Aquinas Articulates Artificial Intelligence
By Mike Schramm
How the Common Doctor Teaches Us Some Common Sense About Human Thinking
The philosophy and theology of Saint Thomas Aquinas has long been considered within the Catholic Church to be a sure footing on which one could walk the path toward truth. One of the staunchest principles in the thinking of Aquinas was the delineation between faith and reason. This was not with the end of keeping these disciplines separate in the minds of his students, but so that they would be able to address audiences accordingly. It is this delineation that will help when comparing and contrasting the topic of human and artificial intelligence. The foundation for the human intellect is the rational soul. While the existence and powers of the soul are part of the Christian faith, they can also be known by human reason. When Aquinas addresses the principles of human intellect, he does so on the basis of philosophical argumentation and not from Christian revelation. This makes his thinking important in the conversation about Artificial Intelligence, which is developed and operates independently of Christian revelation.
Artificial Intelligence is defined by Dr. Copeland and Britannica as the ability of a “digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.” This article also acknowledges, “The term is frequently applied to the project of developing systems endowed with the intellectual processes characteristic of humans, such as the ability to reason, discover meaning, generalize, or learn from past experience,” but, “Despite continuing advances in computer processing speed and memory capacity, there are as yet no programs that can match full human flexibility over wider domains or in tasks requiring much everyday knowledge.” Even the most widely accepted definition acknowledges that the “I” in AI is largely metaphorical.
Taking a classical philosophical approach to the question of the nature of AI, one discovers the nature of a substance by examining its four causes: the material, formal, efficient, and final. It makes sense that the material cause, the stuff of which AI is made, is the data itself. That’s what AI is working with, similar to us when we think. The formal cause, which shapes the action as thinking, we would call the intellect. In AI this would have to be the code or operation that tells it to think. Skipping ahead to the final cause as there is a point of agreement here, the telos of AI one could also say is to perform functions. This is true for us when it comes to practical knowledge, but not speculative knowledge. AI has no capacity to desire to know for the sake of knowing, for contemplation. This shows that there is no agreement in efficient cause, that is, the agent providing the result. Any artificial intelligence finds its origin in human intelligence but communicated through a machine.
Are Humans Just Slow Computers?
The modern comparison of human thinking to Artificial Intelligence sees the human as a machine running off a slow computer. While there are strong parallels between computation in the brain and logical inferences drawn in a computer, this assumption is reductive. It must be addressed in a way that does not appeal to special revelation nor succumb to the mind/body dualistic trap of Descartes. What Aquinas has to say, which will conclude with the distinctions between human thinking and Artificial Intelligence, begins with the essential connection that thinking has with the soul. Because it rests on this foundation, we must first address the nature of the soul.
Aquinas presents “the nature of the soul” as that which is “the first principle of life of those things which live” (ST I. Q 75. A 1. Resp.). Aquinas provides some insight into what life is that helps in making these distinctions. Earlier in the Summa, when discussing the life of God, Aquinas speaks “of life” as “self-movement and the application of itself to any kind of operation” (ST I. Q 18. A 2. Resp.). In contemporary philosophical language this self-movement or “self-perfection” is called “immanent causality.”1 This is characterized precisely by the internal self-movement that flows from the nature of the substance itself. One could consider the capacity for life “built-in” to the DNA of the living thing in question. Immanent causality is contrasted with “transeunt causality,”2 which is understood as movement that is externally powered into the substance. Even if the agent of movement exists within the body of the substance, a battery inside of a machine for example, but the agent is not inherently part of that body, it would still be transeunt causality.
One cannot argue that there is self-movement of immanent causality in any machine because any movement afforded it is produced by something outside of it, either a separate human agent or energy supply. The soul is not simply the battery for the machine because of the soul’s unity with the body. The battery is never part of the substance of the machine, but the soul is the form of the body. There is a much closer unity between the soul and the body in a person than any part of a machine. Even the enlivening part of a machine like the battery cannot be said to form the machine like a soul does for a body.
There is nothing in Artificial Intelligence, which is organized information, that is naturally ordered toward self-movement. This means that Artificial Intelligence could never have the property of life, which means it could never have a soul. Because it cannot have a soul, it cannot have a rational soul. Without a rational soul, AI cannot have the power of intellect. Without the power of intellect, a rational soul, or life itself, AI can only imitate the nature and functions of humanity, but cannot have human intelligence per se, even if it can perform some of these functions to a very high degree.
One could argue that Artificial Intelligence is really the network of information, or the connections between that information. If this is the case, then the important question arises of whether AI can exist without a “body” like a physical computer or bits of information. To strain an analogy to swallow a conclusion, is the information like the cells and the AI like the soul that enlivens it?
The most significant connection Aquinas makes between the intellect and the soul is analogous to the relationship between the soul and the body. When discussing the union of body and soul he says, “the intellect which is the principle of intellectual operation is the form of the human body. For that whereby primarily anything acts is a form of the thing to which the act is to be attributed: for instance, that whereby a body is primarily healed is health, and that whereby the soul knows primarily is knowledge; hence health is a form of the body, and knowledge is a form of the soul” (ST I. Q 76. A 1. Resp.). The form of AI is not anything related to the intellect itself but simply the code input into the machine by, eventually, a human agent.
Going beyond just the gathering of data, which it need not be argued is possible in AI, Aquinas still affirms that the body and soul must work together in order for “the intellect to understand actually, not only when it acquires fresh knowledge, but also when it applies knowledge already acquired, there is need for the act of the imagination” (ST I. Q 84. A 7. Resp.). Aquinas articulates the difference between data and sense knowledge using the example of “the nature of a stone,” but he says this applies to “any material thing,” which, “cannot be known completely and truly, except in as much as it is known as existing in the individual” (Ibid.). However, the individual is first known “through the senses and the imagination” (Ibid.). Applying this to how AI “knows” data, it is not sensing the information, but only has the information imprinted on it. It is like having a picture printed on paper. The paper has the information, but it is not sensing anything about it.
Can Computers “Think”? Does It Matter?
Once we have established the uniqueness of the rational soul, which is due to its intellect, and that the intellect necessarily operates through imagination, and that the imagination requires sense imagery in order to form a phantasm, then the next question for Aquinas and for us and that the distinctiveness of human intellect relies upon is that of which the sensitive part consists.
For Aquinas, “the sensitive part” consists of “four interior powers” that makes thinking possible (ST I. Q 78. A 4. Resp.). These powers are the “common sense, the imagination, and the estimative and memorative powers” (Ibid.). The most important of these for our purposes is the common sense. Aquinas calls the “common sense” that which acts as the “root and principle of the exterior senses” (ST I. Q 78. A 4. Ad 1.) This means that the common sense is tied to the physical senses, but it is the one that “discerns” amongst them. Not only is the common sense necessary to know when one is seeing or smelling, and not only what one is seeing or smelling, but that seeing or smelling that thing is part of what makes it that thing. The common sense is the one that unites the others together and informs the intellect as to the substance of it.
With this in mind, there is no way one could grant that AI is capable of the common sense. As priest and AI expert Fr. Eric Salobir mentions, “we don’t know how to code this common sense that makes a child quickly know that a cat is a cat, that nettles sting and that a flame burns.” Fr. Salobir quotes Marvin Minsky saying, “there is no computer that is aware of what it is doing, but most of the time, neither are we.” The only thing close to mimicking this action is when one puts together a typical list of features, products of the exterior senses, and assigns them a certain substance. If (I see) it walks like a duck and (I hear) it quacks like a duck, then (I know)...you get the idea. But this is just human intellect applied to a machine. We are still coding, or to return back to Aristotle’s philosophical language, forming the intellect.
It seems like the closest thing AI does to “thinking” is the type of thinking sometimes done when we are on auto-pilot. Our brain gets used to the uniform data and “discerns” at the minimum level, which is why we are primed to make mistakes. It is when we are doing this sort of sub-thinking that machines are like us. The difference is when we do it, we know it is a defect. We catch our mistakes, or someone else does, and we readjust. When AI does it, however, it is an achievement.
Why Does AI Fail Metaphysically?
The Turing test is a famous test for determining what AI is. Posited by Alan Turing, it tests whether a person can have a conversation with an AI unit without realizing the unit is AI. I have always found it ironic that the true test of human intelligence is a machine’s ability to lie. But the machine itself is not actually lying because it is not an “unreality” from the perspective of the unit. To the unit, it is still doing what it is programmed to do, which is the “realest reality” it knows. It is only by standing outside the situation that we would see the scenario as deceptive. This is the difference between human, rational intelligence and artificial intelligence. It is not data, but wisdom. It is seeing how all the pieces fit together. This unifying principle Aquinas recognizes is the intellect.
The philosophy of Saint Thomas Aquinas, especially that which relates to the soul, its essential power of intellect and its relationship to the body, are not based in Christian revelation but in human reason. What we need now is for the Christian, who presupposes the uniqueness of the rational soul and the intellect that proceeds from it, to converse intelligently and challenge the mechanical notions of intellect that the Artificial Intelligence movement purports.
Mike Schramm lives in southeastern Minnesota with his wife and seven children. There, he teaches theology and philosophy at Aquinas High School and Viterbo University. He earned his MA in theology from St. Joseph’s College in Maine and an MA in philosophy from Holy Apostles College. He has written for Busted Halo, Homiletic and Pastoral Review, and the Voyage Comics Blog. Read his other WWNN pieces.
For a contemporary treatment of this, see Edward Feser’s Immortal Souls: A Treatise on Human Nature, p. 118, 210–18.
Ibid.