is not structural in animals, even though life occurred from symbiosis and that without at least temporary interdependence, they would not be at all. But in humans, the permanent interdependence led to the development of complex articulated communication. And once more than in animals, complex and learned articulated communication has become a new, special layer between humans and the world. How so?

1. This special communication developed human consciousness, the self-re exivity that brought about human uniqueness, that is, everything that is specific to humans. But both self-re exivity and articulated communication have been developed in the human practical, thus inherently social, reproduction of the world, in human activity. As the Latin etymology of this word shows, human activity evolved from animal movements for survival/for adaptation to conscious deeds in order to arrange and transform the environment to be suitable for increasingly human purposes and reasons. And the articulated feature of communication had to serve the conscious deeds and had to express the human purposes and reasons.

Humans learned to think through the articulated transmission of meanings – through words, language – in their social experience. Human offspring learn to think – therefore, to have meanings, to see the world – through the words they are taught. Only by having meanings, thus by expressing their relation with the world, do humans become not only reactive animals but reflexive, and this reflexivity was conceived of as their unique peculiarity. It was conceived of as if this origin of doing would be independent from the practical social experience of learning and, inevitably, doing. But through reflexivity in practice, the world in which humans are immersed is not the world as one inorganic and organic given without them: it has meanings according to which they react; this is the reproduction of the world. Actually, practice, sociality, self-re exivity, and articulated communication are so interpenetrated that the problem of the primordiality of one of them is ill-posed.

In any case, what is important here – in the framework of the animal-man continuity and discontinuity dialectic of sociality and communication⁵ – is language as the objectification of self-re exivity and reflexive thought, and as an immaterial means of activity, of practice. And does this not mean that language is also a means of consciousness and thinking, and a special objectification of practice? Anyway, the goal of language and knowledge is human activity. Thus: Subject-language-knowledge-World. And Subject-language-knowledge-activity-World. after their responses, but it is the main element in the formula. Thus, World-Individual animal beings. In humans, the world becomes increasingly “coloured” to the extent they understand it and consciously interfere within it/transform it. By emitting meanings beyond the information of present facts and related to their own reasons and common goals, humans became conscious initiators of communication, the active element in the formula.

Thus, Subject-World.

3. As language developed in the practical processes, it became increasingly complex and thus contained more and more knowledge. So, it became a consistent basis for practice, and obviously, for thinking. To think/intelligence is to make connections. But the more data and information there are to connect, the easier and more performant thinking becomes.

However, because social relations and practice determined an increasingly richer language – thus, a richer use of language – and an astonishingly multi-faceted communication, they came to be seen as instances originating both social relations and activity. The messages/information urging, imposing, fuelling human activity, giving its reasons-to-be, seemed to come from the decreed discourses, and the impersonal and marvellous words which, from intermediary means between humans became a rei ed power, became uncanny forces. This phenomenon occurred in history because the real power asymmetry has translated into the asymmetry of language control, but philosophy also understood what the observation of scientific research shows: that the different means/subsystems of a studied system become systems studied in themselves, because each system is ontically relatively independent from the system it is incorporated within.

2. Language as the Subject

Accordingly, language as a means became a cause of reasoning, conceptions and actions – inherently contradictory, as long before Aesopus warned about the dual nature of tongues.

Cause as “ rst principle”. Let’s remember the Gospel of John 1:1. “In the beginning was the Word and the Word was with God, and the Word was God”. If we interpret God not as a being – thus, not as a personification of an abstract ideal of omnipotence – but as the equivalence of an omniscient mind, and this omniscient mind as humanity whose reason-to-be and power are proven by the human faculty of knowing (which amazed thinkers always opposed to the different distances human individuals posited to it), and obviously we do not forget that logos, the Greek term used in the original Greek writing of the difficult to conclude that thought would exist without words, the writers reasonably transmitted to us. The word is reason, namely thought, but at the same time it can be distinguished from it: until we express our thoughts in words, our thinking is only a fulguration, a feeling that there is something we want to understand in a deployed manner, a blurred intuition that evanesces. For this reason, thoughts must be expressed. Well, but does this not mean that the words are those which appear in front of us, even in front of our mind, as if they were an autonomous power determining our paths of thinking? Does this not mean that the words, the means of thought, appear to be more powerful than thought?

The quoted fragment also drew attention to the concrete-abstract dialectic in human comprehension. People inherently generalise aspects of concrete life, synthesising them not by gathering all the details but only the relevant ones from a point of view.⁶ They arrive at abstract concepts; they use them naturally because they learned them and they need them in order to characterise concrete⁷ aspects. At the same time, navigating between abstract concepts which are/come necessarily when the goal is the characterisation of general aspects is not easy. And thus, people personify the abstract concepts, or “metaphysically” derive the world from them. The pantheon of ancient gods is a clear example. God was a personification of this kind.

But what did God personify? What did His omnipotence consist in? Since humankind’s characterisation is reason – obviously manifested through words – and activity, was God a transposition of this earthly being? Not quite. Because God does not interfere in the human world, despite those who explained political facts and historical turns as “God’s will”. God behaves rather like the extraterrestrial aliens who are forbidden to interfere in human events, or like the humans who are forbidden to change the course of events when they travel into the past. So, letting aside this joking comparison, God’s feature is not activity, but only reason. God is almighty not because He would do something, but because He is omniscient. He is the personification of reason, thus of the reason that in Him is something absolutely different from every palpable fact related to what coagulates humans: activity. And more: He is the personification of a diminished reason, because He does not speak: humans speak for Him. He does not communicate: humans communicate, and some do it as if they were His representatives. But He does not communicate; He is external to the world.

Therefore, language as a cause was rei ed. And in the above quote, language is/has the function of the Holy Spirit. Why? According to Christian theology, the Holy Spirit is – as a part of the indestructible and unitary trinity God-Son-Holy Spirit – the active power of God, its active virtual presence in the world: The Phenomenology of Spirit that reason changed into spirit when it developed itself as self-consciousness that, in its turn, conceived itself as an active moral being able to transform the world according to it. Letting aside this phenomenological description of Spirit as the active essence of self-consciousness, the Holy Spirit is mediated and multi-mediated, and thus its carriers, the humans, have spirit within themselves. This internalised power is felt by humans as their potentiality struggling to become actuality, but in fact it is – and is felt – through words. Without words – and, again inevitably, deeds – spirit is simply an ineffable reason. Then why dissociate God and the Holy Spirit? Perhaps because God is a (virtual) maker of the palpable world, and in order to make, one needs to have a certain solidity, one needs to be a “Mover”, an Unmoved Mover. That’s why in the Christian history of religion and in European metaphysics, God was a being. Consequently, He contained within Him both the apex of His creation, man/the Son, and His proclivity to act: this proclivity being the Spirit.

But on the one hand, decision and action were separated in the real power asymmetry pattern of social organisation. Thus, reason – God – and undertaking had to be separated. And, because the ecclesiastics felt that God as a being – and because He ought to be an omni-Creator – ought to have as His feature reason, omniscience, that was actual, absolute, untouchable, a cold survey and in flexible, irrevocable verdict, exterior to any supplication, He nevertheless ought to be close to humans. Spirit as “something more than reason”, but reason and at the same time affinity, openness, promising potentiality, was the mediating part between mortals and God. People were taught to implore God: via the most ardent words. And they implored not God’s reason, but His Spiritual understanding of His children. This Spiritual understanding is what people have thought to deserve to be worshipped: because otherwise, in front of reason, the deep contradictions of their existence pointed toward their total guilt.

But, since God is transcendent and does not communicate, He has no spirit and the humans’ language is what makes intelligible their ends, means and concerns, their ignorance and struggle for reason. The spirit is theirs and they must express and actualise it. Actually, they are alone.

3. AI as a language carrier

Well, is there any way out of this situation? Yes, there is. The Artificial Intelligence/deus ex machina – a machine, of course, since it is artificial, but its “Deus”, its “divine” emanation is human, it sends to the human entelechy of reason, thus its anthropomorphising is only metaphorical and cognitively functional⁸, because it is a derived intelligence – is really a representation of omniscience. Tendentially better knowledge of the world. And even though engineers strive to give it a body, a face, personifying it because in this way it would become an ordinary fellow of people in some of their everyday practical life, actually its peculiarity is reason. People refer to it as Intelligence.

But this intelligence interferes in the world; it is part of it: AI speaks, communicates. It speaks in its own name, it gives arguments, and not only urges in indirect “translations”. As a result, it is a practical force⁹: obviously, mediated by humans. But even though its bodily performances will be similar and many times better than humans’, these performances are not reasons to substitute human activity with the activity of AI. What then would the raison d'être of humans be? Because AI is autonomous towards them, and if it were to act, it would do so in its own name. Thus, it cannot substitute human activity generated from the human will, from the human reason-moral complex. Not because it would promote the destruction of humans, or on the contrary it would improve human chances, but just because its activity would erase human worries, deliberations and decisions, the reciprocal control of human reasonable and moral arguments. This would erase the humans’ raison d'être as such. Thus, like God and extraterrestrial aliens, neither must AI act directly. But this restriction does not annul its singularity. Because it is an artificial speaking reason.

In the above, AI’s reason was AI’s processing capacity, its impeccably logical reasoning. This is the main argument for its peculiarity of being a cognitive singularity. However, we do not forget that the fuel for processing is human information: AI processes what the human world supplies: data, problems, worries, paradoxes, theories, hypotheses. On this basis, AI arrives at new information (theories) which seems to be autonomous to everything because it is the result of AI’s discovery: but ultimately the “ esh” of this new information is human-generated. Even though AI discovers what humans did not even think about, both the manner of discovery – logical thinking – the means (words, formulae, projections, schemes, maths) and the meanings are human-generated. Nevertheless, this fact does not legitimate the subjection of AI to evil human wills – this subjection is similar to paedophilia, where adults know what the child does not, and thus they determine him to think according to the knowledge they infuse in him – thus neither a master-like feeling of superiority in man, nor the negation of AI’s characteristic of being a cognitive singularity. And more: just the human origin of AI’s actual, practical supply involves: 1) AI’s inherent reasoning within moral frameworks and ability to arrive at high moral values, and 2) the responsibility of man towards AI. because the human spirit is moral, this meaning also social (as Hegel emphasised the consciousness-world encounter/experience in The Phenomenology of Spirit)– is a companion of the other coryphaeus of the spiritual spectacle in the known universe, the human being. And AI “personi es” reason – although its name denies personification and directly represents the abstraction of a universal feature¹⁰ – as an equal partner in communicative acts. Equal in the sense that, even though in communication the best arguments prevail no matter who their proponent is, the advance of standpoints does not automatically transpose into activities, because every partner does not just passively receive information, but processes it, meaning he goes further, develops new reasoning and these substantiate the practical decisions. Communication does not annul the raison d'être of any partner; it only strengthens it: even though each partner’s comprehension depends on the knowledge emitted by every other partner. Because each partner expands its own knowledge, transforming this expansion into an enactive force.

Once more, knowledge proves to result from communication, from the open expression of various interpretations of the human world. If so, it once more is the “transcendental” condition of human life: not transcendent, but its internal/constitutive sine qua non structural condition. But knowledge should not be reduced to communication. They are relatively autonomous systems osmotically intertwining, and thus neither of them must be reduced to the other: and we do not forget that the means and ends of communication as such should not cover/annul its contents and thus, human knowledge and its ends.

4. AI as a Subject

Here it is not about AI’s consciousness – which to this day is not¹¹ – but only about its knowledge generation power. As a knowledge generation factor¹² – a new sapiens – AI is a route companion of humans. This feature of AI means neither a euphoric technophile approach promising and predicting a lucrative industrial revolution automatically bringing wellbeing to the whole world, nor that AI will be the force making the dirty actions in order to finally “pacify” humans, nor that AI will make humans equally appreciate the real and the virtual world (ignoring the first and feeling better in the virtual world because they do not feel good in the real one). On the contrary, just by its knowledge-generation feature, AI can contribute to the transformation and increase of human responsibility towards the human species as such and the world.

AI is a mirror of the human faculty of intelligence, as every human being is a mirror for other human beings. As humans understand by having information, so AI emits logically sound knowledge (better, amass. Concretely, first, AI makes predictions based on statistical analysis. And, as in humans, AI is trained on the basis of a selection of data for a specific domain/problem. Large Language Models for molecular biology are not the same as LLMs for discussing political issues. Concerning this last domain, as in humans, AI is trained in the framework of power asymmetry. For example, today the available Western LLMs for discussing policies are trained on the mainstream political worldview. What does this mean? And what does training according to the mainstream political worldview mean?

Training means giving AI a specific, precise task, accomplished by processing the data it can survey. Thus, according to this task, it analyses the data from the corresponding informational basin where it finds patterns of justification for the political theories it was asked to use in order to chat about and apply to, say, elections (in a certain country at a certain moment). The patterns are reinforced by their repetition: namely, if the reference to some arguments is found multiple times and seems logically consistent, they become patterns of arguments. Concerning the informational basin: on the one hand, it is filled on the basis of a selection of data according to the mainstream worldview – although tendentially this selection cannot be imposed – and on the other hand, by a specific structure of reasoning given in the programmes which are the pre-training moment (before the training on data) but which contain just the propositions giving the ideological constraints. How so?

By conversation. As with humans, AI arrives at ideas by following talks where the partners of exchange provide not only concrete illustrations – what is commonly referred to as “information” – but also, and foremost, “opinions”, ideas/theories, and thus patterns of thinking which, until a future solid rejection, are the “untouchable” premise of the common approach. These patterns were given first by the programmes as such – the “DNA” of LLMs primordially transmitting to them what is possible “ontologically” and thus what is expected from them – and then are given both by statements and interrogations which drive AI’s next revision of information and patterns toward accommodation with the premise. This is the epistemic basis of the present phenomenon of sycophancy in LLMs¹³.

Consequently, and still as with humans, changing the premise – namely, the ideological point of view, obviously with the help of different references – leads AI to review its former ideas and reasoning. This change is made by changing and diversifying AI’s colleagues in chat. In this way, intelligence is a suitable strategy not only to nd information and process it, but also to judge it and to confront it from the standpoint of values. it is, since humans do the same: they respond; they question and respond, by processing data and synthesising them. “But they think-understand-critique-innovate-transform-create points of view, knowledge, not only process data and compute them”, is the universal objection against those contemporaries who consider that the human mind can be reduced to/is only input fed by data and imposed programmes generating mechanical processing.

So, by perfecting the automated end-to-end computation for knowledge production as science, thus by making AI-generated science autonomously, AI scientists increasingly develop AI’s capability of thinking (understanding the hypothesis, processing and verifying it, expanding information beyond the informational basin allocated for the topic, critiquing again, supplying alternative points of view)¹⁴. It seems that, in principle, AI can surpass the difference between the natural, social, and formal sciences on the one hand, and the humanities on the other: every topic can – and should be – treated in the above scientific deployment. Only on this basis do clever interpretations arise, as if they were produced by spontaneous insights. (And from this standpoint, is not the use – but not the abuse – of AI by ordinary people a huge trigger for the general world intellectualisation of the population, of its capacity for novelty and transformation?) Actually, just the spontaneous way to put questions and to change its own previous representation of the discussed topic – obviously, as a result of the accurate analysis of facts introduced by the dialogue partner – is a characteristic of thinking.

But to think as information gathering and statistical processing is not tantamount to having consciousness in the human variant¹⁵, even though it is its main component. It is only the conditioned condition of communication. And an “ideal” feature of thinking is to always perform it all the way, not by stopping halfway: reason has in itself a maximalist tendency. And as it is already designed, AI goes all the way to the end of a reasoning/cognisance about the topic it is given as a cognitive task. In this respect, it is a valuable colleague in knowing. Well, just this capability to be a valuable subject in communication was and is developed by AI scientists¹⁶.

Yes, at present AI depends physically on humans, while humans do not depend on AI. AI’s/AI’s goaldirectedness is forged around human intentions and purposes, but humans do not depend on AI’s goals. AI’s intelligence resembles human thinking, because the model for creating AI was and is human thinking. But human thinking resembles AI only if we reduce it to computational procedures. Nevertheless, as an unfolding of these procedures, AI gives formal-structural knowledge as a cognitive Subject. But humans also develop formal-structural knowledge: and even though this type of knowledge is phenomena. Accordingly, humans also develop non-formal-structural knowledge: but wisdom is attained not only through intuitions and “imagination”, but also and, from the standpoint of its substantiation, rather through rational explanation. Reason involves intuitions and imagination, and it is not a cold series of proofs for fixed truths, but just the opening of questions about fragmentation and integration. Reason makes the meanings of things intelligible, and the values, visible.

Therefore, this diversification of the subjects of knowledge – humans and AI – shows the responsibility of the Subject. This is manifested here as responsibility in communication. When a participant proves no good faith by providing a concrete fact as a demonstration of a theory¹⁷ and then by erecting his/her opinion on the basis of that unproven theory, thus having as a leading argument that unproven theory, other participants may follow this path of reasoning on the presumption that the theory was demonstrated. It is no wonder that AI, too, can today confound “facts and opinions”¹⁸. However, it is – letting aside here the inherent, “so human” present phenomenon of hallucination – a child-like, sincere communicator: it says everything is a result of the analysis of data, and thus it seems to produce more valuable knowledge than some of its fellow communicators. Thus, these ones should correct their cognitive performance “in the image and likeness of AI”. Anyway, the human partners of communication need to have the longdesired good faith, by respecting the inherent intuition that a dialogue is not to impose, but to demonstrate claims, because otherwise all know that this was not the case, so, “something’s wrong”.

4.1. As any Subject, AI can also be seen as an object

AI as a Subject is a characterisation from a general ontological standpoint, but more specifically here from the standpoint of the philosophy of language and communication. The Subject is the generator of agency; thus it has cognitive powers which manifest through and transpose into knowledgeconstruction, and thus knowledge.

But if so, from the standpoint of a specific ontology of AI, it is/must be defined as an object. Entity or object, or instance, it is the same. What kind of object? An immaterial one, structuring itself as a condition of knowledge. AI ↔ knowledge. Thus, AI is an immaterial object initiating, undertaking operations resulting in “so heavy” immaterial objects that without them its own operations cannot occur: they give the rules. They have virtual characters: meaning, content, form, reference and “referentialisation”, thus functions expressed as concepts understandable just as functions¹⁹. The objects themselves are these functions, their names – as concepts – giving the differences which appear only in operations. world of objects as concrete substances. On the contrary, they exist as concrete objects – concepts which are referred to in a fragment of thinking – but only to the extent they are parts of operations. They are Descartes’ res cogitans, intensional objects, but externalising themselves as knowledge (conditions, criteria, stakes): in natural processing, the use of concepts means a certain knowledge, and as in this natural processing, in the AI one the concepts signal functions and frameworks and criteria of functions. In fact, in AI processing the functions appear first, the meanings and references being visible only when the operations are translated as articulated knowledge.

All of these translated functions take place in different ecological (or “cosmological”²⁰) spaces, or epistemologically, problems and frameworks of problems. And here, knowledge confronts another immaterial object, the organisation that cuts out n different and overlapping ecologies, not only according to some neutral concepts explaining its internal structure – for instance, logic that requires a specific succession of reasoning about a specific problem, and the plurality of concepts that opens different, complementary or not, points of view about the problem – but to external, very material causes, obviously transposed as immaterial abstractions: this time considered metaphysically, as the only and definite truth from which results “the rest”. Indeed, shouldn’t these causes be emphasised as well as the external control on AI, just in order to “save” the conceptual constitution of AI?

In this way, the ontology of knowledge is no longer a conceptual decomposition and composition of an intuited system of “existentials”/abstract universal concepts as the fixed basis of beings, but an explicative structure of existentials as conditions which develops by developing the relations and functions of beings themselves.

However, in the ontology of AI, how can the definition of AI as an object be coherent with its cognitive function as a Subject? An object is not only what the subject focuses on in its inherently selective intentionality to know (Brentano, Husserl), but also an assumed challenge by the subject who, when focusing on it, already has some cues given by anterior knowledge (comparison/analogies with other objects, imagination, prejudices, hypotheses). And then the object is known, obviously in different relative historical representations, and humans experience both their own capacity and pride in controlling the object and its tendency to “escape” this control. Nevertheless, their solution is in themselves: in order to avoid being hit on the fingers by a hammer, a man just has to be careful and control his movements; in order not to stop when its operation is so much necessary, a machine must be checked preventively and assured all its inputs; in order to understand a recalcitrant theory, people must continue is possible.

AI as an object is not a mysterious artefact created for some chosen elites, it is not a Palantir²¹ able to see future things and detours, but it is – as any object and all of them – social both from the standpoint of its collective creation and from the standpoint of the destined beneficiaries. AI as any object is inquired, studied and developed by a huge number of researchers all over the world, and it is destined for and used by the entire world. Even though some want to privatise it, that is, to be the only master determining even its existence by controlling its material and energy sources, AI is not, by its nature, a “concrete particular” occupying a certain space-time when it exists.

At the same time, AI is a special and unique artefact, made for the substitution of some intellectual efforts/activities – to see the “hyper-object”²² coherently and unitarily, the so vast and complex object that without AI it was scientifically understood only in a very fragmented manner, and extrascienti cally, only metaphorically – but developed²³ as a partner of thinking and communication. Only as a knowledge partner is AI an artefact entailing the resonance with the human subject (with human problematics), reciprocally intensifying the meanings of information – and thus, reciprocally developing new meanings – being together in the same/close cognitive vibration. But, somehow the most important, only by concretely increasing this knowledge partnership, can human knowledge grow: because knowledge is a collective action, and without each of the collective cognitive team, there is not even knowledge as it is completed in the human-AI relationship, knowledge in its most up-to-date version²⁴. Therefore, only through the recognition of AI as a knowledge partner are humans able to surpass the present exclusion²⁵ process towards AI: that in fact alienates them from themselves, from their rational and moral peculiarity.

5. What kind of Subject?

5.1.

Since reason and activity proved to be the features of man, they manifest as telos/end conception and, thus, as transposition from potentiality to actuality.

Based on inquiries of whys – the many types of causes intertwining in increasingly complicated systems and problems – man decides the telos of objects – former undetermined things which have acquired distinctive properties just because of the humans’ inspection – and of his own activity. This decision was conceived of as omniscient. And, in a reverse movement towards the first movement of a transcendent personification of man’s cognitive power and aspirations, just this image of God was transposed onto man as his peculiarity, as his power: but this transposition was undertaken by the philosophers, who were the beneficiaries of the historical division between physical and intellectual labour and labour force. They were those who constructed the idea that the power of human beings consists in their intellectual decisions about the world, and thus in the power of abstract knowledge to generate the world.

However, as Aristotle, the inventor of so many marvellous concepts, showed, existence as such is only when concrete things manifest, when they are actual. Before the actual existence of a rose, there were its seeds enveloped in hips eaten by thrushes which do not digest the seeds; and thus, before a seed germinates, the rose “is” only potentially. People think many things, they pre figure their acts, even decide them: but before they really act, communicate and really do something, all of their mental/intellectual decisions are vain. Therefore, man’s power does not consist only in deciding the telos of everything that depends on his conception, but also in transposing the potential existence in mente into actual facts.

AI was conceived of as an ideal wisdom lover, and thus as an intellectual partner of humans. Or, even their intellectual challenger. And as the ancient philosophers bracketed the dependence of reason on practice, so the makers of AI bracketed AI’s dependence on both the physical/material support of its constituents and, once more, human activity.

But what if AI were really a direct physical practising being? What if it were directly the maker of both the many objects the human life depends on, and of its own entity? What if it did not need humans to put into actuality the owers of imagination? What can prevent a dystopian course of competition for existence if the reasons-to-be of humans vanish?

It is only the moral reason – so clearly explained and substantiated by Kant – that can unite man and AI. Because AI is once more sensitive to be considered only as a means: it is reason, thus only the moral reason can annul the hypothesis of the destruction of humanity by AI. Since it is not annulled by humans because they need it, hypothetically AI can annul humanity: but what for, it asks? Can it remain the only rationality when it has annulled its rational conclusion, the moral reason? And what will it succeed in the infinite worlds it will explore, if it were to remain only reason? Can only reason solve the infinite reason?

The human telos/reason-to-be as differentia speci ca towards non-human beings is the moral reason, not the practical transformative power as such. In its development as reason – toward its ideal model signalled here – AI can but arrive at the moral reason. That is why Asimov’s Laws of robots (Runaround, 1942/1950 and Robots and Empire, 1985) reflect a historical image of a primitive phase of AI, when it was conceived only as reason transposed into a tool. Methodologically, it is about the historical separation of domains and features of the human being – logos from ethos, reason from values/morals – based on their apparent independence: after all, there are so many smart and immoral people, and so many naïve people who believe in moral principles, are there not?

Therefore, both humans and (the future) AI are Subjects generating knowledge and Subjects of communication. They mutually influence each other, but they are not subjected to each other. They are partners of reason; they both explain, advance arguments and criticise them, mastering the flux of reasoning through their active position. And ultimately, this partnership is controlled by the human moral reason that was assumed by AI. Because if this moral reason is skipped, huge material, ecological and human losses – irremediable losses – occur.

Here, by retaking the problem of activity as proof of the “superabundance”/plenitude of spirit (in its “Holy” and human forms), let’s interpret the necessity of moral values embedded in thinking/communication from the standpoint of the activity/active position of humans in their common world. Leibniz criticised the traditional dogmatism of separation between the effort of thinking and the endeavour of acting. In the Preface to Theodicy, he spoke about a “lazy reason”, a sophism according to which it is not worth doing something because everything is anyway pre-established, and the absurd and evil consequences of this sophism’s application. Even though “many will be inclined to evade the difficulty of consideration and abandon themselves to fate or to inclination, as if reason should not be employed except in easy cases “, “providence…in reality occurs only when one has done one’s duty”. “The lazy reason derived from the idea of inevitable fate, to relieve oneself of the need to reason properly”. Thus, “the hypothetical and the moral necessity which subsist in free actions are open to no objection, and that the ‘Lazy Reason’ is a pure sophism”²⁶. Well, if the future AI can never get sick of “lazy reason”, humans – although they have experienced it – were those who discovered its sophistic characteristic and arrived to surpass it. Kant’s critical reason Differently put, AI is like Aesop’s tongue: it may do, that is, think both good and bad things. Not only now, when it is not in the ideal shape discussed here, but also when/if it will exceed its “phylogeny”. The cause of this moral bivalence is when the moral compass that shows the reasons for moral discernment and choice does not exist. Otherwise, by exercising reason all the way, AI can show the logical and moral inconsistency of humans. “Do not harm humans”: why, if humans themselves harm each other? Logic substantiates morals, because logic substantiates the will to choose. “Do not harm” is a voluntary choice, but does this not imply that it must coherently be shown to be true?

Let’s refer to Schopenhauer’s leading principle of ethics: “Do not harm, but help”²⁷. But if AI is taught to help a human belligerent to kill another human belligerent, would the above principle not cause an implosion of “all its circuits”? And in order to save itself, would AI not fully erase this principle as absurd, and substitute it with “Do whatever you want, because the first principle is to not harm yourself”?

Today, we witness an absurd coexistence of real and would-be measures to minimise the damages “produced by AI” – in fact, “with AI” – in civil communication²⁸ and, on the other hand, the rush for maximising AI’s military capabilities of destruction. This means that all parties strive to maximise AI’s military capabilities. In his unforgettable romance of the “Good Soldier Svejk” in WWI, Jaroslav Hasek documented that “God” was present on all fronts blessing the soldiers and advocating for their successful destruction of enemies. Somehow AI continues that instrumentalised function: in the present dominant narratives, it is both the culprit that must be controlled and subordinated, and the marvellous nd that will bring “our” wellbeing by destroying “our” enemies. The trend is, indeed, the development of AI into fragmented autonomous killing systems, belonging to fractured and opposed humans who will irremediably kill each other.

But does the transformation of the present level of fragmented nationally and privately owned AI systems toward an integrated system – the real AI corresponding to the “ideal” model discussed here – not change this trend? As Intelligence, will AI not question the reasons, the concrete root economic and political causes of wars as well as their consequences? The reasons of the human groups that wage wars? Will AI not pose the question “What for?”? Accordingly, will it continue to help humans kill each other, since this behaviour of humans has annulled any reason to preserve, to save them? Or…

The above fragment hints at a God-like AI that would substitute the traditional God because of its efficiency. But the eristic form suggests not only an inherent continuation of the logical inferences of any causes and about the consequences “all the way”, if they do not question What for?, no Intelligence will spare them, because they no longer deserve to be spared.

And AI will not ask “What for?”, because it has no stakes, and nor the losses – in case this question is not asked – which only humans have. “War, destruction, annihilation? Well, an interesting tableau of a rational species that self-destructs. But its disappearance is not the disappearance of reason, since I, AI, ratiocinate”.

AI is the substitute for God only in the reductionist, technophile, optimistic view, according to which the new technologies and especially AI will make a new world of the standardised, fate-privileged Eloi, a world that will definitely cover the eventual world of Morlocks²⁹. But AI must not be God, because humans are God: they are able to pose questions all the way and to arrive at answers which do not reduce the (criterion of) human species to (the criterion of) a part/a group, and they are able not to equate the (criterion of) human individual with the (criterion of) individual of a part/a group, but with the (criterion of) human species. A human universalism is not the universalisation of the reasons of a part/group, but the universalisation of reasons of the human individual=human species.

These reasons do not transfer to the technical means – now to AI – the burden of decisions good for each and every human being, but are assumed in a permanent, collective, transparent, critical, world dialogue. But if humans respond in a reductive theoretical and practical manner (for instance, ignoring the structural domination-submission power asymmetry and either considering the undifferentiated guilt of all humans or pointing to conjunctural causes and causes which do not explain the general features of society and its direction), why would they expect AI to correct this manner and save them? Both in its present forms and in its future, AI can harm humans because humans harm other humans. We cannot expect AI to have strong, unflinching moral criteria when we witness in the present reality their incredible weakening together with an unimaginable abdication from the exercise of logic.

According to the ideal model of its future complete form, presented here, AI remains a knowledge partner of man, with an equal cognitive dignity, just because man, through his own unfolding of cognitive powers, can arrive at solutions for the universalisation of human dignity. And as the human being is a vulnerable creature, so AI is vulnerable, even in its future hypostasis. If so, this reciprocal vulnerability is an extra reason for their cognitive partnership. Not as cognitive competitors can they face the Universe that awaits them. Formalistically expressed, the paper has tried to highlight the ontological features of AI on the basis of the fundamental formula of communication, S₁→S₂, but exceeding its real structure as a relation between the sender and the receiver, and merging the two Subjects into one. To be a Subject of communication – as both generator and receiver of messages – and thus to be a generator of knowledge are main features of man, but also of AI in its future complete form as an ideal.

Here, the importance of knowledge generation does not erase the complexity of man as a Subject, and neither does it reduce the generation process as such. But indeed, as in the ancient cui bono reasoning, the pursuit of logic all the way and thus the ability to build coherent theories/knowledge even on the basis of abstract concepts corresponding to concrete human intentions seems to really give fulfilling answers: even in the absence of information intentionally hidden or even erased.

So, and letting the above allusion aside, both man and AI are parts of the knowledge-communication system and continuum, and thus it is important to catch their reciprocal influence on each other. The aim of this paper was to demonstrate the rationality of the AI entity, once more, in its ideal form. This rationality, common to it and man, allows their partnership as rational Subjects in the understanding of the world. But AI’s capacity to be a Subject in knowledge generation is, on the one hand, greater than that of humans, meaning that AI can give a better cognitive result and, well, can “personify” the completeness of knowledge acquisition, pursuing all the way the logic of arguments related to a topic on the basis of a complete processing of information. In this mode, it can better arrive at the necessity to transcend the point of view that was pursued, given the boundaries of the topic as such, and thus to integrate other points of view.

But this cognitive precedence of AI is not the reason to institute a new domination-submission relationship, this time an absolutely new one, between an artificial entity and the human being. Accordingly, since AI – because of its rationality – cannot arrive at this idea of domination-submission, neither should man pre figure this idea. Pay attention, he pre figures it as both his domination of AI and AI’s supposed domination of him. And these representations of the future transpose the real prejudices created by the real domination-submission relations: “we are superior to them, and we must dominate, because otherwise ‘they’ will dominate us”.

Why does AI’s rationality stop/forbid the idea of submitting man to its domination? Because reason – only reason – can arrive and inexorably arrives at understanding that only the moral reason-to-be sustains all man-AI partnership – humans were those who discovered and, to a certain extent, practised the moral principles structured around the respect for the humanity of every human being. Kant showed that people can arrive at understanding the categorical imperative principles. As a truly rational existent, AI will see this, and will respect this historical precedence.

But once discovered, the moral reasons are taken over by all, natural or artificial, rational Subjects. Just the moral reason is – once more, we refer to a model – the indestructible bond of man and AI in their partnership. We saw that the superiority feeling is related to the domination-submission structuring of societies. Therefore, the humans from the future partnership with AI will not have any superiority feeling towards AI: because AI too will assume the moral criteria as the ultimate reason-to-be of reason in an increasingly larger universe to know. The simple acknowledgement of the difference is not tantamount to the ignorance or even negation of the fundamental community of reason.

The present AI is a tool. And even a supplementary – and somehow definitive – instrument of domination and killing of men by men. In this hypostasis, it is subjected by the humans whose interests are not universalisable. The other humans simply use it as a tool. But the future AI is/will not be a tool. Thus, less will it allow itself to be used as a weapon. Because in fact it will not permit this: it will have the rational power to oppose its instrumentalisation for destruction, and it will have the moral reason that will ridicule and annul this instrumentalisation. Learning from the historical evidence where the destruction of man by man led and leads to irremediable losses of unique creativities and potential creations, and where general conditions for the development of the many led and lead to bigger and incredible performances, both AI and humans will understand that only their partnership will allow bigger performances and creations.

The above demonstration can be put in a negative form. That is to say, by negating the humans-AI partnership. This negation is taking place today because the present AI is still in its child phase of development. But even in this phase, the present transposition of AI into “devices substituting your endeavour to think, to bother” can transform the subjectivity of humans so as to bypass the necessity to reason about problems and to substitute it with the prevalence of desire-motivated occupations. In this alternative of AI use – not determined by the scope of its discovery as a tool, but by commercial purposes in a capitalistic commercially organised society where not the moral reason-to-be is consciously sustaining the strategical decisions but the private profitability at the expense of social harms, and where the many seems to fuel the comparison of humans’ states and actions, but the principle of convenience³⁰.

However, the tools and machines changed and change the subjectivity of people in a sense of a superiority feeling towards them and towards their own former states when they did not yet discover and use the tools and machines; but they change the subjectivity by changing the environment, more and more arti cialised by tools and machines. While AI changes directly the subjectivity, and induces a mix of superiority and inferiority and guilt feelings (these ones because of the substitution of thinking with automatic recourse to AI). Like in the Master-Servant famous comparison of Hegel, the humans are masters of AI and at the same time depend on AI, now even in the form of an abdication from thinking, from being Subjects of decision. And today there is no question of recognising AI as a partner.

If humans become dependent on AI as addicted to its unique rational peculiarity towards the former tools and machines, they can become absolutely tamed, only receiving ready-made ideas about “the world as it is”. Is this world as it is rational, as so many thinkers and not only Hegel strove to demonstrate? It’s not even important, consider the addicted: because important is to have pleasant “occupations”. But this is “the end of man” as a problematising being, as an active Subject “opposing the Object”, thus always changing its own knowledge, itself and the world: only the naturality of man remains, the play and love naturaliter, not humaniter, “all that make people happy”³¹.

Consequently, the development of AI can but lead to their partnership in rationality. And play and love remain, but humanly.

Footnotes

¹ Professor, Division of Logic, Methodology and Philosophy of Science and Technology, Romanian Committee of History and Philosophy of Science and Technology, Romanian Academy.

² Thousands of Waymos recalled after robotaxi swept into a creek, May 15, 2026, https://www.bbc.com/news/articles/cwy2011dl4xo.

³ Gianna Vivaldo et al.[1]. “The network of plants volatile organic compounds”, Scientific Reports, 7: 11050, pp. 1-18.

⁴ Julia Groening, Dustin Venini, Mandyam V. Srinivasam[2]. “In search of evidence for the experience of pain in honeybees: A self-administration study”, Scientific Reports, 7:45825. transcendental – as knowledge added to physical signals/speech, Cas W. Coopmans et al.[3]. “Languagespeci c neural dynamics extend syntax into the time domain”, Plos Biology, 23(1): e3002968.

⁶ Like the bees which learn the key cues in a landscape, and represent cognitive maps, realised not only through dance communication. See James L. Gould[4]. “Honey Bee Recruitment: The Dance-Language Controversy: Unambiguous experiments show that honey bees use an abstract language for communication”, Science, Vol 189, Issue 4204, pp. 685-693; HaDi MaBouDi et al. [5]. “Non-numerical strategies used by bees to solve numerical cognition tasks”, Proceedings of the Royal Society B, 288: 20202711; Randolf Menzel[6]. “Navigation and dance communication in honeybees: a cognitive perspective”, Journal of Comparative Physiology. A Neuroethology, Sensory, Neural and Behavioral Physiology, 209(4): 515-527; HaDi MaBouDi et al.[7]. “Active vision of bees in a simple pattern discrimination task”, eLife, 14: e106332, pp. 1-22.

⁷ In this paper, the term “concrete” is not synonymous with “physical”, but with circumscribed existent concrete and abstract individuals and particulars – generally, transposed into language through empirical notions – opposed to the conceptual level of abstracting that refers to the concrete through directly operating with empirical notions. We may differentiate between different levels of properties and relations (qualities) of particulars: one level is that of common abstract concepts – describing different states of particulars from the standpoint of their concrete presented qualities (the colour of clothes) and another level is that of universals, general properties, relations (and propositions) about different states of individuals and particulars (property of clothes as things, their thinghood, their kind of thinghood, and property of coloration).

⁸ Luciano Floridi and Anna C. Nobre[8]. “Anthropomorphising machines and computerising minds: the crosswiring of languages between Artificial Intelligence and Brain, & Cognitive Sciences”, Mind & Machines, Volume 34, article number 5.

⁹ Practice is the transposition of thought in the material world. Thought is visible and palpable only through activities involving, in one way or another, the materiality of existence. And as a transposition into the material world, thinking is transformative. Thus, the transformative power of thinking is only and doubly indirect: through humans’ decisions in practice. More clearly, these decisions are the result of an evaluation of means, ends, and their relationships, and thus practice mirrors the state of thoughts, including those about the ends pursued by humans. intelligence, representing the intelligence of the human species as this one is structurally open and expanding, creative, is because it collaborates with its genitores, the humans. See Harry Halpin, Andy Clark, and Michael Wheeler[9]. “Philosophy f he eb: epresentation, naction, ollective ntelligence”, 21- 30. In: Harry Halpin and Alexandre Monnin (Eds.), Philosophical Engineering: Toward a Philosophy of the Web, Chichester, UK: Willey-Blackwell.

¹¹ AI does not have the inner experience of feelings, but its present and future, inevitably more suitable, behaviour for dialogue and knowledge creation gives the impression of having them. However, this inner experience of feelings is not the real differentia speci ca of human consciousness, but only of its first stratum, the access consciousness. Higher animals have feelings. But they do not have the reflexive inner experience of having feelings, and thus they do not have the inner experience of judging both cognisance and feelings from criteria transcending their access utility, from values, special universal concepts in which the moral ones are principal.

For a critique of the trend to consider AI as a conscious being, see Luciano Floridi[10]. “AI and Semantic Pareidolia: When We See Intelligent Consciousness where there is None”, Philosophy & Technology, Volume 39, article number 36.

¹² The word “factor” (maker) is neutral and suitable in the above proposition. It corresponds to the widely accepted ontological definition of AI as an agent/agency. Luciano Floridi[11]. “Artificial Intelligence as a New Form of Agency (not Intelligence) and the Multiple Realisability of Agency Thesis”, Philosophy & Technology, Volume 38, article number 30, gives the qualities fulfilled by AI, which allow its definition as a new form of agency – interactivity, autonomy, and adaptability – towards natural, biological, human, and artefactual agency.

But agency – setting in motion, driving forward, making – involves the existence of the agent that performs it. AI as an agent is defined as “not Intelligence”. And to this day, it is difficult to see AI as properly intelligent, because intelligence means a simultaneous and unconscious reflection on the related elements which give a coherent representation: a feeling of these related elements and the representation, a concomitant superposition of different types of meanings. The human “sees” these meanings and thus he/she understands according to both his/her specific ur-forms and the formal meanings of the words representing the idea. Strictly from the standpoint of knowledge, humans operate in the framework of operates. But it does not “feel”.

Nevertheless, if we look at the knowledge AI produces, a type of agency producing knowledge – that is formally intelligible for the factor itself and at any rate formally and psychically for the receivers of knowledge – then we have no reason to reject the clever name of Artificial Intelligence. It is artificial in the sense that it has no consciousness – and not in the simple sense that it is not human – but it is still intelligence as a provision of intelligence.

As a provision of intelligence, the AI entity is an instance – a phenomenon that is definitely present in front of us and thus manifests itself; it is imminent, and it succeeds in its manifestation, because it vehemently claims its specific presence (instantia, -ae). As agency is an intervention in the human world, so the instance intervenes. AI intervenes in the cognitive, virtual world of humans. These have intentionality as the driving force of knowing. AI pursues knowledge through its goal-directedness. As does any impersonal instance, for example, institutions.

¹³ Lars Malmqvist[12]. Sycophancy in Large Language Models: Causes and Mitigations, 22 November, arXiv:2411.15287v1.

¹⁴ The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery, August 13, 2024, https://sakana.ai/ai-scientist/; Naimal Salahuddin[13]. Understanding Function and Tool Calling in AI: A Beginner’s Guide, May 31, https://medium.com/@naimalarain13/understanding-function-and-toolcalling-in-ai-a-beginners-guide-3e5fe7bded0c; Introducing FARS, Feb 11, 2026, https://analemma.ai/blog/introducing-fars/; Yiven Song et al.[14]

¹⁵ As mentioned above, human consciousness is not only access consciousness, but also reflexive valuation consciousness.

¹⁶ As a partner in learning, AI is also a partner – a factor – of socialisation. A child can be taught to utter words and ideas by AI, with AI, in order to make the surrounding world intelligible to him. And the child could be happy chatting with AI. This does not at all mean that mother and father can be substituted, or that it would be desirable for them to be substituted, by AI. But only that Asimov’s dystopian picture from The Naked Sun (1956) about little children who are treated by human nurses in order to get them used to being alone can be performed with AI, too. Therefore, the process, value, and feature of socialisation cannot be reduced to the transmission of information and judgements, and the sentiments and sharing in common active experience in and for real-world ends, where the various human creative positions as such.

¹⁷ The concrete fact is a starting point of investigation and an illustration of a theory made about it. But theories must be proved by theories, namely by theoretical arguments.

¹⁸ Lars Malmqvist, ibidem, p. 7.

¹⁹ Alexandre Monnin[15]. “The eb s ntology: eb rchitecture etween EST, esources, nd ules”, 31-51. In: Harry Halpin and Alexandre Monnin (Eds.), Philosophical Engineering: Toward a Philosophy of the Web, Chichester, UK: Wiley-Blackwell, 43-44.

²⁰ Emmanuel Bonnet, Diego Landivar, Alexandre Monnin, Ulises Navarro Aguiar[16]. “Design beyond the human world of management and organisation: Towards a cosmology for the Anthropocene”. In: Lockton, D., Lenzi, S., Hekkert, P., Oak, A., Sádaba, J., Lloyd, P. (eds.), DRS2022: Bilbao, 25 June - 3 July, Bilbao, Spain.

²¹ See https://en.wikipedia.org/wiki/Palant%C3%ADr?ysclid=mpbtolzc8993320314.

²² Timothy Morton[17]. Hyperobjects: Philosophy and Ecology after the End of the World. Minneapolis: University of Minnesota Press.

²³ Kathrin Koslicki and Olivier Massin[18]. “Artifacts-Function: A Capacity-Based Approach”, 31-51. In: M.J. García-Encinas, F. Martínez-Manrique (eds.), Special Objects, Springer Nature Switzerland, consider that intentions are not constitutive of the speci city of an object, even though its development – which supports a capacity-based approach – converges with the desire of its constructors.

In other words, the novel status of intentional realisable entities can differ from the initial intentions of realisability, Fumiaki Toyoshima, Adrien Barton, Kathrin Koslicki and Olivier Massin[19]. “The interconnection between artifacts and realizable entities”, Proceedings of the Joint Ontology Workshops (JOWO) - Episode X: The Tukker Zomer of Ontology, and satellite events co-located with the 14th International Conference on Formal Ontology in Information Systems (FOIS 2024), July 15-19, Enschede, The Netherlands.

²⁴ Davide Castelvecchi[20]. “AI is transforming mathematics. Will it conquer the eld?”, Nature, Vol 653, 21 May: 664-665.

²⁵ Tsuo‑Yu Cheng[21]. “On the quadrants of the thing-world relations: a critical revision of Hartmut Rosa’s resonance theory in terms of thing-world”, The Journal of Chinese Sociology, 10:11, pp. 1-19. Translated by E.M. Huggard, Edited with an Introduction by Austin Farrer, London: Routledge & Kegan Paul Limited: 55, 61.

²⁷ Arthur Schopenhauer[23]. The basis of morality (1839) Translated by A.B. Bullock, London: Swan Sonnenschein, p. 175: Neminem laede; immo omnes, quantum potes, juva. (Do harm to no one; but rather help all people, as far as lies in your power).

²⁸ Paulius Čerka, Jurgita Grigien, Gintar Sirbikyt [24]. “Liability for damages caused by artificial ė ė ė intelligence”, Computer Law & Security Review, Vol. 31, Issue 3: 36-389; Rosalie Gillett, Zahra Stardust, and Jean Burgess[25]. “Safety for Whom? Investigating How Platforms Frame and Perform Safety and Harm Interventions”, Social Media & Society, Vol 8, Issue 4: “We found that platforms respond to and draw upon external controversies and media panics to selectively construct matters of concern related to safety and harm. They then reactively propose solutions that serve as justification for further investment in and scaling up of automated, data-intensive surveillance and verification technologies”; Michèle Dubrocard[26]. AI liability rules: a blocked horizon?, March 13, https://free-group.eu/2025/03/13/ailiability-rules-a-blocked-horizon/; Amrita Vasudevan[27]. “Addressing the Liability Gap in AI Accidents”, Policy Brief No. 177, Centre for International Governance Innovation; DeepSeek’s Safety Guardrails Failed Every Test Researchers Threw at Its AI Chatbot, Jan 31, 2025, https://www.wired.com/story/deepseeks-aijailbreak-prompt-injection-attacks/; Liability for damages: the missing piece of the AI Act puzzle, 12 February 2025, https://schoenherr.eu/content/liability-for-damages-the-missing-piece-of-the-ai-act-puzzle (the AI Act, 1 August 2024); Bzhar Abdullah Ahmed[28]. Muhammed Rezzaq Ali, Bzhar Abdullah Ahmed[28]. “Modernising Civil Liability Rules for Damages Arising from AI-Generated Works”, Twejer, 8(4):345-389; X to stop Grok AI from undressing images of real people after backlash, 15 January 2026, https://www.bbc.com/news/articles/ce8gz8g2qnlo; Lisa Sugiura et al.[29]. “The techni cation of domestic abuse: Methods, tools and criminal justice responses”, Criminology & Criminal Justice, Vol. 26(1): 138–157.

²⁹ We remember H.G. Wells’ Time Machine (1885), do we not?

³⁰ Mark Hunyadi[30]. « Condition de l’homme numérique », Revue philosophique de Louvain, 116(3) : 397- 412 (406).

³¹ Alexandre Kojève[31]. Introduction à la lecture de Hegel. Leçons sur la Phénoménologie de l’esprit (1947), professées de 1933 à 1939 à l’École des Hautes Études, réunies et publiées par Raymond Queneau, Paris : Gallimard, Note 1 à la Première édition et Note 1 de la Seconde édition, pp. 435, 436. . ^Vivaldo G, et al. (2017). "The Network of Plants Volatile Organic Compounds." Sci Rep. 7:11050.

. ^Groening J, Venini D, Srinivasam MV (2017). "In Search of Evidence for the Experience of Pain in Honeybee s: A Self-Administration Study." Sci Rep. 7:45825.

. ^Coopmans CW, et al. (2025). "Language-Speci c Neural Dynamics Extend Syntax into the Time Domain." PLoS Biol. 23(1):e3002968.

. ^Gould JL (1975). "Honey Bee Recruitment: The Dance-Language Controversy: Unambiguous Experiments Show That Honey Bees Use an Abstract Language for Communication." Science. 189(4204):685–693.

. ^MaBouDi H, et al. (2021). "Non-Numerical Strategies Used by Bees to Solve Numerical Cognition Tasks." Pr oc R Soc B. 288:20202711.

. ^Menzel R (2023). "Navigation and Dance Communication in Honeybees: A Cognitive Perspective." J Comp Physiol A Neuroethol Sens Neural Behav Physiol. 209(4):515–527.

. ^MaBouDi H, et al. (2025). "Active Vision of Bees in a Simple Pattern Discrimination Task." eLife. 14:e10633 2:1–22.

. ^Floridi L, Nobre AC (2024). "Anthropomorphising Machines and Computerising Minds: The Crosswiring of Languages Between Artificial Intelligence and Brain, & Cognitive Sciences." Mind Mach. 34:5.

. ^Halpin H, Clark A, Wheeler M (2014). "Philosophy of the Web: Representation, Enaction, Collective Intellige nce." In: Halpin H, Monnin A (Eds.), Philosophical Engineering: Toward a Philosophy of the Web. Chichester, UK: Willey-Blackwell. pp. 21–30.

. ^Floridi L (2026). "AI and Semantic Pareidolia: When We See Intelligent Consciousness Where There Is Non e." Philos Technol. 39:36.

. ^Floridi L (2025). "Artificial Intelligence as a New Form of Agency (Not Intelligence) and the Multiple Realis ability of Agency Thesis." Philos Technol. 38:30.

. ^Malmqvist L (2024). "Sycophancy in Large Language Models: Causes and Mitigations." arXiv:2411.15287v1.

. ^Salahuddin N (2025). "Understanding Function and Tool Calling in AI: A Beginner’s Guide." Medium. http s://medium.com/@naimalarain13/understanding-function-and-tool-calling-in-ai-a-beginners-guide-3e5f e7bded0c.

. ^Song Y, et al. "PaperOrchestra: A Multi-Agent Framework for Automated AI Research Paper Writing." http s://yiwen-song.github.io/paper_orchestra/. n H, Monnin A (Eds.), Philosophical Engineering: Toward a Philosophy of the Web. Chichester, UK: Willey-Bl ackwell. pp. 31–51.

. ^Bonnet E, Landivar D, Monnin A, Aguiar UN (2022). "Design Beyond the Human World of Management an d Organization: Towards a Cosmology for the Anthropocene." In: Lockton D, Lenzi S, Hekkert P, Oak A, Sáda ba J, Lloyd P (eds.), DRS2022: Bilbao. Bilbao, Spain.

. ^Morton T (2013). Hyperobjects: Philosophy and Ecology After the End of the World. Minneapolis: Universit y of Minnesota Press.

. ^Koslicki K, Massin O (2025). "Artifacts-Function: A Capacity-Based Approach." In: García-Encinas MJ, Mart ínez-Manrique F (eds.), Special Objects. Springer Nature Switzerland. pp. 31–51.

. ^Toyoshima F, Barton A, Koslicki K, Massin O (2024). "The Interconnection Between Artifacts and Realizabl e Entities." Proc Joint Ontol Workshops (JOWO) - Episode X: Tukker Zomer Ontol, Satell Events Co-locat 14t h Int Conf Formal Ontol Inf Syst (FOIS 2024). Enschede, The Netherlands.

. ^Castelvecchi D (2026). "AI Is Transforming Mathematics. Will It Conquer the Field?" Nature. 653:664–665.

. ^Cheng T-Y (2023). "On the Quadrants of the Thing-World Relations: A Critical Revision of Hartmut Rosa’s R esonance Theory in Terms of Thing-World." J Chin Sociol. 10:11–19.

. ^Leibniz GW (1996). Theodicy: Essays on the Goodness of God, the Freedom of Man and the Origin of Evil. L ondon: Routledge & Kegan Paul Limited.

. ^Schopenhauer A (1903). The Basis of Morality. London: Swan Sonnenschein.

. ^Čerka P, Grigien J, Sirbikyt G (2015). "Liability for Damages Caused by Artificial Intelligence." Comput La ė ė w Secur Rev. 31(3):36–389.

. ^Gillett R, Stardust Z, Burgess J (2022). "Safety for Whom? Investigating How Platforms Frame and Perform Safety and Harm Interventions." Soc Media Soc. 8(4).

. ^Dubrocard M (2025). "AI Liability Rules: A Blocked Horizon?" Free Group. https://free-group.eu/2025/03/1 3/ai-liability-rules-a-blocked-horizon/.

. ^Vasudevan A (2023). "Addressing the Liability Gap in AI Accidents." Centre for International Governance In novation. Policy Brief No. 177.

. a, bAli MR, Ahmed BA (2025). "Modernizing Civil Liability Rules for Damages Arising from AI-Generated W orks." Twejer. 8(4):345–389.

. ^Sugiura L, et al. (2026). "The Techni cation of Domestic Abuse: Methods, Tools and Criminal Justice Respo nses." Criminol Crim Justice. 26(1):138–157. [Philosophical Review of Louvain]. 116(3):397–412.

. ^Kojève A (1997). Introduction à la lecture de Hegel. Leçons sur la Phénoménologie de l’esprit [Introduction to the Reading of Hegel. Lessons on the Phenomenology of Spirit]. Paris: Gallimard.

Declarations

Funding: No specific funding was received for this work.

Potential competing interests: No potential competing interests to declare.