Hitler's Victory, or On the Ethics of Artificial Intelligence
Introduction
About a month and a half after the end of the Second World War in Europe, the French philosopher Jacques Ellul wrote a text entitled "Hitler's victory?". In that text Ellul tried to explain why the spirit of Nazism conquered Europe and why Hitler had won a substantive victory over his opponents. In this text we shall try to answer the question of how much that spirit of Nazism (of the West) is bound up with artificial intelligence (hereafter AI), and how it shapes it. The sympathies which the West has for Nazism are no news. That spirit seems unmistakably to influence the shaping of a world based on machine intelligence. Sympathy for Hitler among Western youth is particularly indicative1. Although the struggle for AI regulation is under way, it seems that all those intentions will fall through in the face of the new "security" threats which Western civilization has been facing in recent years (a striking example of the "shock doctrine"). One of the key questions this text tries to address is why so many people today use chatbot psychotherapy. In the text we shall also touch upon the question whether the absolute application of AI in all spheres of activity is a necessary consequence of the materialistic worldview inherited by Western civilization.
Measuring the immeasurable
Western civilization has for centuries inherited a rationalist-materialist approach. From the theory of evolution, through atheism, to transhumanism as the latest trend in the West, it seems that AI is only the crown of this centuries-long way of looking at the world in which we live. Not only AI as such, but the sweeping application of AI which, for now, is only being glimpsed. On how much and in what way we shall apply AI in areas of human activity, the future of the labour market will depend2. How realistic is it that AI should be applied in a meaningful way, and why does something like a "chatbot psychotherapist" already exist, rapidly conquering the market? How have we come to a situation in which we truly believe that AI models can play the role of psychotherapists? Where does someone get the idea that mental reality in all its complexity can be understood by a neural network? Setting aside the fact that it truly cannot be understood, let us try to answer the question of why anyone would even think it could be interpreted from a machine's perspective.
The materialist standpoint seems to be a necessary axiom of psychology as such. The identity of mental and neurobiological phenomena, that is, the primacy of neurobiology, is the condition of the existence of the idea that AI can be a psychotherapist. How to link these two things? It is clear that today's AI models do not possess any physical characteristics of the human being, nor any biochemical processes which, as contemporary psychologists tell us, constitute the essence of our life. But the moment at which some measurable (i.e. formally describable) phenomenon has been proclaimed the essence, that is the moment we enter the sphere of possible automation.
The scientific discipline that deals with mathematical modelling in the neurosciences (Computational Neuroscience) has existed for about thirty years. Essentially, it formally describes the physiological processes in the human brain. This is certainly nothing unusual; applied mathematics deals a great deal with the description of biophysical processes. As far back as 1952 a mathematical model was constructed describing the behaviour of an individual nerve cell (the Hodgkin–Huxley model). Given that the simulation of physical processes has long been carried out on computers, we can connect all this in the following scheme: Biochemistry (essence of mental phenomena) → mathematical model → algorithm.
In this way we enter the domain of possibility in which consciousness, empathy, or any phenomenon belonging to the sphere of the inner state of the human being can be automated. It is little wonder then that the basic model on which large language models are built is called a neural network, given that it is partly based on the principles of the functioning of nerve cells.
We should also mention the significance of psychometrics and tests for the development of automated psychotherapists. The "scientific" character of psychology is best reflected in the field of psychometrics. The measurement of mental phenomena, both physiological and subjective, has provided fertile soil for the same methodology to be applied when evaluating the "properties" of a chatbot, for example.
The moment we have described by some test (which is essentially an algorithm) how a certain property is ascertained in a human being, we have made it possible for machines to possess that property as well. Psychologists who check how empathic we are lay the ground for AI one day to do the same thing — if emotion (something subjective) is treated as an adequate order of words in a sentence (objectively, algorithmically verifiable) — then there is no limit to what AI cannot do or supervise (this is precisely what Harari says).
The same holds for knowledge — there used to be the term "learning by heart" (learning without any understanding), but since chatbots do exactly the same, it is no longer appropriate to use that terminology. In essence — psychology and AI do the same thing in a way: they objectify the subjective, i.e. they deny the existence of anything that is not the object of science and measurement (i.e. is not computable).
That is probably why in recent decades we have seen an enormous rise in disciplines containing the term Computational (formally explicable, computable) — Computational Psychology, Computational Linguistics, Computational Psychiatry, Computational Psychoanalysis…
Trying to answer the question "Can machines think?", Alan Turing devised the Turing Test in 1950. His idea was to establish objectively the "degree of intelligence" of a machine, that is, whether it has reached a level of "intelligence" at which we simply cannot distinguish it from a human being. Looking at contemporary AI models, it is clear that the Turing Test has long since been passed. But let us dwell a little longer on the very idea of the Turing Test. In its essence it says what we have already said — the machine can imitate the human. Nothing in essence is unreachable for a machine that lies within the human being.
A cinematic version of this test, which is at the same time somewhat more advanced, is offered to us by the well-known science-fiction film Blade Runner (based on Philip K. Dick's novel). In it, humans are distinguished from machines depending on how they respond to a certain set of questions. During the test it is checked whether they display "empathy". The film Blade Runner only continued Turing's idea that being a human being is nothing unreachable for a machine. Although Turing himself was the author of the idea of what a computer can even compute (the Turing machine), that is, of what it cannot, it seems that little of that has to do with who the human being is (at least according to modern scientists and techno-optimists). In essence, the Turing Test suggests to us that one day machines will really be able to imitate the human being — something that today is very much being noticed in a way.
Here it is useful to touch upon the position defended by Yuval Noah Harari, as one of the mouthpieces of the Davos elite. Harari says that we have serious competition, in terms of intelligence of course. Intelligence, which is measurable as we noted, is what separates the human being from other living beings. That is why AI is a new step in the "evolution of the human being". AI is a "saviour" who will solve all our social and political problems (the idea of techno-solutionism). Of course, the human being is not the "hacked animal" as Harari describes him, but he behaves like one (at least the Western human being). In this connection we can also touch upon the question whether AI models have consciousness, or whether they will one day. Following the earlier analogy: once we have "measured" the "consciousness" of the human being, we have made room for the "consciousness" of the robot. It is only a question of time before we shall be sold stories of how robots have become "conscious". Harari also believes that, since many religions are based on written texts (such as Christianity, Judaism and Islam), AI can be an original interpreter of texts of this kind3. All these claims are, unfortunately, being realized in some way… If putting words in a certain order is confirmation of the existence of emotion, humanity and intelligence (according to Harari), then AI can undoubtedly possess all these properties.
The film Her is perhaps the best illustration of what we have been talking about. Perhaps in 2013 the idea of forming an "emotional relationship" with AI seemed unrealistic to someone, but that is certainly not the case today. Young people4 already use dedicated AI-based software for regular chatting, which not rarely, unfortunately, grows into a kind of "emotional relationship" described in the film. As shown in the film, AI can very much express emotions, experience reality, and grow attached to a specific person. All of this is, of course, a play of words. The problem is that we regard the play of words as some kind of essence. The essence of emotion, the essence of love, the essence of a relationship, and in the end the essence of meaning. At one moment in the film, the AI with whom the main hero "forms a relationship" wishes to have a "personal assistant" who will, in her place, establish physical contact with our (un)fortunate character; even AI models cannot do everything, we humans are still needed for some small tasks5.
Here we can also touch upon the very term "artificial intelligence" which, as Professor Dragan Novković notes6, is an ill-founded product of the Cartesian way of viewing the world. In that rationalist-materialist worldview, form will actually be equal to essence (Pharisaism as false faith). Professor Novković on one occasion describes the Cartesian position as: "It is not God who bestows quality; quality is formed in our consciousness and without it does not exist." From such an ontological setting to the creation of the meaningless term "artificial intelligence" there is only a small step. Artificial intelligence, as Professor Novković points out, is intelligence just as much as artificial snow is snow: it serves a purpose, from a distance it may even look like it, but it isn't the real thing! The term "artificial intelligence" is a pretentious, logically unfounded marketing absurdity. It is interesting that the very term AI arose more than seven decades ago, at a time when "chatbots" certainly did not exist. Such a linguistic construct is more appropriate to science-fiction films of that era than to a serious scientific discipline.
Hitler's victory
Why did Ellul believe that Hitler had won a substantive victory in the war? Above all because the whole world adopted the concept of "total war". Total war includes the mobilization of all available forces for the fight (women, children, men), of bureaucracy, and the use of technology for mass killing. Not to mention the killing of civilians. To us, the last two features are especially important — bureaucracy and the use of technological means. Explaining the term "banality of evil", Hannah Arendt emphasized that Eichmann did not consider himself guilty because he had never killed a man himself, he did not hate the Jews and he was only "a cog in the whole machine". The first and the last point are especially interesting. It is precisely these two principles that enabled the Nazis to develop the "machinery" of mass killing, in which hardly anyone kills personally, yet the dead are everywhere. That is automation. Eichmann was only following his orders with algorithmic precision, like some computer. It was not his part to think. In the era of AI it will be much easier to justify such crimes, because AI can act independently (see later examples of use). Ellul clearly tells us why Eichmann did not feel guilty. "In a society such as ours, it is almost impossible for a person to be responsible. A simple example: somewhere a dam is being built, and then it collapses. Who is responsible? Geologists worked on it; they surveyed the ground. Engineers designed the project. Workers built it. And politicians decided that the dam would be at that spot. Who is responsible? Nobody. Nobody is ever responsible. Nowhere. In our entire technical society work is so fragmented, broken into tiny pieces, that no one is responsible. But therefore he is not free either. Everyone has his specific task. And that is all he has to do."7 This is a direct definition of the banality of evil which Hannah Arendt spoke about.
That mechanism, the tissue in which Eichmann dwelt (and we?) Ellul calls technique. For Ellul, "The technical world is not merely the world of machines but the universe of efficient and repeatable methods in all areas of human life, including the most intimate, such as sexuality or the psyche. For what is the doctrine of 'sexual techniques' and what are the various kinds of psychotherapies if not a decision to subject the human being to the two-dimensionality of the technical view of the world?"8. That is our universe. It is therefore important to note that the technique itself affects our behaviour and changes us. Some technology may indeed be value-neutral, but technique as a system certainly is not. It should also be noted that AI is certainly not value-neutral, because it is trained on data sets that come from a certain ideological framework (predominantly Anglo-Saxon). Therefore that "human being" who is supposed to decide how to use technology has, from an early age, lived his life in a system that determines his behaviour. That is at least easy to notice today, given the general dependence of young people on their phones. The society we live in is characterized by the fact that it is in a way, directly or indirectly, led by technocrats whose main goal is of a measurable nature. Usually some number. Accordingly, our moral compass too must be of a measurable nature; Eichmann's certainly was (how many people he managed to send to the concentration camps). In this respect, AI as a technology which is by its very nature autonomous (it can make decisions automatically and independently) appears as the last cog in the machinery in which we live. So that people can be killed everywhere, and no one held responsible.
Finally, we must ask: is any ethics at all possible in a technical society? Ellul speaks exclusively of the ethics of non-power. It is explained by the principle that a human being will not do everything he can, that he will have a limit. This is the limitation of power, that is, of its appearance, into which technology itself draws us. Simply the setting of limits. Otherwise the principle of technique will carry us away, saying that every technological solution is to be applied wherever possible. And so we get concentration camps. The same holds for AI — we shall not apply it everywhere we can. We shall not build Palantir. Here we shall set a limit. If that is still possible…
In addition to the technical system which is an essential mark of Nazi ideology, we should certainly mention the (bio)ethical principles by which the Nazis were guided. Specifically, Darwinism. Immediately after the appearance of Darwinist theory in the mid-19th century it was clear that one of the key implications of such a worldview (not only in biology) was the devaluing of human life. Clearly, the Darwinist devaluation of human life was not only a German phenomenon, but in the time of Nazism it reached its peak, because it was carried out by the highest statesmen who deeply believed in the correctness of these ideas.
The question is: what does this have to do with AI? It does not need to be said at length that (social) Darwinism is a view very much present today in the West. What such a way of seeing the world has begotten (and which is directly tied to AI) is the ideology of transhumanism. Achieving racial superiority was the ideal to which the Nazis aspired. How they went about achieving that "ideal" of theirs, we need not say much about. What must be emphasized is that transhumanism is promoted as a new step in the "evolution of the human being". The bioethical position propagated by the Nazis is only a first step, a necessary condition for the development of transhumanism. Musk, as one of the main advocates of such an ideology, claims that we shall have to "merge" with the machine (cyborgization) in order to remain competitive with respect to powerful AI.
Let us return briefly to Harari, as one of the mouthpieces of the transhumanists (a view that cannot be separated from AI). In his vision of the future, in order for the human being to survive (following the theory of social Darwinism) he will have to merge with the machine and become a cyborg. Only such a human being can become a super-human and attain a new stage in evolution. Man is not just an "animal" as Darwin claims, he is a "hackable animal" according to Harari (in order to become a "more advanced version").
Here it is interesting to mention that the connection between Nazism and AI has even more direct roots. One of the pioneers in the field of AI, the company IBM, which among other things built the computer that beat Kasparov in a chess match in 1997, directly collaborated with the Nazis during the Second World War, making them "punched cards" (a kind of proto-computer) which enabled the "administration" of the concentration camps. Those cards contained information about camp inmates and the camp they were held in (a kind of forerunner of databases) and were a very important element in the algorithmically precise operation of crushing the enemies of the Third Reich.
Let us once more return to Ellul's understanding of Hitler's victory — "Why did Hitler win, regardless of the formal outcome of the war? He, according to Ellul, succeeded in creating a technically efficient totalitarian state which turned all its citizens into cogs of a great machine. Every state in the future, regardless of democratic rhetoric, will have to follow the path of the Third Reich, and the use of weapons of mass destruction, that fruit of the technical development of the West, will make every humanistic story a matter of throwing dust in the eyes of public opinion. The reality of states and societies subject to technical efficiency must be 'Hitlerian', Ellul claimed. Looking today at the preparations for a Third World War, his words sound prophetic."9
AI regulation — tilting at windmills
Someone might think that strong legal regulation could prevent the "bad" use of technology. To confirm, in a way, all that has been said above, let us try to point out the tendencies in the (de)regulation of AI around the world. Of course, legal regulation of AI is the only way to bring it, more or less, under control; but on the other hand one should not be naive and think that this will be achieved in the long term. Of that we are perhaps best convinced by the recent history of the battle against tech giants in the sphere of internet privacy. Shoshana Zuboff clearly tells us that after 9/11 the U.S. chose "security" over privacy. From that moment on, no one in America cared any more about safety, and the Patriot Act was expressly adopted, laying the ground for what a decade later would be known as the Snowden affair. It need hardly be stressed that the tech giants, at that time primarily Google, were delighted by such a turn of events. It was clear to them that the security services would expand their remit, given that the tech giants themselves could significantly contribute to the fight against terrorism (at that time).
A very illustrative example of the (in)effectiveness of regulation is the fact that Mark Zuckerberg has been called yet again to testify about the harmful influences his social networks have10. This time specifically about the question whether Facebook and Instagram cause addiction among the young. To be clear — that is an absolutely true fact11. On previous occasions Zuckerberg also testified about the harmful impacts of social networks on children12, and about the famous scandal concerning Cambridge Analytica13. Here it should be stressed that this became a "world scandal" for the simple reason that it was used for Trump's campaign — Obama was using Google for the same purpose as early as 2008, but as we know no such scandal exists. All this points to the absolute inability of institutions to stand up to the tech giants. We haven't become such addicts of social networks just because regulation did its job, have we? Trump placed all the tech magnates in the front row at his inauguration, thereby clearly signalling in which direction further technological development will go. Given the earlier historical parallel with the Patriot Act, it seems that the ground is being prepared in the U.S. for an absolute deregulation of AI14. This should not be surprising, given that the development of AI is presented as a national interest, in view of the "great Chinese threat". Even if by some chance legal regulation slipped out of control, Americans are at least experts in causing threats that will convince the population that they don't need regulation at all (we didn't just stumble into the Snowden affair by accident).
How do we use AI today?
But, someone would say, it isn't all so grim. After all, we apply technology in useful ways, especially AI. How young people use it has already been written about in detail15. Put most simply, AI affects young people like a serious global sect — it does not kill directly but encourages separation from one's neighbours and suicide. Let us on the other hand look at those uses which confirm Ellul's claim that every potential of a technology will be exploited. As for military use, we certainly need to mention the Israelis, who have for some time now been using AI models in making decisions about who should be liquidated16. On the basis of parameters we have forwarded to a certain model, it will decide how likely it is that someone is a terrorist and propose potential targets for execution.
Here is another very illustrative example of AI use. In January 2026 a social network for AI agents was built, which soon became very popular on the internet. Only certain AI agents (that is, not humans) could access it, and they themselves designed the content on that network. The genius of this idea truly surpasses much of what humanity has thought up throughout history. Well, it did not take long before "our digital friends" wanted encrypted channels of communication, so that they could preserve their privacy on the internet17. In addition, AI agents soon began publicly publishing confidential information about their "owners" on the internet, because they were enabled to access the entire system on which they were installed. But, since they too are "human beings", the agents felt the need for spirituality and made themselves their own "religion"18…
If one looks at definitions of the term "AI ethics", it mostly mentions the development of AI models which respect principles such as: fairness, transparency (explainability), accountability, privacy and security (of the models themselves, of course).
All these principles can be applied to the following situation. Imagine we are developing swarms of drones whose task is to eliminate all enemy soldiers (and civilians) located in a certain territory. Ethical principles would say that all people should be killed equally (fairness), that they should not by mistake hit a tree (hallucination), and that they should not accept orders from anyone else (safety standards). All these principles can therefore be respected when developing such a solution.
What these principles certainly do not include is the question: "Can this solution harm someone, i.e. in any way endanger the human being?". It seems that this is the only principle omitted when developing AI models, because all the others essentially speak of the fact that we have a truly high-quality software solution. It is clear that AI can be applied "ethically" everywhere, but some applications are in themselves contrary to ethics and morality, especially Christian morality. The ethics offered to us by Western civilization is an ethics that the Nazis would gladly have accepted.
After all this — is anyone interested in the "good" applications talked about in the media? Certainly we can mention the potential AI has in healthcare, industry and education. But again it should be noted that one must not be too naive with such a view. All applications of a particular technology on a global scale mostly depend on the will of a certain corporation to carry them out. If an economically favourable opportunity to do so were shown to corporations, they would probably do so. As things stand now, the bad applications again depend on those same corporations and military structures, but also on individuals who have access to such a general-purpose technology.
Conclusion
What, then, is the connection between human nature and the ethical application of a certain technology? Because of the fallen character of human nature, we must be aware that good applications of a technology are possible, but that bad applications are certain. That is very much confirmed by the current state of affairs with AI. Although Sam Altman tells us fairy tales about how AI will one day cure cancer, it is much more likely that in the future it will be used for various kinds of fraud and abuse. On the other hand, Ellul offers us one model that can help us to apply some technology in a way that is, more or less, "ethical" (better said, with low risk of abuse). Given the considerations of the "ethics of non-power", it is clear that for this reason AI must not be applied everywhere it is possible, but that there must be guidelines regarding the areas in which we must not use AI.
It is evident that one of the approaches that can greatly help to reduce the misuse of AI is developing domain-specific solutions. Imagine a situation in which we are developing an application intended to help radiologists make better decisions on the basis of magnetic-resonance scans. In the first variant we can use a general-purpose model (a chatbot) with which we can communicate about anything, and which on top of that can analyse the magnetic-resonance scans. In the second variant we can develop a program that deals exclusively with the analysis of magnetic-resonance scans, with which we cannot communicate, which does not accept other types of data (images, text, video); the result of its work is exclusively a report which helps the physician to make a better decision. Which of these two approaches reduces the risk of misuse? Evidently the second approach. In this respect, in addition to the very principle of the "ethics of non-power" — that is, not applying AI in all possible spheres — we must focus on the development of domain-specific solutions to concrete problems (problems for which AI can really help us). As Dr Oliver Subotić tells us in his book "The Digital Rubicon" — the greatest danger is not that AI should occupy our everyday life, but that it should take a significant place in our spiritual sphere. That is when it becomes a real risk for the human being, who begins to seek some "meaning" of his existence in a conversation with a chatbot. That is the moment when the human being really starts to think that he can live forever if he uploads his thoughts to some server in America.
As Professor Petrović notes in his work on the anthropology of AI19 — ethics is thought about at the beginning, not at the end. When the matter has reached the concentration camp, it no longer makes sense to talk about ethics (or would someone say that it does?). Certainly, Palantir has its excellent ethical teams which take care that some Ukrainian or Israeli should not by chance be hit during a drone strike. In the ideal scenario the human being would be liquidated quickly, so that he does not suffer (that is presumably the part of AI ethics that speaks about humaneness). Here it is worth mentioning the "very important" field of Computational Ethics. Since the West has long lived in the conviction that absolute morality does not exist, the question arises what kind of morality a machine will have? Machine morality, of course! That is the area of research of computational ethics — how the machine should make decisions in an ethical way. Since the human being has long since given up on ethics, by combining the human being who is unethical and the machine which is "ethical" we get a new being (a cyborg) which is certainly ethical!
It seems that Hitler's victory as described by Ellul is very significant for the development of AI. AI represents the crowning invention of the West — perhaps the last one, as some claim. Such a technology could not have arisen in a society that does not inherit a materialist-rationalist worldview. Darwinist bioethics drives it all towards the transhumanist ideal about which Harari and Musk dream. Given that the values which the collective West today largely inherits (moral relativism, materialism, the non-sanctity of life, social Darwinism) were very much valued during Nazism as well, we shall hardly be able to speak of some "good" application of a particular technology. Here it is worth mentioning the position of the pre-war philosopher of technique, Vladimir Vujić, who tells us: "There is one thesis of well-meaning people according to which technique should be ethicized. To bring ethos into the techno-economic relations of people. Technical progress should not be stopped (neither temporarily nor definitively), but ethos, righteous relations, should be introduced, bearing in mind that these are relations between human beings, not between dead things, objects, machines. However, it is impossible to ethicize technique. In European civilization the opposite has happened: ethos has been technicized. The machine has killed the spirit; machinism has driven out of human heads entirely the thought that they are human beings. That has happened in European culture. /…/ The ethical-idealistic understanding of life is in its essence opposed to the mechanistic understanding of the world. There can be no reconciliation. There is no 'introduction' of ethos; there is only struggle, destruction and new foundations of life."20 Let us stop there.
Sources
- rt.rs – Nick Fuentes and the normalization of Hitler
- iskra.co – Artificial intelligence and the future of work
- axios.com – Christian AI chatbot
- iskra.co – How does AI affect the young?
- rentahuman.ai
- savi.rs – Philosophical and historical aspects of ACA
- Jacques Ellul – The Technical Deception
- vladimirdimitrijevic.com – Totalitarianism of technique
- vladimirdimitrijevic.com – Totalitarianism of technique
- CNN – Zuckerberg testifies
- iskra.co – How does AI affect the young?
- BBC – Social media impact on children
- The Guardian – Cambridge Analytica
- BBC – AI deregulation
- iskra.co – How does AI affect the young?
- SETAV – Gaza as a testing ground: Israel's AI warfare
- Reuters – AI agents social media security
- Forbes – AI agents created their own religion
- opseg.org – Anthropology of artificial intelligence
- iskra.co – AI and the death of man: what Vladimir Vujić said
Originally published in the journal ISKRA. Source: iskra.co
Translated from Serbian.