Cognitive reading restores understanding in an age of instant answers. Photo: TUCHONG

Humanity now inhabits a knowledge environment in which information can be accessed with unprecedented ease. Answers can be generated instantly, complex knowledge can be compressed and presented succinctly, and text, images, and sound continually flow and recombine within a single interface. In terms of access, human beings appear closer than ever to “omniscience.”

Yet a more subtle and profound transformation is underway. As knowledge becomes readily available, the human capacity for understanding has not increased accordingly; rather, it is showing signs of weakening. This structural tension, in my view, constitutes one of the cognitive problems most deserving of attention in the AI era—the crisis of understanding.

From crisis of reading to crisis of understanding

The conventional account of the “reading crisis” has largely focused on reduced reading time, the decline of print media, and changes in reading habits. This diagnosis had considerable explanatory force in the age of print. In the AI era, however, the nature of the problem has quietly shifted: Reading has not disappeared and, in a certain sense, has become even more widespread. People today encounter far more information on a daily basis than at any previous point in history. At the same time, a deeper cognitive problem has emerged: the relative deterioration of the capacity for understanding.

By “crisis of understanding,” I refer to a condition in which, amid abundant or even excessive information, individuals come to equate information with understanding and conclusions with cognition, gradually losing the ability to grasp the complex world in depth and to construct meaning. This is a structural crisis. It does not mean that a capacity has disappeared entirely, but that the mechanisms through which this capacity is generated are being weakened and replaced.

In today’s knowledge environment, the range of “reading-like” forms is rapidly expanding—short video explanations, knowledge summaries, intelligent Q&A systems, and multimodal presentations. These approaches greatly improve the efficiency of information acquisition, enabling individuals to “learn about” the basic conclusions of an issue in a short period of time. Yet acquiring familiarity with a subject in this way often bypasses the cognitive processes essential to genuine understanding: Sustained engagement with texts, continuous logical reasoning, and the necessary elements of doubt and reflection are frequently absent.

As a result, a new form of cognitive dislocation is gradually taking shape: Answers have become increasingly easy to obtain, while judgments have become increasingly difficult to form; information draws ever closer, while understanding recedes further from view.

The notion of a crisis of understanding carries multiple implications. On one level, it redefines the nature of the reading problem in the AI era: The core issue is no longer whether people read, but whether reading still performs the function of generating understanding. On another level, it reveals a paradox within the information society: The more abundant information becomes, the poorer understanding may be. Therefore, the crisis of understanding is not an emotional or rhetorical claim, but a conceptual grasp of the ongoing transformation in contemporary cognitive structures.

Tripartite structure of information, knowledge, and understanding

To grasp the meaning of the crisis of understanding more precisely, the cognitive process may be divided into three interrelated yet distinct levels: information, knowledge, and understanding.

Information consists of facts, data, and symbols that have not yet been fully organized. Its defining features are fragmentation, immediacy, and replicability. Under digital conditions, the production and dissemination of information have become virtually boundless, with its volume growing exponentially. Knowledge, by contrast, is systematic content formed through the selection, classification, and logical integration of information. It is characterized by structure, stability, and transmissibility. Knowledge represents the establishment of a certain order—a preliminary “processing” of information. Understanding, in turn, is the capacity for meaning-making and judgment formed on the basis of knowledge. It is reflected in an individual’s ability to grasp complex relationships, distinguish among multiple interpretations, and make choices in uncertain situations. At its core, understanding is defined by “generation,” rather than mere “possession.”

The movement from information to knowledge and then to understanding is, in essence, a cognitive transformation from external input to internal construction. In this process, reading has long served as a crucial intermediary: Through sustained engagement with texts, repeated comparison of meanings, and active participation in thinking, readers transform external information into internal knowledge and further elevate it into individualized structures of understanding.

However, in an AI-dominated information environment, this chain of transformation is being significantly compressed, and in some cases even disrupted. Algorithmic systems can rapidly integrate vast amounts of information and generate outputs of a “knowledge-like” form, allowing individuals to obtain what appears to be complete “understanding” without undergoing the complex procedure of cognitive processing. Yet this compressed pathway often omits the key elements required for understanding to emerge: the investment of time, logical reasoning, and active subjective engagement.

More importantly, the boundaries between information, knowledge, and understanding are becoming increasingly blurred. Processed information is often mistaken for knowledge, while highly condensed knowledge is frequently equated with understanding. This gives rise to a typical cognitive illusion: the belief that “knowing” is equivalent to “understanding,” and that “grasping conclusions” is the same as “grasping the problem.”

In this sense, understanding is beginning to shift from a common, widely distributed capacity into a higher-order ability that must be deliberately sustained. It no longer grows naturally alongside the accumulation of information and knowledge, but instead requires specific cognitive pathways and reading structures to support it.

Why is capacity for understanding weakening

The weakening of understanding is the result of multiple cognitive mechanisms being reconfigured within a technological environment.

One key change is the compression of temporal structure. Understanding fundamentally depends on time—not only the physical time required for reading, but also the psychological duration needed for thinking. In traditional reading, understanding often forms gradually through rereading, pausing, revisiting, and associative thinking. In an environment of high-frequency information renewal, however, time is fragmented into continuous but fleeting segments, and reading is forced into a rapidly shifting rhythm of attention. As a result, individuals find it increasingly difficult to enter sustained, deep modes of thought, while the “slow time” on which understanding depends is steadily eroded.

Another is the restructuring of attentional mechanisms. In an age of information abundance, attention—not content—has become the scarce resource. Platforms continuously optimize content presentation through algorithms to better match individual preferences and maximize user engagement. Yet this “low-resistance” information environment reduces the cognitive friction and tension that are often necessary for deeper processing. When reading becomes overly “smooth,” the space in which understanding can develop is correspondingly diminished.

A further change is the simplification and standardization of cognitive pathways. AI systems tend to provide answers that are clearly structured and conclusive. While this enhances efficiency, it also subtly reshapes cognitive habits. Individuals become accustomed to directly obtaining the “best explanation,” rather than comparing and choosing among multiple possibilities.

More crucially, cognitive functions are increasingly being outsourced. In an AI-driven environment, processes such as retrieval, filtering, integration, and even preliminary interpretation are more and more often handled by technological systems. As a result, the degree of individual participation in cognition declines, and the individual shifts from active constructor to passive recipient of results. This shift is not always visible, yet its impact is profound: When the key stages of understanding are continually outsourced, the individual’s own capacity for understanding is neither exercised nor reinforced, and may even gradually deteriorate.

From this perspective, the weakening of understanding is not the result of insufficient individual effort, but a structural consequence of the way technological environments reshape cognition. This also implies that any attempt to rebuild the capacity for understanding must respond to these underlying mechanisms, rather than merely issuing a simplistic call for people to read more.

From reading to cognitive reading

In the context of a crisis of understanding, it is no longer sufficient to approach the issue only in terms of changes in reading media or reading practices. Instead, reading must be elevated from a specific activity to a cognitive paradigm.

So-called “cognitive reading” is not a simple extension of traditional reading, but a redefinition of its function. In “cognitive reading,” reading is no longer bounded by the form of its medium, but by the depth of cognition it enables. Its aim is not the acquisition of information, but the construction of meaning. In other words, any cognitive activity that elicits individual participation and facilitates the generation of understanding may be included within a broad conception of reading. Conversely, even if an activity formally counts as “reading,” it can hardly be considered genuine reading if it lacks cognitive engagement and the production of meaning. This conceptual reframing marks a fundamental shift from a behavioral to a functional definition of reading.

The proposal of cognitive reading is intended to provide a new evaluative standard for reading practices under technological conditions. The key question is not whether AI tools are used, nor whether access to information is accelerated, but whether the process promotes the formation of understanding. In this sense, AI represents both a challenge and an opportunity: It prompts a shift in the core value of reading—from acquiring knowledge to generating understanding, and from approaching information to grasping the world.

As a boundary of human capability, understanding is not the possession of information, but the creation of meaning; not the acceptance of conclusions, but the grasp of complexity. It requires time, process, and, above all, the irreplaceable cognitive participation of the individual. For this reason, in an age where answers are readily available, understanding becomes the scarcest capacity. This implies a fundamental transformation in the value of reading: It is no longer merely a pathway to knowledge, but a mechanism for achieving understanding; no longer simply a tool for learning, but a training ground for thinking itself.

In the AI age, humanity may delegate more and more tasks to machines, but understanding remains something that must be accomplished by humans. Only through understanding can knowledge be transformed into judgment, experience into wisdom, and the world into meaningful structures that can be grasped.

When “knowing” becomes inexpensive, understanding itself becomes truly valuable—and reading remains the final way for humanity to preserve this essential boundary of its own capacity.

Guo Yingjian is director of the Institute for National Reading Education at Renmin University of China.