《认知主体之后》中文版序言节选

By Geoffrey Chen

我写《认知主体之后》(After the Knowing Subject)，并不是因为我对既有哲学传统失去了耐心，更不是试图宣布某些哲学派别已经过时。相反，正是因为我长期浸润在这些传统之中——从大卫·休谟对认识基础的根本动摇，到伊曼努尔·康德通过批判哲学对认知主体的重建，再到二十世纪哲学试图将知识从心理确信中逐步抽离出来的各种努力——我才逐渐意识到，在人工智能已经深度介入现实运作的今天，我们面对的问题，已经不再完全落在这些思想最初所设定的提问方式之内。

作为一名独立研究者，我并不试图建立一个与既有哲学体系彻底对立的新学派。我更关心的，是从一个更基础的层面重新审视这样一个问题——在人工智能时代，哲学究竟在做什么，我们又应当如何继续进行哲学的工作。这种追问本身并不否定传统，恰恰相反，它必须被嵌入在已有的哲学框架之中，以它们作为背景、参照与对话对象，否则任何新的尝试都会失去坐标。

如果回看近代哲学的脉络，问题其实早已逐步显现。休谟已经敏锐地意识到，知识并不天然扎根于理性主体之中，而更多依赖于习惯、关联与经验的稳定性；康德正是在这一断裂处介入，通过对先验条件的分析重新确立了认知主体的中心地位，使知识得以被“拯救”。而在二十世纪，随着威拉德·冯·奥曼·蒯因对分析与综合之分的瓦解、威尔弗里德·塞拉斯对“给予的神话”的拆解，以及卡尔·波普尔将知识理解为独立于心理确信的客观结构，知识开始被越来越多地理解为一种嵌入整体系统之中的结构性存在，一种可以被讨论、检验、修正，甚至在相当程度上脱离个体理解而存在的对象。

这些思想并未彼此否定，它们共同构成了现代认识论的张力背景。但在人工智能出现之前，一个前提始终没有被真正松动——无论是经验主义、批判哲学，还是整体论与客观化的尝试，认知主体始终被默认为认识的最终承载者，理解始终被视为知识得以成立的隐含条件。

正是在这样的背景下，我开始意识到，一个长期被默认却很少被单独审视的前提，正在悄然成为当代思想的瓶颈。几乎所有主流哲学体系，都在不同程度上预设了主体的理解作为认识的核心条件。世界如何被理解，知识如何成立，判断如何获得正当性，这些问题被反复讨论，却很少有人追问，如果理解本身暂时缺席，认识是否仍然发生。

这个问题在传统时代或许显得抽象，但在人工智能系统已经能够在无人理解的情况下生成稳定判断、指导行动、并产生现实后果的今天，它不再只是一个理论假设，而是一种日常事实。模型可以在无人真正理解其内部机制的情况下持续预测成功，系统可以在无人解释其推理路径的情况下做出有效决策，制度可以在无人认同其意义的情况下持续运作。面对这些现象，如果哲学仍然坚持把主体的理解当作一切认识的必要前提，那么它要么不得不忽略现实中正在发生的事情，要么只能不断退回到防御性的解释之中。

因此，在这本书中，我选择了一种方法上的克制——暂时把认知主体放在一边。这并不是对人的否定，也不是对既有哲学传统的不敬，而是一种必要的清场。通过这种清场，我希望考察的是，当我们不再预设“必须有人理解”，知识是否仍然可以成立，判断是否仍然具有约束力，世界是否依然能够被描述、预测并嵌入行动。

在这个意义上，《认知主体之后》并不是一部关于意识或内在体验的哲学，而是一部关于有效性、结构与可运行性的哲学。它试图把知识从心理状态中松动出来，把认识从主体的内在视角中释放出来，使之重新成为一种可以被系统生成、维持并运作的东西。理解在这里并未被否认，而是被放回到一个不再具有特权的位置上。

我并不认为这种转向会使哲学变得贫乏。恰恰相反，当哲学不再承担为主体提供最终安置的任务时，它才第一次被迫正面面对现实结构本身。错误从何而来，责任如何分配，风险如何形成，这些问题不再可以简单地归结为某个主体的认知状态，而必须被放回到系统、流程与嵌入方式之中加以思考。

《认知主体之后》因此并不是一部反传统、反人类的作品。它并不需要通过反复援引这些哲学家的权威来证明自身的正当性，也无意取代任何既有的哲学框架。它只是试图在这些思想已经开启的方向上继续向前——在休谟动摇、康德重建、而蒯因、塞拉斯与波普尔已经松动主体中心的位置上，将这一松动推向一个在人工智能时代不再可以回避的结论。当人类不再是认识的唯一中心，当理解不再是知识的必要条件，我们或许才能更清楚地理解，在一个并不等待我们理解的世界里，哲学仍然能够、也必须继续发挥什么样的作用。

I wrote After the Knowing Subject not because I have lost patience with existing philosophical traditions, nor because I wish to declare any school of philosophy obsolete. On the contrary, it is precisely because I have long been immersed in these traditions—from 大卫 · 休谟’s radical destabilization of the foundations of knowledge, through Immanuel Kant’s reconstruction of the knowing subject within critical philosophy, and onward to the various twentieth-century efforts to detach knowledge from psychological certainty—that I gradually came to realize that, in an age where artificial intelligence has deeply entered the operation of reality, the problems we face no longer fall entirely within the original forms of questioning established by these frameworks.

As an independent researcher, I do not seek to found a new school in opposition to existing philosophical systems. What concerns me instead is a more fundamental re-examination of a prior question—what philosophy is doing in the age of artificial intelligence, and how philosophical inquiry itself should continue. This line of questioning does not negate tradition; on the contrary, it must be embedded within existing philosophical frameworks, taking them as background, reference, and interlocutor. Without this, any new attempt would lose its coordinates.

If we look back at the trajectory of modern philosophy, the problem had in fact already begun to emerge. Hume had perceived with great acuity that knowledge does not naturally take root in rational subjectivity, but depends instead on habit, association, and the stability of experience. Kant intervened precisely at this rupture, re-establishing the centrality of the knowing subject through an analysis of transcendental conditions, thereby “saving” knowledge. In the twentieth century, as reflection deepened on distinctions such as analytic and synthetic, inner and outer, psychological fact and normative reason, knowledge increasingly came to be understood as a structural entity embedded within an overall system—something that could be discussed, tested, revised, and to a significant extent exist independently of individual psychological conviction.

These lines of thought do not negate one another. Together, they form the field of tension that defines modern epistemology. Yet prior to the emergence of artificial intelligence, one assumption was never truly loosened—whether in empiricism, critical philosophy, or holistic and objectivist approaches, the knowing subject was always tacitly presumed to be the ultimate bearer of knowledge, and understanding was always treated as an implicit condition for knowledge to count as such.

It was against this background that I began to see how a long-standing but rarely isolated assumption had quietly become a bottleneck for contemporary thought. Nearly all major philosophical systems, in one way or another, presuppose the subject’s understanding as the core condition of knowledge. How the world is understood, how knowledge is constituted, how judgments gain legitimacy—these questions have been discussed repeatedly, yet rarely has the following been asked: if understanding itself is temporarily absent, does cognition still occur?

In earlier philosophical contexts, this question might have seemed abstract. Today, however, when artificial intelligence systems are able to generate stable judgments, guide action, and produce real consequences in the absence of understanding, it is no longer a theoretical supposition but an everyday fact. Models can continue to predict successfully without anyone genuinely understanding their internal mechanisms; systems can make effective decisions without anyone being able to explain their inferential paths; institutions can continue to operate even when their meaning is no longer consciously endorsed. Faced with these realities, if philosophy continues to insist on treating subjective understanding as a necessary condition of all knowledge, it must either ignore what is taking place or retreat into increasingly defensive explanations.

For this reason, in this book I adopt a form of methodological restraint—I temporarily set the knowing subject aside. This is neither a negation of the human nor a gesture of disrespect toward philosophical tradition, but a necessary clearing of the ground. Through this clearing, I seek to examine whether knowledge can still stand, whether judgments can still exert constraint, and whether the world can still be described, predicted, and embedded in action once we no longer presuppose that understanding must be present.

In this sense, After the Knowing Subject is not a philosophy of consciousness or inner experience, but a philosophy of effectiveness, structure, and operability. It attempts to loosen knowledge from psychological states and to release cognition from the subject’s internal point of view, allowing it to reappear as something that can be generated, maintained, and operated by systems. Understanding is not denied here, but relocated to a position that no longer enjoys epistemic privilege.

I do not believe that this shift impoverishes philosophy. On the contrary, when philosophy ceases to take responsibility for providing a final resting place for the subject, it is for the first time compelled to confront the structures of reality themselves. Where errors arise, how responsibility is distributed, how risks are formed—these questions can no longer be reduced to the cognitive states of particular subjects, but must be returned to systems, processes, and modes of embedding.

After the Knowing Subject is therefore neither anti-traditional nor anti-human. It does not need to rely on repeated appeals to philosophical authorities to establish its legitimacy, nor does it seek to replace any existing framework. It merely attempts, while respecting the historical background and problem-consciousness those frameworks provide, to take one step further—at the point where 大卫 · 休谟 unsettled, Immanuel Kant rebuilt, and where Willard Van Orman Quine, Wilfrid Sellars, and Karl Popper had already loosened the subject’s centrality. It is to carry that loosening toward a conclusion that can no longer be avoided in the age of artificial intelligence. When human beings are no longer the sole center of cognition, when understanding is no longer a necessary condition of knowledge, we may finally be able to see more clearly what role philosophy can—and must—continue to play in a world that does not wait for us to understand it.