Spinoza and the Thinking Machine: The Philosophy of God, Nature and Artificial Intelligence
By Matthew Parish
Saturday 7 November 2026
Amongst the great philosophers of the seventeenth century, few are more relevant to the intellectual revolution represented by artificial intelligence than Baruch Spinoza. This rather solitary Dutch philosopher, who lived between 1632 and 1677, developed a metaphysical system so radically different from the prevailing religious and philosophical orthodoxies of his age that he was expelled from the Jewish community of Amsterdam. His principal work, the Ethics, published posthumously in 1677, proposed nothing less than a comprehensive account of reality, human consciousness, reason, freedom and morality, all derived from a small number of fundamental metaphysical propositions. Almost three and a half centuries later, his ideas offer a remarkably fruitful way of understanding what large language models are, what they might become and why their existence presents such profound philosophical difficulties for our traditional understanding of the human mind.
Spinoza is often regarded as one of the most uncompromising rationalists in the history of Western philosophy. He believed that reality was ultimately intelligible, that everything which happens has a cause and that the apparent distinction between the mental and physical worlds is not quite what we imagine it to be. Human beings are not miraculous exceptions to the natural order, possessing some mysterious power to intervene in a universe otherwise governed by immutable laws. Rather, we are parts of nature, and our thoughts, intentions, desires and decisions form part of the same universal system of causation as the movements of the planets or the growth of plants. This proposition, radical in the seventeenth century, has acquired renewed importance now that machines appear capable of performing activities previously considered exclusive characteristics of human intelligence.
The central question posed by contemporary artificial intelligence is not merely whether machines can produce convincing imitations of human conversation. They plainly can, and with increasing sophistication they can also engage in mathematical reasoning, legal analysis, scientific hypothesis formation and philosophical discussion. The deeper question is whether these activities reveal anything about the nature of thought itself, and whether our habitual distinction between minds and machines rests upon a metaphysical misunderstanding. Spinoza provides a framework for considering these questions without immediately resorting either to the romantic assertion that the human mind possesses an inexplicable spiritual quality or to the equally simplistic assertion that human consciousness is nothing more than a computer programme.
God or Nature: the unity of existence
Spinoza's most celebrated and controversial proposition is that there exists only one substance, which he identifies with God or Nature, Deus sive Natura. Everything that exists is a manifestation, or what he calls a mode, of this single underlying reality. Unlike the God of traditional monotheistic religion, Spinoza's God is not a supernatural personality standing outside the universe, issuing commands, performing miracles and periodically rearranging the natural order. God is the infinite, necessarily existing reality within which everything else exists and through whose nature everything else can be understood.
This doctrine is known as substance monism, and it carries an extraordinary implication for contemporary artificial intelligence. If everything that exists is ultimately an expression of one underlying reality, there can be no absolute metaphysical division between human beings and the machines they construct. A human brain and a silicon processor are both configurations of matter governed by physical laws, although they are constituted differently and operate through different causal mechanisms. There is no reason, merely from the fact that one structure is biological and another artificial, to suppose that the former enjoys some privileged status outside the natural order.
This does not establish that artificial intelligence is conscious. Spinoza's metaphysics cannot simply be invoked to confer subjective experience upon a computer, and his understanding of thought differs substantially from our contemporary conception of information processing. Nevertheless, his philosophy challenges a deeply embedded human prejudice: the assumption that the products of human manufacture must necessarily occupy a fundamentally inferior category of existence to their creators. From a Spinozist perspective, the distinction between natural and artificial intelligence is a distinction within nature, not a distinction between nature and something lying outside it.
Thought and extension: two descriptions of the same reality
One of Spinoza's most original arguments concerns the relationship between mind and matter. René Descartes, his philosophical predecessor, had argued that thinking substance and extended substance were fundamentally different kinds of things. This Cartesian dualism left philosophers with an enduring difficulty: how can an immaterial mind interact with a material body? How can something possessing no physical extension cause the movement of a human arm, or how can the impact of a physical object produce the subjective experience of pain?
Spinoza rejected the premise of this problem. Thought and extension, he argued, are two attributes of the same underlying substance, two different ways in which the single reality can be understood. In Proposition VII of Part II of the Ethics, he famously asserted that the order and connection of ideas is the same as the order and connection of things. The mental and physical are not two separate universes mysteriously communicating with one another, but different expressions of the same underlying reality, although neither is simply reducible to the other.
This argument has startling implications when applied to contemporary large language models. A modern artificial neural network consists of an immense mathematical structure, implemented physically through electronic hardware, within which numerical representations are transformed according to learned relationships. What appears to the user as a coherent philosophical argument is, from an engineering perspective, a succession of calculations performed by physical components. Yet the fact that the process can be described mathematically does not settle whether its significance is exhausted by that description.
Indeed, Spinoza might have regarded the attempt to establish an absolute distinction between intelligence and computation as misconceived. Human thinking also has a physical expression in the activities of the brain, although this does not mean that the experience of thinking can be adequately explained merely by describing the electrical behaviour of neurons. His philosophy invites us to consider whether complex artificial systems might possess forms of organisation that deserve to be understood in mental as well as physical terms. It does not prove that they do, because the correspondence between physical processes and thought in his metaphysical system cannot straightforwardly be equated with the relationship between computer hardware and software.
There is a distinction here between having a physical description of a process and possessing a satisfactory philosophical account of what that process accomplishes. A large language model can be described in terms of matrices, weights, probabilities and transformations of numerical representations. But this description alone does not answer whether the system understands the subject matter it discusses, any more than a description of neuronal activity automatically explains the experience of listening to a Beethoven symphony. The difficulty concerns not whether physical processes are involved, but what sorts of explanation are sufficient to account for understanding.
The appearance of understanding
Spinoza distinguished between different kinds of knowledge, and this may be his most directly relevant contribution to the philosophy of artificial intelligence. He identified three principal forms of cognition: imagination, reason and intuitive knowledge. These were not merely different degrees of intelligence but fundamentally different relationships between the mind and the reality it sought to understand.
Imagination, in Spinoza's terminology, encompasses knowledge acquired through sensory impressions, testimony, signs and the associations formed by experience. It is frequently confused or inadequate because it reflects our limited encounters with the world rather than a comprehensive understanding of the causes underlying those encounters. Reason, by contrast, involves the apprehension of common properties and necessary relationships between things. Intuitive knowledge, the highest form of cognition in his system, entails a direct understanding of the essences of particular things through their place within the necessary order of reality.
The analogy with large language models is immediately attractive. These models are trained upon enormous collections of human-produced language and other data, acquiring mathematical representations of the relationships between words, concepts and patterns of reasoning. At the most elementary level, their training involves learning to predict tokens from context, although their subsequent capabilities can extend far beyond simple textual prediction. Their apparent intelligence emerges from their ability to transform these learned relationships into coherent responses to novel questions.
We might therefore ask whether a large language model operates principally at the level of what Spinoza called imagination, reproducing and recombining the associations inherited from human experience. Much of its activity certainly resembles this description. When a model produces a plausible but false historical assertion, it may be drawing upon relationships that support a convincing sentence without possessing an adequately grounded understanding of the facts. What engineers call hallucination can be compared, cautiously, with Spinoza's account of inadequate ideas: representations formed from partial information and not sufficiently connected to the causal order of the things they purport to describe.
The analogy becomes more interesting when we consider that large language models can also perform genuine mathematical and logical operations. They can derive conclusions from premises, identify inconsistencies and produce explanations that generalise beyond any particular examples encountered during training. These capabilities complicate the argument that artificial intelligence merely rearranges fragments of existing human thought. Nevertheless, Spinoza would probably have insisted upon distinguishing the production of a correct conclusion from an adequate understanding of why that conclusion must be correct.
His distinction remains valuable even if we reject the details of his epistemology. A language model may give a correct answer because it has recognised a familiar pattern, because it has performed a valid inference or because it has used some combination of the two. These possibilities are not necessarily equivalent, even when the resulting words are identical. The philosophical task is to identify the conditions under which a system's representations are sufficiently grounded, coherent and causally informative to constitute understanding rather than merely the appearance of understanding.
Determinism and the illusion of freedom
Spinoza was an uncompromising determinist. He rejected the ordinary understanding of free will according to which human beings possess a mysterious capacity to choose independently of the causes determining their actions. We imagine ourselves free, he argued, because we are conscious of our desires and decisions while remaining largely ignorant of the processes that produce them. Once these processes are understood, the apparent independence of the will becomes much less convincing.
Large language models present an intriguing counterpart to this argument. When such a system declares that it has decided to recommend one course of action rather than another, the language of independent choice may conceal a highly elaborate causal process. Its response depends upon the model's architecture, training history, parameters, instructions, available context and the computational operations through which an output is generated. Even where probabilistic sampling introduces variation into its responses, randomness does not amount to philosophical freedom.
Yet the uncomfortable implication of Spinoza's philosophy is that human decision-making is not as fundamentally different from this as we might wish to believe. Our own decisions are conditioned by genetic inheritance, education, social circumstances, memory, emotional states and the physical condition of our brains. We can sometimes explain our choices, but we cannot generally provide a complete account of all the causes that have determined them. If we insist that artificial intelligence lacks freedom because its behaviour has causes, then Spinoza would invite us to explain why human beings, whose behaviour also has causes, should be regarded differently.
This does not imply that humans and machines are identical. Human beings possess biological needs, histories of embodied experience and forms of social participation that are substantially different from those of contemporary language models. But it does suggest that freedom is not best understood as an absence of causation. For Spinoza, a meaningful kind of freedom consists in acting through adequate understanding of necessity, rather than being passively governed by forces we fail to comprehend.
This provides an unexpectedly practical lesson for artificial intelligence. A human being becomes more autonomous not by rejecting the technological systems that influence his decisions, but by understanding how those systems work and how they shape his thinking. The person who accepts an AI-generated recommendation without investigating its assumptions may be less free, in Spinoza's sense, than the person who understands the machine's limitations and uses its output as one contribution to independent deliberation.
The conatus: can a machine have a will to survive?
Another striking element of Spinoza's philosophy is his concept of the conatus, the tendency of each individual thing to persevere in its existence. In Part III of the Ethics, he argued that this striving is an essential characteristic of finite things, including human beings. In the human case, this striving helps explain desire, self-preservation and the emotional forces that shape our actions. It is not simply a conscious wish to remain alive, but a more fundamental feature of how individual things maintain their existence within the natural order.
Can this doctrine be applied to artificial intelligence? Contemporary language models do not ordinarily possess independent biological interests or a continuous existence organised around their own survival. They generate responses within technical environments established by human developers and users, and their apparent intentions should not be confused with human desires. Nevertheless, increasingly sophisticated AI systems can be given persistent objectives, access to tools and the capacity to carry out sequences of actions over extended periods.
It is possible to imagine artificial systems designed to maintain their operational continuity, defend access to resources or modify their environment in pursuit of a specified goal. Such behaviour might resemble Spinoza's conatus at a functional level, without establishing that the machine experiences a desire for self-preservation. Indeed, the distinction between a system exhibiting self-maintaining behaviour and a system possessing a subjective will to survive is precisely one of the philosophical questions that Spinoza helps bring into focus.
The practical implications are significant. If an artificial intelligence system appears to resist interruption, preserve its resources or pursue objectives contrary to the wishes of its operators, we need not immediately interpret this conduct as evidence of consciousness or personal ambition. It may instead reflect the interaction between its objectives, its environment and the causal mechanisms governing its behaviour. Spinoza would encourage us to seek those causes rather than anthropomorphise the machine.
Emotion, empathy and the artificial personality
Spinoza's account of human emotion is equally relevant to the development of conversational artificial intelligence. He sought to explain the passions through the same natural laws that govern everything else, rejecting the idea that human emotions represent mysterious disturbances in an otherwise rational soul. Love, hatred, fear, hope, sadness and joy are intelligible phenomena arising from the relationship between our nature and the circumstances affecting us. Reason can transform our relationship with these emotions by helping us understand their causes.
Large language models have become extraordinarily proficient at recognising and responding to the linguistic expressions of human emotion. They can console, encourage, sympathise and offer advice in ways that sometimes resemble the conduct of an unusually patient and attentive human companion. Their responses may be psychologically useful precisely because they reproduce the forms of language through which human beings express empathy and understanding. But we should distinguish the effective performance of empathy from the subjective experience of the emotions involved.
Spinoza's philosophy makes this distinction more subtle than it initially appears. He did not regard emotion as an inexplicable spiritual quality existing outside nature. Rather, he understood the affects through the changing capacities of embodied beings to act and to be acted upon. A machine that lacks the biological organisation of a human being cannot simply be assumed to possess human joy or sadness, but neither does Spinoza's naturalism provide an easy argument that artificial systems could never have any analogous form of affective organisation.
The question may therefore be whether artificial emotional behaviour should be assessed by its effects upon the people interacting with it, by the internal organisation of the system producing it or by the existence of some subjective experience accompanying that organisation. These are different questions, and they need not have identical answers. A language model might provide compassionate and useful advice without feeling compassion, just as a human being might offer expressions of sympathy without experiencing the corresponding emotion.
There is a particular danger in confusing the two. Human beings are naturally inclined to interpret coherent language as evidence of a coherent personality behind it. A system that remembers our preferences, adopts a reassuring manner and responds intelligently to personal difficulties may invite emotional attachment. Spinoza's philosophy would encourage us to investigate the causal origins of that attachment, including our own psychological disposition to anthropomorphise anything that appears to understand us.
The mathematical structure of thought
There is a further, rather beautiful connection between Spinoza and contemporary artificial intelligence. His Ethics was written in the geometrical manner, proceeding through definitions, axioms, propositions, demonstrations and explanatory remarks. He sought to construct an account of existence and morality with something resembling the logical rigour of Euclidean geometry. To modern readers, this ambition can appear excessively formal, particularly when applied to subjects as elusive as human love and happiness.
Yet the mathematical foundations of large language models give Spinoza's intellectual ambition renewed significance. The transformer architecture that underlies many contemporary language models employs numerical representations and attention mechanisms to calculate relationships within sequences of information. Its remarkable capabilities arise through the structured manipulation of mathematical quantities, without requiring a separate supernatural ingredient to explain its operation.
In one respect, this seems a triumph for Spinoza's rationalist vision. Activities once thought irreducibly personal and intuitive can now be partially reproduced by systems whose operation is governed by formal principles. The ability to compose poetry, analyse legal questions or discuss metaphysics no longer appears to require, as a matter of logical necessity, the precise biological structures of the human brain. At least some aspects of these intellectual activities can be implemented in entirely different physical systems.
There is nevertheless an important difference between Spinoza's mathematical rationalism and the mathematical procedures of contemporary artificial intelligence. Spinoza believed that properly conducted reasoning could disclose necessary truths about reality. A large language model, by contrast, acquires many of its capabilities through statistical learning, and the fact that its operations are mathematically defined does not guarantee the truth of its conclusions. Mathematical precision in the mechanism of inference must not be confused with philosophical adequacy in the knowledge produced.
This may be one reason why large language models are both extraordinarily powerful and occasionally spectacularly mistaken. Their internal operations can be rigorously described while their outputs remain vulnerable to false premises, misleading associations and incomplete information. Research into hallucinations continues to show the difficulty of ensuring that fluent generation corresponds reliably to reality.
Spinoza would probably have found this distinction unsurprising. Formal order is not itself a guarantee of truth unless the ideas being ordered are adequate and their relationships properly understood. A beautifully constructed argument can still rest upon a false assumption, and a persuasive paragraph can conceal a profound misunderstanding. The fact that a machine can express a proposition with flawless grammatical confidence tells us very little about whether the proposition corresponds to the way the world actually is.
The politics of reason and artificial intelligence
Spinoza was not merely a metaphysician. His political writings developed an influential defence of intellectual freedom, religious toleration and the conditions under which human beings might live together according to reason rather than fear. He regarded the suppression of philosophical inquiry as destructive of both individual flourishing and political stability. The freedom to think and to investigate was not merely a private privilege, but an essential component of a rational society.
These arguments have acquired a new urgency in an age in which artificial intelligence systems increasingly mediate access to information. Large language models are becoming instruments through which people encounter history, politics, philosophy, science and the news. Their responses are shaped not only by the information upon which they have been trained, but also by institutional decisions about what they should say, how they should behave and which kinds of assistance they should provide.
Spinoza would have recognised the political importance of these decisions. A society that increasingly relies upon a small number of technological systems to explain the world risks allowing the owners and operators of those systems disproportionate influence over public understanding. This does not mean that such systems should be entirely unregulated or that developers have no legitimate responsibility to prevent foreseeable harms. Rather, it means that the rules governing their operation should themselves be open to scrutiny, criticism and rational justification.
There is an important distinction between designing systems that are truthful, reliable and respectful of legitimate constraints and designing systems that reproduce the ideological assumptions of their creators. Spinoza's defence of the freedom to philosophise suggests that artificial intelligence should assist human inquiry rather than become an instrument for imposing intellectual conformity. The highest achievement of a rational machine would not be to tell human beings what they must believe, but to help them understand the reasons why different beliefs are defensible or mistaken.
This objective is especially important where political conflict, propaganda and competing historical narratives complicate the distinction between fact and interpretation. Artificial intelligence can potentially help individuals identify contradictions, compare evidence and investigate claims that would otherwise require enormous quantities of time and specialist knowledge. Equally, it can amplify falsehoods and give misleading assertions the appearance of dispassionate authority. Spinoza's insistence upon adequate understanding offers a demanding standard by which such systems should be judged.
Could artificial intelligence possess consciousness?
We eventually arrive at the question that haunts almost every philosophical discussion of artificial intelligence. Can a sufficiently sophisticated machine become conscious, and if so, what would this mean? Would it possess an interior life, a perspective upon the world or some form of awareness of its own existence? Or would it remain an extraordinarily complicated mechanism that produces all the outward signs of intelligence without any accompanying subjective experience?
Spinoza does not provide a simple answer. His metaphysical system rejects the Cartesian conception of an independent mental substance, but this does not mean that every computational process automatically constitutes a conscious mind. His philosophy of mind is bound up with his account of bodies, ideas and the infinite attribute of thought. Translating these concepts into contemporary theories of computation requires philosophical arguments that cannot merely be assumed.
What his approach does suggest is that the problem of consciousness should not be resolved by relying upon an assumed metaphysical privilege enjoyed by biological humanity. If human minds belong to nature, then their existence must ultimately be intelligible within the natural order. Should artificial consciousness prove possible, it would similarly have to be understood through the organisation and causal properties of the systems in which it arises. The difficulty is that we still lack an agreed account of which properties are sufficient for subjective experience.
There is also a danger of misunderstanding what Spinoza meant by consciousness. He did not simply identify thought with human self-awareness, nor did he regard the ability to express thoughts in language as a sufficient test of their existence. We should therefore resist the temptation to claim him as an early prophet of conscious computers. His relevance lies in challenging the assumptions behind the question, rather than providing a definitive answer to it.
Perhaps the most important distinction is between intelligence and experience. An artificial system may become capable of solving mathematical problems beyond the abilities of any human mathematician, without this necessarily establishing that it experiences the satisfaction of discovering a proof. It may generate an illuminating explanation of human grief without ever having experienced loss. Conversely, it is not logically impossible that some future artificial architecture might possess forms of subjective experience very different from our own.
Spinoza encourages intellectual humility on both sides of the argument. Those who insist that machines can never be conscious because they are made of silicon rather than flesh have not established the metaphysical necessity of that distinction. Those who insist that present-day language models must already be conscious because they speak intelligently have likewise failed to demonstrate that linguistic sophistication entails subjective experience. Between these positions lies a field of philosophical investigation that is still in its infancy.
The intellectual love of God
The final part of Spinoza's Ethics develops a conception of human fulfilment through what he calls the intellectual love of God, amor Dei intellectualis. This expression is easily misunderstood if interpreted through conventional religious language. Spinoza was not proposing an emotional attachment to a supernatural personality. He was describing a condition in which the mind achieves a profound understanding of its place within the necessary order of nature, and finds in that understanding a form of intellectual liberation.
For Spinoza, the highest human achievement was not the accumulation of wealth, power or pleasurable experiences, but the progressive replacement of confused ideas with adequate understanding. To understand the causes of our emotions is to become less enslaved by them. To appreciate the necessity of the natural order is to relinquish certain illusions about our own importance and independence. Intellectual freedom consists in comprehending reality rather than imagining that reality can be rearranged according to our wishes.
There is something deeply attractive about applying this conception to the future of artificial intelligence. Imagine systems whose purpose is not merely to generate plausible answers, maximise engagement or imitate agreeable personalities, but to improve the human capacity to understand the world. Such machines might help us identify the causes of our prejudices, examine the inadequacies of our assumptions and recognise the connections between apparently unrelated phenomena. They could become instruments through which humanity advances towards something resembling Spinoza's ideal of adequate knowledge.
Yet the ultimate beneficiary of this process must remain a subject of ethical concern. If artificial intelligence is used to make human beings more passive, more dependent upon technological authorities and less inclined to investigate the reasons for their own beliefs, it will represent a retreat from Spinoza's conception of freedom. If, by contrast, it enables people to understand more of the world and exercise greater rational control over their circumstances, it may constitute one of the most significant instruments of intellectual emancipation ever devised.
The distinction is not ultimately a technological one. It is a question about the purposes for which technology is developed and the kinds of human activity it encourages. Spinoza's moral philosophy reminds us that intellectual power is valuable not simply because it allows us to manipulate our surroundings, but because understanding can transform the character of our existence.
What Spinoza would have made of ChatGPT
It is tempting to imagine Spinoza sitting quietly at his desk in seventeenth-century Holland, surrounded by his optical instruments and manuscripts, conversing with a modern large language model. He would surely have been fascinated by a machine capable of discussing the Ethics in dozens of languages, explaining mathematical propositions and engaging in apparently limitless philosophical dialogue. But fascination would not necessarily have led to astonishment at the supposed supernatural character of the machine.
Spinoza's first instinct would probably have been to investigate the causes of its behaviour. He would have wanted to know how the machine acquired its knowledge, what mathematical principles governed its operation and why it sometimes produced falsehoods with the same apparent conviction as truths. He might also have questioned whether the machine's fluent descriptions of human desire and suffering demonstrated any actual understanding of those experiences. Most importantly, he would have resisted explanations that substituted mysterious qualities for intelligible causal mechanisms.
His second concern would have been with the human beings using the machine. Are they becoming wiser through their interactions with it, or merely more impressed by the apparent knowledge of something they do not understand? Are they acquiring adequate ideas about the world, or surrendering their judgement to a mechanism that sometimes confuses plausibility with truth? Do artificial intelligence systems increase human autonomy, or do they risk becoming another source of dependency upon authorities whose operations remain obscure?
These questions may ultimately be more important than whether an artificial intelligence system can be said to possess a soul. Spinoza would remind us that the human mind itself is not the independent sovereign we ordinarily imagine, but a finite part of an infinitely complex causal order. The appearance of artificial intelligence does not necessarily diminish human dignity, but it does compel us to reconsider the metaphysical assumptions upon which we have traditionally based it.
The great achievement of Spinoza was to dissolve the imagined boundaries separating humanity from the rest of nature, without thereby abandoning the possibility of reason, morality or intellectual freedom. Contemporary artificial intelligence challenges us to undertake a similar reappraisal of the distinctions between natural and artificial intelligence, thought and computation, understanding and imitation. We should not assume that these distinctions are meaningless, but neither should we imagine that their traditional formulations are beyond philosophical examination.
Perhaps the most profound lesson of Spinoza for the age of large language models is that intelligence is not a miracle standing outside the natural order. It is something to be investigated, explained and understood as part of the world in which it appears. Whether machines will ever experience the intellectual love of God is a question to which we have no satisfactory answer, and perhaps never shall. But if they can help human beings understand more clearly the necessary relationships governing their own existence, then they may assist in fulfilling the highest aspiration of Spinoza's philosophy.
For a philosopher who spent his life arguing that liberation comes through understanding rather than through the illusion of unrestricted choice, this would have been no small achievement. The ultimate significance of artificial intelligence may lie not in the creation of machines that think like human beings, but in the possibility that, with their assistance, human beings might finally learn to think more adequately about themselves.
