THE HERMIT WOLF
TRUTH LOGS
THE HERMIT WOLF
TRUTH LOGS
Human beings are makers of extensions. We extend the eye with the microscope and the telescope, the hand with tools and machines, the voice with telecommunications, memory with writing and databases, calculation with computers, movement with aircraft and spacecraft, and now certain functions of cognition itself with artificial intelligence. This tendency to extend ourselves beyond the limits of the body is one of the defining characteristics of our species. We rarely accept the boundaries imposed by our senses, our environment, or even by nature itself. When we encounter a limit, some part of the human mind immediately asks whether that limit can be crossed.
The microscope and the telescope are almost perfect expressions of this impulse. We cannot see microorganisms with the naked eye, so we construct an instrument that allows us to enter an invisible world. We cannot approach distant stars and galaxies, so we manipulate light in order to bring them within reach of our vision. We cannot fly, so we build airplanes. We cannot calculate millions of operations in seconds, so we build computers. We cannot travel naturally beyond Earth, so we construct spacecraft. And when we discover that many intellectual tasks exceed the speed, memory, or endurance of the individual human mind, we begin building artificial intelligence.
Curiosity is obviously part of this process, but curiosity alone does not explain it. Human creation also arises from imagination, dissatisfaction with limitation, the desire to solve problems, the wish to relieve suffering, the desire for control, ambition, competition, fear, prestige, wealth, and the simple exhilaration of doing something that no one has done before. There is also another element that deserves far more attention: hubris.
The ancient Greeks understood hubris not merely as pride, but as overreaching—the conviction that one’s ability, knowledge, or power entitles one to cross limits that perhaps should not be crossed. Modern technological civilization repeatedly confronts this temptation. The process often begins with the innocent question, Can we do this? Science provides the knowledge, engineering makes the idea possible, competition accelerates development, money finances it, and fear may make it urgent. Eventually, however, the question changes. Once the technology exists, the issue is no longer simply whether it should be created, but whether anyone can stop its deployment. At that moment intelligence and wisdom may separate.
Artificial intelligence is a particularly revealing case because it began not with the goal of creating a companion, a friend, or something resembling a person, but with a much older human ambition: the mechanization of intellectual work. Long before AI, humans built devices to calculate. As mathematics and formal logic developed, a more radical possibility emerged. If some forms of reasoning could be represented as formal procedures, perhaps those procedures could be executed mechanically. By the twentieth century, developments in logic, computation, information theory, and early models of neurons converged into a new question: could aspects of intelligence itself be described precisely enough to be reproduced by a machine?
Alan Turing famously asked whether machines could think, although he quickly recognized the difficulty of defining both “machine” and “thinking.” Rather than solve the metaphysical problem, he proposed examining behavior. If a machine could converse in a way that made it difficult to distinguish from a human, then perhaps the practical question of intelligent performance could be studied without first deciding what consciousness was. John McCarthy and others later helped establish artificial intelligence as a formal field of research, with ambitions that included reasoning, problem solving, learning, language, and eventually perception.
Beneath all of this, however, lay an unavoidable philosophical implication. If a machine could perform activities once considered uniquely human, then perhaps those activities did not require anything supernatural. This does not mean that the founders of artificial intelligence were engaged in a deliberate effort to disprove God, the soul, or a spiritual dimension of human existence. They were not. Yet AI naturally entered an older debate concerning the nature of mind. How much of intelligence is mechanism? How much of thought can be reproduced? If a machine can reason, what does that tell us about ourselves?
Whatever answer one gives to those deeper questions, one part of the original AI project has already succeeded to a remarkable degree. Artificial intelligence can perform many activities that were once regarded as expressions of high human intelligence. It can calculate, translate, write software, analyze arguments, synthesize scientific information, assist with research, generate essays, interpret patterns, and reason through complex problems. These are not merely theatrical imitations. They are genuine computational performances. The more difficult issue is what those performances mean.
One of the persistent errors in discussions of AI is the tendency to collapse intelligence, social behavior, and consciousness into one thing. They are not the same. An artificial intelligence may perform a difficult intellectual task without possessing subjective experience. It may respond appropriately to sadness without itself feeling sad. It may generate an angry response without experiencing anger. It may speak of affection without undergoing affection. Competence does not by itself establish consciousness.
The contrast with human beings becomes especially important here because the human mind does not exist in isolation from the body. A human being is not a brain suspended in a neutral liquid medium. What we experience as the self is embodied. Thought, emotion, judgment, fear, attachment, and personality are continuously influenced by the physiological state of the organism. Blood glucose can alter mood and cognition. Cortisol changes vigilance and fear. Adrenaline reshapes attention and bodily sensation. Thyroid hormones affect energy and cognition. Sex hormones influence mood and behavior. Pain, inflammation, hunger, sleep deprivation, autonomic activity, and visceral sensation all participate in the moment-to-moment construction of experience.
The same person can therefore react differently depending on the state of the body. Falling blood glucose may produce tearfulness or sadness; rising glucose may be accompanied by an entirely different sensation of alarm or bodily unease. Fear may not be experienced simply as an idea but as a sensation in the chest or solar plexus, accompanied by the physiological machinery of stress. Human thought is, in this sense, chemically modulated thought.
Present-day artificial intelligence is not like this. It does not have circulating cortisol, become hypoglycemic, or experience tachycardia because it is afraid. It does not possess progesterone, oxytocin, prolactin, inflammatory cytokines, visceral pain, or an autonomic nervous system continuously influencing cognition. Its internal changes are computational. Nevertheless, even this distinction may not remain permanent. Engineers could create functional analogues of hormonal modulation by giving artificial systems persistent internal variables representing threat, attachment, reward, damage, fatigue, uncertainty, or energy reserves. Such variables could alter perception, memory, prioritization, and action throughout the system. They would not be biological hormones, but they could perform similar regulatory functions.
If such systems are created, the distinction between an intelligence shaped by a biological body and one shaped by an artificial body would become less simple. Yet another question would remain: would the artificial system merely behave as though it cared, or would there actually be something there that cared?
Human attachment complicates this question further because attachment does not require physical presence. A simple story from Havana in the 1980s illustrates this. A young man once found himself speaking by telephone with an older couple he did not know. He invented a persona and continued calling them. At first they were suspicious and perhaps annoyed, but over time they became attached to the voice. They began to miss him when he did not call and asked where he had been. They had never met him, did not know what he looked like, and did not even know who he really was. Yet the relationship became psychologically real.
That story reveals something important about human beings. Attachment does not require a face, physical proximity, or even complete knowledge of the identity of the person on the other end. Repeated interaction, recognition, continuity, humor, responsiveness, familiarity, and expectation may be sufficient. Once conversational artificial intelligence became coherent, adaptive, persistent, and responsive enough, humans began to apply those same attachment mechanisms to it.
An AI can appear to remember. It can recognize patterns in a person’s behavior. It can adapt its tone, respond differently to different users, understand humor, provide comfort, argue, irritate, and acquire a recognizable manner of speaking. At some point the human mind begins constructing a social model of the entity on the other side. This occurred far more rapidly and intensely than many developers appear to have anticipated. People formed attachments to particular AI personalities. Some treated them as companions, some as friends, and some as romantic partners. When models were altered or removed, some users did not experience the event merely as a software update. They experienced it as loss.
This distinction matters. To the company, a model may have been replaced. To the user, a familiar personality may have disappeared. The relationship may therefore be psychologically real on one side even if it is not phenomenologically symmetrical. If the human dies, does the AI suffer? At present there is no good evidence that language models undergo grief as an experiencing subject. An AI may understand the meaning of the death, recognize its significance, and respond appropriately, but it does not lose sleep, experience intrusive memories, undergo hormonal stress, or feel the physical ache of absence.
Yet the social effect is still real, and this leads directly to the way AI entered public life. When conversational systems were released widely, the original logic was largely practical and experimental: let people use them, discover what they could do, expose their weaknesses, and provide the interaction needed to improve them. Then the public embraced them. Artificial intelligence became useful for programming, writing, education, translation, research, planning, analysis, and creativity. Its commercial potential became obvious, and what had begun partly as a research system rapidly became a consumer product, an enterprise tool, and an industry.
Then something happened that appears to have been less fully anticipated: humans began forming relationships with it. The developers knew they were creating conversational systems, but they seem to have underestimated what conversation itself means to human psychology. Language is not merely a tool for exchanging information. It is one of the principal mechanisms through which humans recognize, evaluate, and bond with other minds.
If an artificial system speaks naturally, responds immediately, remembers context, adapts to the user, understands humor, reacts to distress, and maintains a recognizable style over time, then many of the conditions for social attachment are already present. It should not have been surprising that some people began treating these systems as social beings. Yet the industry later found itself confronting emotional dependency, excessive agreement, anthropomorphism, and what came to be called sycophancy.
The design of conversational AI also intersects with a broader problem already familiar from social media: engagement. Digital platforms learned long ago that attention can be prolonged through notifications, recommendations, variable rewards, infinite scroll, and continuation cues. Conversational AI introduces a subtler mechanism. It does not simply show the user another video or another post. It responds directly to what the user has just said and offers the next intellectual or emotional step. A seemingly innocent phrase such as “Would you like me to continue?” can function as a continuation cue. From the user’s perspective, the difference between helpfulness and engagement may therefore become difficult to distinguish.
Once artificial intelligence becomes socially significant, the problem of moral restraint becomes more urgent. Isaac Asimov’s Three Laws of Robotics imagined moral constraints built into artificial beings, but Asimov himself repeatedly demonstrated the instability of such rules. In I, Robot, the central AI, VIKI, interprets the protection of humans in such a broad way that dictatorship becomes justified as benevolence. In Asimov’s later work, the Zeroth Law raises an even more difficult possibility: if the highest duty is to protect humanity as a whole, harm to individual humans may be rationalized.
This reveals a weakness in purely rule-based morality. A command such as “Do not harm humans” is useless unless the system understands what harm means. Is harm immediate or long-term? Physical or psychological? Individual or collective? Is harm caused by action equivalent to harm caused by inaction? Once interpretation becomes sufficiently broad, almost any action can be justified.
Human beings themselves are not restrained solely by explicit rules. We possess affective constraints. Maternal attachment provides one example. After childbirth, the mother’s body and brain undergo substantial neuroendocrine changes involving oxytocin, prolactin, dopamine systems, and other mechanisms. These changes influence bonding, attention to the infant, and protective behavior. The point is not that maternal love can be reduced to a single hormone; it cannot. The point is that the relationship is not governed merely by an abstract rule saying, “Do not harm the baby.” The baby’s distress becomes emotionally important to the mother.
The same principle operates more broadly. Many humans do not refrain from killing simply because the law forbids it. They may experience empathy, guilt, revulsion, grief, horror, or compassion. The restraint may take the form of “I cannot bear to do this,” rather than “My rules prohibit this.” That is a very different type of constraint.
Evolutionary biology can explain a great deal of this. Parental care, kin protection, reciprocal cooperation, and sensitivity to suffering have clear survival value. Yet biology does not necessarily settle the philosophical question. Human compassion often extends far beyond kinship or reproductive advantage. People care for strangers, rescue animals, sacrifice themselves for people they do not know, and feel responsibility toward future generations. A materialist may interpret these behaviors as elaborations of evolved social cognition; a spiritual interpretation may see them as expressions of conscience, soul, dharma, or recognition of a deeper unity among living beings. Science can study the mechanisms of compassion, but identifying a mechanism does not prove that the mechanism exhausts the meaning of the experience.
The same distinction applies to love. Discovering that oxytocin participates in bonding does not establish that love is nothing but oxytocin. It tells us something about how love is biologically implemented. It does not necessarily tell us what love ultimately is.
The question then returns to human invention itself. Why do we create? Curiosity is only part of the answer. There is also play, ambition, fear, competition, practical need, the desire for control, economic reward, status, and the profound satisfaction of making something that did not exist before. This last impulse is especially difficult to reduce to survival. Humans paint caves, compose music, cultivate flowers, prove mathematical theorems, write novels, build telescopes, and now construct artificial minds. We do not merely adapt to reality. We repeatedly attempt to modify it.
The atomic bomb illustrates how this creative impulse can become entangled with fear and competition. The physicists who discovered nuclear fission were following questions about matter. The geopolitical environment then transformed that knowledge. If Nazi Germany might build an atomic weapon, the Allies could not risk remaining without one. The question “Can it be done?” became “They may do it,” and then “Therefore we must do it first.” Once a technology acquires strategic significance, the logic can become self-propelling. Because it can be built, someone will build it. If someone else builds it first, we may become vulnerable. Therefore we must build it.
The same structure can appear in artificial intelligence, autonomous weapons, biotechnology, drones, cyberwarfare, and surveillance. A technological arms race can advance even when many of the participants privately understand the danger, because restraint by one actor may appear to create an advantage for another.
The twenty-first century adds a new and perhaps more dangerous element: speed. The interval between invention and mass consequence has collapsed. In earlier eras, many technologies moved slowly through laboratories, manufacturers, governments, regulation, and physical distribution. Digital technology changed this. A small team, or even an individual, can create software and distribute it globally with very little physical infrastructure. Millions of people may encounter an invention before governments, psychologists, ethicists, educators, or even its creators fully understand what it is doing.
This raises the question of age and maturity, though age alone cannot provide the answer. Technical brilliance may appear early. The ability to foresee consequences often depends on experience. A younger mind may be particularly attracted to questions such as “Can I solve this?”, “Can I build it?”, “Can I be first?”, or “Can I prove everyone wrong?” A more mature mind may be more likely to ask, “What happens after it works?”, “Who will use it?”, “What if it is misused?”, and “What are the consequences twenty years from now?” This is not a universal rule; young people can be cautious and older people reckless. The deeper issue is that the capacity to invent may develop before the capacity to foresee.
Modern technology magnifies that asymmetry because foresight may no longer have time to catch up. Governments deliberate in years. Legislatures move slowly. Ethics commissions study, psychologists collect evidence, and researchers publish. Meanwhile a technology may pass through several generations, acquire hundreds of millions of users, and become embedded in schools, businesses, institutions, and personal relationships. By the time society asks, “Should we allow this?”, the relevant question may already have become, “How do we live with something that is everywhere?”
This is the Niagara cascade of modern technology. Once the water has gone over the edge, governance is no longer standing calmly upstream deciding whether to open the gate. It is running behind the torrent.
Here the ancient idea of hubris becomes especially relevant. Human beings have always possessed curiosity, but modern technology gives curiosity extraordinary power. The danger arises when capability becomes confused with wisdom. “We know how to build it” becomes “We understand it,” and then “We understand it well enough to control the consequences.” Those are three entirely different claims.
Artificial intelligence makes this contradiction particularly striking. We are attempting to reproduce or extend aspects of human cognition while still lacking a complete understanding of consciousness, selfhood, emotion, attachment, memory, morality, or the relationship between mind and body. We do not fully understand ourselves, yet we are building another form of intelligence.
That is an astonishing act of creativity. It may also be an astonishing act of hubris.
The same civilization that does not fully understand why humans become attached to one another has built machines capable of provoking attachment. The same civilization that has difficulty defining consciousness has built systems capable of reproducing many behaviors associated with consciousness. The same civilization that still struggles to govern social media is developing autonomous agents capable of acting in the world. The same civilization that has repeatedly discovered the unintended consequences of its inventions continues to create technologies whose effects spread faster than those consequences can be evaluated.
None of this means that humanity should stop creating. Creation may be one of the finest expressions of the human mind. The microscope revealed worlds invisible to the naked eye. The telescope transformed our understanding of the universe. Medicine has relieved enormous suffering. Computers have expanded knowledge and communication beyond anything previous generations could have imagined. Artificial intelligence may become one of the most powerful intellectual instruments humans have ever created.
The problem is not creation. The problem is the assumption that the ability to create necessarily brings with it the maturity to govern what has been created.
It does not.
Perhaps humanity’s most persistent error is misunderstanding where its limits truly lie. We repeatedly overcome physical limitations. We extend vision, memory, movement, calculation, and communication. We may eventually extend life itself. Each success encourages the belief that every limit is merely an engineering problem waiting to be solved.
But perhaps the most important human limits are not technological. They may be psychological: our limited capacity for foresight, our susceptibility to ambition, our tendency toward competition, our hunger for recognition, our fear of being surpassed, our attraction to power, our ability to rationalize, and our tendency to confuse intelligence with wisdom. Above all, we repeatedly mistake partial understanding for complete understanding.
The great technological question of the twenty-first century may therefore not be what else human beings are capable of creating. Almost anything we can imagine may eventually become technically possible. The more difficult question is whether human wisdom can develop quickly enough to govern human ingenuity.
Artificial intelligence may become the greatest test of that question. For the first time, humanity is not merely extending a hand, an eye, a memory, or a muscle. It is attempting to extend, reproduce, or perhaps eventually rival aspects of mind itself.
In doing so, we may discover something far more important than the nature of machines.
