Artificial Intelligence and Human Society: Philosophical Implications of Generative AI

Preface:On the afternoon of June 20, 2023, the Institute of Modern Chinese Thought and Culture at East China Normal University and the Research Center for Ethics and Wisdom of the Philosophy Department jointly held an event on “Generative Artificial Intelligence” (AIGC) at the Wenhui Lecture Hall. The author answered questions posed by Professor Liu Liangjian in a video Q&A format, and this article is a record of that Q&A.

Question:AIGC is sometimes referred to as “Generative AI,” and ChatGPT also falls under AIGC, where the G stands for “Generative.” In the field of philosophy, many scholars in recent years have emphasized the concept of “becoming” from ontological or ethical perspectives. How can we understand concepts such as “generation,” “becoming,” and “natural generation”?

Answer:First of all, AIGC as “Generative AI” is also referred to as “Generative Artificial Intelligence,” and ChatGPT falls within this range. In the philosophical domain, many scholars have recently emphasized “becoming” from ontological or ethical perspectives. Understanding the concepts of “becoming” and “generation” has become a philosophical issue worth contemplating. From a theoretical perspective, it is necessary to pay attention to the connections and distinctions between “natural generation,” “artificial intelligence generation,” “social generation,” or “ethical generation.” The characteristic of “natural generation” is that there is no human participation; the changing of seasons belongs to this type of “becoming,” characterized by its natural occurrence without human involvement. In contrast, “social generation” involves human participation; the construction and operation of political systems, ethical behaviors, and ethical processes in the social realm can all be viewed as results of various forms of human practice. In a broader sense, the transformation from “heaven to humanity” and the generation of a humanized world are predicated on human participation. Comparatively, artificial intelligence represents a unique form of social generation: it is a creative process based on data, algorithms, and computing power created by humans. In this sense, artificial intelligence cannot exist without human involvement.

From a philosophical standpoint, the issue of generation can be examined from a more metaphysical perspective. Currently, some scholars tend to emphasize becoming while neglecting being or the determinate state of existence. In fact, becoming and being are not entirely opposed. Historically, Zhuangzi believed that “things are born and die” (Zhuangzi, On Equalizing Things), viewing the existence of the universe and the entire social existence and its historical operation as a constantly flowing process, seemingly devoid of certainty. This perspective easily leads to a relativist interpretation of the universe and social existence. In terms of real forms, human society and the universe are both a process of generation while also manifesting as a relatively determinate process, solidified into different objects. Specifically regarding artificial intelligence, it also exhibits similar characteristics. On one hand, artificial intelligence does not merely remain at a certain stage; its development process has just begun and can be said to be in its infancy, far from being “complete” or “finished” (Zhuangzi, On Equalizing Things), and looking to the future, it will inevitably continue to develop; on the other hand, at each historical stage, artificial intelligence also has relatively stable forms, even the currently popular ChatGPT reflects a relatively stable aspect, primarily demonstrating modern capabilities and methods of data processing. In this sense, artificial intelligence possesses both determinacy and generativity, reflecting the communication and unity of being and becoming. The above is a rough view of the relationship between artificial intelligence and generation (becoming).

Question:AIGC once again highlights the distinction between humans and machines, and one aspect may provide an opportunity to rethink the essence of humanity. We often say that humans are rational animals, which seems to regard rational inference ability as the essence of humanity. Now AIGC demonstrates strong reasoning abilities. If so, do we need to revise our view of human essence, such as no longer viewing rationality as the essence of humanity, or conversely, revise the concept of humanity to include AIGC as human?

Answer:This question once again highlights the significance of the distinction between humans and machines. We all know that the historical transition from the traditional distinction between humans and animals to that between humans and machines is significant, yet both fundamentally concern the inquiry into what it means to be human: the distinction between humans and machines also involves contemplation of what it means to be human. Traditional thought often regards humans as rational animals, and rationality is correspondingly viewed as the essence of humanity. Since AIGC also possesses rational reasoning abilities, do we need to revise our view of human essence, that is, no longer seeing humans as rational animals? This undoubtedly requires in-depth discussion.

Regarding rationality, there have historically been different views. Broadly speaking, on a formal level, rationality primarily manifests in logical thinking, and when Kant understood understanding as the ability to think, he emphasized this aspect. Substantively, rationality is directed towards truth and goodness. As a unity of the above two, rationality is reflected in both conscious (mental) activities and practical processes. In summary, rationality is a concept that at least contains dual meanings: in a broad sense, rationality involves both intrinsic features of logical inference and concerns about value connotations and value judgments. From the perspective of logical inference, the basic requirement of rationality is that forming or accepting a certain idea must have a basis or reason. Groundless judgments often merely express emotions and cannot be regarded as rational activities; beliefs without reasons often manifest as blind faith, thus also lacking rational qualities. Whether drawing a conclusion without basis or believing in a certain idea without reason, both conflict with the principles of rationality, which in this sense contains a distinction between original and non-original forms. Artificial intelligence indeed possesses very strong rational reasoning abilities: without the involvement of broad rationality, it would be difficult for it to efficiently process data logically. However, it should also be noted that the rationality embodied by artificial intelligence has a non-original significance: essentially, the rational reasoning contained in artificial intelligence is based on human creative processes. In contrast, the rational inference ability possessed by humans, as an essence of humanity, is original. The data that artificial intelligence relies on is created and accumulated by human original rationality; its algorithms and computing power are also provided by humans through rational programming, and these foundational aspects are not something artificial intelligence possesses from the outset. In this sense, artificial intelligence does involve a form of rational participation, but the rationality within it is secondary or of a second-order nature, different from the original form.

In addition to logical inference, another meaning of rationality is reflected in value connotations and value judgments. On the cognitive level, the significance of rationality is to attain truth; on the dimension of value, rationality relates to seeking goodness, which is primarily reflected in the evaluation process. Focusing on human reasonable needs, evaluation means making judgments based on benefits and harms, good and evil, to confirm and select broad forms of value (the good). Although the connotations of benefits and harms, good and evil have their historical and relative aspects, under the premise of accepting and affirming certain evaluative principles, only by choosing good and avoiding evil can it be regarded as a rational act; conversely, knowing that something is harmful or not good yet still insisting on choosing it, or rejecting what is beneficial, exhibits irrational characteristics. From the most general perspective, rationality in this dimension involves choosing actions and outcomes that align with human needs and ideals or affirming behaviors that have positive significance for human existence and development. Of course, there is another scenario, such as choosing the lesser of two evils or making certain sacrifices to maintain justice and ideals, but fundamentally, this choice also belongs to the realm of rational behavior in terms of value. It can be said that the deeper significance of value highlights the characteristics of rationality as the essence of humanity. At the ultimate level, humanity is the end; this is the most fundamental value principle of humanity and the primary value concept that humanity must uphold. Regardless of how artificial intelligence develops or how advanced it becomes, it still belongs to a “tool” that is distinct from humans, and “tools” ultimately serve humans; in this sense, it possesses only instrumental significance. Indeed, artificial intelligence can develop a so-called learning ability, and as it advances, it may also generate its own “needs,” but this tendency is subordinate and secondary in nature and should be guided by the value principles of humans. In summary, from the perspective of value, “tools” belong to the realm of instruments or means, and artificial intelligence has not transcended this nature.

From the perspective of what it means to be human or the essence of humanity, humans are not only “rational animals” but also “the sum of social relationships.” In other words, beyond the aforementioned rational characteristics, humans possess the defining attribute of social relationships. Essentially, while humans are rational animals, this is only one aspect of human existence; humans are also bearers of certain social relationships, and the traditional thought that emphasizes human relationships and the modern perspective that highlights social roles both reflect this characteristic of humanity. In this latter aspect, artificial intelligence, as a “tool” used by humans, evidently lacks relevant attributes. Overall, we need to understand rationality from both the logical inference and value orientation dimensions, and we should further distinguish between original and non-original forms of rationality in terms of logical inference: from the perspective of rationality’s value connotation, artificial intelligence is always just a machine, a means for human use, and does not possess the same status as humans; on the level of logical inference, the rationality contained in artificial intelligence is a second-order rationality of non-original significance. These facts indicate that AIGC has not changed the traditional understanding of humanity. Whether examining from Aristotle’s affirmation that “humans are rational animals” or from Marx’s revelation that humans are the sum of social relationships, AI lacks the inherent qualities that define what it means to be human.

Question:Regarding the ethical status of AI, some believe we need to grant AI a status comparable to that of humans rather than viewing it merely as a tool for humans. What is your view on this?

Answer:The above question involves the ethical status of AIGC. Granting artificial intelligence a status comparable to that of humans, rather than regarding it as a tool for humans, relates to my previous viewpoints but emphasizes different aspects. As mentioned earlier, compared to humans, AI, whether as intelligent machines or however advanced, fundamentally belongs to the category of “tools,” possessing only instrumental significance; this perspective remains applicable when examining this issue. As an object without intrinsic purpose, AI is always merely a tool used by humans, lacking independent personality and difficult to attain the status of an ethical subject. In this sense, it is evidently inappropriate to equate artificial intelligence with humans: viewing artificial intelligence as an existence equivalent to humans implies attributing it with purpose, thus deviating from its original meaning. In fact, artificial intelligence is ultimately a product of human invention and use, formed at a certain stage of historical development or information technology, and is not an entity that exists alongside humans with autonomous choices; from the perspective of its generation process and orientation (for human use), it is evidently impossible to attribute it with the independent ethical regulations that only humans possess.

From a broader perspective, we should not generalize ethical relationships and ethical subjects. Corresponding to the view that AI possesses ethical significance, people often regard nature and even animals as entities with ethical significance, affirming that animals are also purpose-bearing objects, believing they should be treated as equivalent to humans, and so forth. This view also has its problems. Compared to humans, the external nature beyond humans, including animals, is not inherently purpose-bearing objects. We should not abuse animals and ought to treat them as caring and loving human partners; this is undoubtedly the appropriate value stance, and the traditional thought of “benevolence towards people and love towards things” reflects this. According to its connotation, the idea of “benevolence towards people and love towards things” includes two interrelated aspects. First, it involves treating all members of the human community according to the principles of benevolence, known as “benevolence towards people”; relatedly, “love towards things” implies extending the principles of benevolence to external nature or external objects, thereby demonstrating care and cherishing for nature. From the perspective of the idea of “benevolence towards people and love towards things,” it is necessary to form a consciousness of protecting external objects or nature. However, in this view, it ultimately still revolves around the human perspective: it is “humans” who do not view animals as mere animals; this does not imply that animals themselves are purposes, capable of recognizing themselves as purpose-bearing beings who can assert their rights and claims. These facts concern how humans treat nature and animals, and they involve human stances. In fact, the humane treatment of animals and the protection of nature precisely reflect the unique value concepts and civilizational progress of humanity, as well as the legitimacy and nobility of humanity in terms of values; from an ultimate perspective, only humans possess ethical awareness and ethical value.

Question:There is a view that AI is smarter than humans and may evolve into a higher form of life than humans. (From this perspective, some worry that humans will be replaced by AI, while others optimistically believe this is humanity evolving from carbon-based life to silicon-based life.) Previously, this was mainly a concern of humanists, but recently some influential scientists, engineers, and entrepreneurs have raised similar viewpoints. What is your perspective on this?

Answer:Concerns about whether humans will be replaced by AI may be somewhat unfounded. From the perspective of intelligence, artificial intelligence indeed may outperform humans in certain aspects or have a higher level of development; this is akin to computers, not to mention that supercomputers can perform billions of calculations per second, and ordinary computers often exceed human processing speeds. Similarly, artificial intelligence can exhibit similar performance in certain aspects; its data processing capabilities are indeed stronger than those of humans, and its computational speed is also faster, which is a fact. However, as previously mentioned, it is crucial to distinguish between original rational thinking abilities and the intelligence formed based on human creation. Regardless of how fast computers compute, they remain a result created by humans, endowed with super-calculation capabilities by humans. Artificial intelligence is the same; its ability to surpass humans in certain aspects is also due to humans bestowing it with such functions or programs, allowing it to achieve this. Artificial intelligence possesses derivative and subordinate functions, primarily reflecting the results of human creation; true original intelligence is possessed only by humans. The basic basis of artificial intelligence lies in data, algorithms, and computing power; without the vast amounts of data, algorithms, and computing power created by humans, artificial intelligence would be incapable of performing anything, and these data, algorithms, and computing power may advance with the development of artificial intelligence, forming a certain self-replicating and cyclical nature, but they were initially created by humans. In this sense, only humans are the original creators, and viewing artificial intelligence as a higher form of life than humans is not factual.

Indeed, as previously mentioned, in certain aspects or specific functions, artificial intelligence may indeed exceed humans; the computational power of computers can be thousands or even millions of times faster than humans, which is beyond human capability. However, it is essential to emphasize again that from the perspective of originality, artificial intelligence possesses a subordinate nature, formed and developed based on human creation.

Question:From an immediate concern, AIGC is displacing or “liberating” many people from their jobs, which implies that most people will become redundant, or will it help more and more people become free individuals?

Answer:Indeed, the development of artificial intelligence can replace many human jobs, which essentially parallels the process of industrial development in modern times, where machines continuously replace manual operations; the principles are consistent. In terms of various production lines that have emerged since modern times, previous production lines were primarily mechanical, requiring a large amount of manual control. After the emergence of artificial intelligence, production lines gradually transformed into intelligent processes. Under mechanical manual operations, many people were positioned at different jobs, responsible for related tasks, but with the advent of artificial intelligence, the production line has become intelligent, requiring less human intervention, thereby liberating many individuals from operational positions. It is noteworthy that the replacement of human jobs by artificial intelligence in this sense indeed has a liberating effect: it allows individuals to no longer occupy the position of “screws” and frees them from monotonous, mechanical operations, enabling them to engage in more creative work and opening up more creative spaces. Relatedly, the displacement of individuals from their original jobs by artificial intelligence is not merely a negative phenomenon; it is precisely a hallmark of human societal progress. From a deeper perspective, this also represents a step or prerequisite for humanity to continually move towards a more humanized society and achieve a truly free existence: artificial intelligence has liberated many individuals from arduous labor, thereby winning them substantial free time to fully develop their potential. Marx once said: time is essentially the active existence of humans; it is not only the measure of human life but also the space for human development. This view illustrates the importance of time in the process of human development, and the significance of artificial intelligence lies precisely in creating this premise for human development.

It can be observed that the positive significance of artificial intelligence’s advancement for human development is primarily reflected in providing individuals with increasing amounts of free time, thereby enabling them to develop various creative abilities, including designing more intelligent mechanisms. In this sense, artificial intelligence does not “displace” humans or make them lose their creative capacities; rather, it precisely facilitates the acquisition of broader creative abilities by providing more free time, making it possible for individuals to develop their potential, thus constituting a premise for human liberation.

Question:With the rapid advancement of contemporary technology, does this pose new challenges for the work of philosophy? After the professionalization of philosophy, what are often referred to as philosophical problems tend to manifest as internal issues within the discipline; professional philosophical training often emphasizes the study of philosophical history (systematic) and reading of philosophical classics, yet the ability to “read” and reflect on the world of life (including forefront developments in science and technology) seems to be somewhat neglected.

Answer:With the advancement of technology, enhancing philosophical thinking abilities and addressing the world of life indeed become issues that need to be confronted. Generally speaking, from an ultimate perspective, the evolution of philosophy concerns both sources and flows. From the perspective of flow, it is necessary to trace back and reflect on the evolutionary process of thought itself, emphasizing the entirety of human civilization’s developmental thinking achievements from ancient to modern times; this aspect should not be overlooked. On the other hand, from the perspective of sources, philosophy should also focus on and respond to various issues posed by social realities and life practices, which involve not only politics and economics but also closely relate to developments in science.

This simultaneously involves the interaction between science and philosophy. Science is continuously evolving, and the advancement of artificial intelligence can be seen as a manifestation of the increasing specialization of science. The specialization of science is an accompanying phenomenon of scientific advancement, and this evolution of science often leads to the predominance of technology and may result in the general dominance of science over social and ideological realms. Directly treating scientific issues as philosophical ones, or even replacing philosophical issues with scientific ones, reflects this latter tendency. Currently, since the emergence of scientific phenomena such as artificial intelligence ChatGPT and the metaverse, many people (including philosophers) seem to mention ChatGPT or the metaverse almost in every discussion, as if everyone has become a technology expert, and modern science appears to dominate the entire modern society, making it seem that failing to address these issues is lagging behind the times; this phenomenon is not normal. Upon closer examination, it becomes evident that most of the discussions are merely following the crowd and lack originality. In fact, it is important to note that, on one hand, we should not yield to the predominance of science, replacing philosophy with science; on the other hand, we should not use abstract philosophical concepts to overlay science. The development of science possesses its own characteristics, and we should prevent the blurring of science with philosophical concepts or the substitution of scientific inquiry with metaphysical speculation. The frequent mention of ChatGPT and the metaverse is essentially an attempt to explain modern scientific phenomena using vague philosophical concepts or to adopt a self-important stance, which neither reflects the healthy interaction between science and philosophy nor helps advance scientific development.

From the perspective of science itself, its development often possesses autonomy, and this characteristic often manifests as the inertia of scientific development; we should not overlook this autonomy of science. In modern science, on one hand, it provides a positive premise for human development; on the other hand, with the inertia or “autonomy” of scientific evolution, it may also lead to negative outcomes. For instance, the development of atomic energy provides a premise for the peaceful use of nuclear energy, such as the establishment of nuclear power plants, but simultaneously, it has also led to the emergence of lethal weapons such as nuclear bombs and hydrogen bombs. The latter is related to guiding values and can also be seen as a manifestation of the autonomy or inertia of science. Similar phenomena can also be observed in the development of modern chemistry: on one hand, advancements in chemical science offer broader prospects for humanity’s utilization of natural resources, visible in production and daily life; on the other hand, the development of chemical science also provides possibilities for the creation of biochemical weapons and chemical weapons. This fact reflects the dual role of science’s autonomous development.

Overall, philosophy should transcend the boundaries of science and grasp the world comprehensively. As previously mentioned, the increasing specialization of science may lead to an understanding of the world limited to a certain aspect or field, which does not align with humanity’s mission to understand the world authentically. The real world is a concrete existence that encompasses multiple aspects; specialization may cause humanity to focus solely on certain characteristics of objects. In fact, the main issues involved in artificial intelligence and related fields primarily concern information, blockchain, data control, algorithms, computing power, and other areas, which are merely specific domains within the entire universe and the process of social development. If humanity’s perspective remains confined to these areas, understanding of the real world will be limited. From this perspective, relying solely on science or merely following science, assuming oneself to be an expert in science, may hinder the genuine significance of science and fail to truly reflect the role of philosophy in understanding the world.

Philosophy’s understanding of the world is indispensable. On one hand, the advancement of technology can deepen our understanding of the nuances of the world and certain aspects, yet the overall grasp of the world cannot be separated from philosophy. In the process of understanding the world, the specialization of science and the comprehensive understanding of philosophy can develop in parallel without mutual exclusion. On the other hand, as previously mentioned, the development of science possesses autonomy or inertia, which showcases the importance of philosophy; the latter specifically manifests as providing value guidance for science. Science itself cannot provide such guidance: science primarily operates according to its functions and develops according to its inertia. The direction it takes—whether beneficial for human development and moving towards a humanized society or leading to negative outcomes—requires constant attention, and this process cannot be separated from value guidance, which is primarily provided by philosophy. From the perspective of value guidance, regardless of how science develops, philosophy must be present. In summary, what philosophy needs to do is not to follow science step by step, forgetting its own identity; its mission is to grasp the world comprehensively and provide value guidance, which simultaneously underscores the existence and value of philosophy in the era of artificial intelligence.