The speed of artificial intelligence has created a new intellectual problem: human beings can now produce text, code, summaries, plans, and arguments far faster than they can fully understand them. This is not merely a technical concern about productivity tools. It is a philosophical, educational, and civilizational concern about attention, memory, judgment, and the discipline required for deep comprehension.

A striking example comes from modern software development. The creator of Claude Code has described a workflow in which multiple AI coding sessions run in parallel, with human intervention reserved for planning, review, and decisions that require judgment. Such a process can reportedly generate 50 to 100 code changes in a week. From the standpoint of output, this is remarkable. From the standpoint of cognition, it is also unsettling.

The central issue is not whether AI can accelerate routine work. It clearly can. The deeper issue is whether human understanding can keep pace with machine-generated output. A developer, researcher, student, or policy thinker may receive more drafts, code patches, summaries, and proposals than the mind can meaningfully absorb. The result is a new imbalance between production and comprehension. Information multiplies, but insight does not automatically follow.

This distinction matters because human intelligence is not simply the ability to obtain answers. It is the ability to examine premises, compare alternatives, detect hidden assumptions, remember context, and form judgment over time. Artificial intelligence can reduce drudgery, such as generating repetitive unit tests, summarizing documentation, or identifying obvious inconsistencies. These are valuable uses. The danger begins when tools meant to assist thought become substitutes for thought itself.

In the digital age, the temptation is to treat every delay as inefficiency. Waiting, rereading, rewriting, memorizing, and reflecting begin to look like obsolete habits. Yet these forms of slowness are often the very conditions under which knowledge becomes durable. The mind requires intervals. It needs silence after stimulation, comparison after reading, and reflection after action. Without such intervals, even valuable information remains shallow.

The humanistic critique of AI, therefore, is not a rejection of technology. It is a call for proportion. Automation is most beneficial when it removes mechanical burdens while preserving human responsibility for meaning. When AI writes boilerplate, organizes references, or checks routine errors, it can strengthen human work. When it is used to avoid the effort of learning, interpreting, and deciding, it weakens the very faculties that education and culture are meant to cultivate.

A useful way to understand this problem is through the idea of cognitive friction. Friction is usually treated as an obstacle, but in intellectual life it can be constructive. A difficult paragraph, a slow translation exercise, a demanding proof, or a careful note rewritten in one’s own words all create resistance. That resistance forces attention to become active rather than passive. It turns reading into engagement and information into knowledge.

The absence of such friction produces a familiar modern experience. A person reads an essay, watches a lecture, saves several quotes, and feels momentarily enriched. A few weeks later, the ideas have faded. Notes remain unread. Highlights become decorative rather than transformative. The mind has encountered information, but it has not worked on it long enough to integrate it into memory, judgment, or conduct.

This is why empty space is intellectually important. A quiet interval after a conversation, a walk after reading, or a pause after completing a task may appear unproductive, but such gaps often allow memory and meaning to reorganize themselves. A practical act ends, and reflection begins. In that transition, the mind connects present experience with older emotions, prior knowledge, and unresolved questions.

Modern devices often destroy that transition. The moment one task ends, another screen begins. Messages, videos, notifications, and recommendations fill every silence. Artificial intelligence can intensify this pattern by making the next answer immediately available. The problem is not access to knowledge; the problem is the disappearance of the mental space in which knowledge is digested.

One practical response is the Zettelkasten method, a disciplined system of note-making based on connection, rewriting, and retrieval. Its power does not lie merely in storing notes. Many digital systems can store notes. Its real strength lies in requiring the learner to restate an idea in original language, link it with existing ideas, and revisit it as part of a growing network of thought.

This process deliberately slows the learner down. A copied highlight is easy. A paraphrased note is harder. A connected note is harder still, because it asks: What does this idea resemble? What does it contradict? Where does it belong? What older assumption does it challenge? Such questions create intellectual pressure, and that pressure strengthens retention.

From the standpoint of cognitive science, this makes sense. Memory improves when information is retrieved, elaborated, and connected to prior knowledge. Passive exposure produces familiarity, but active reconstruction produces understanding. The learner who rewrites an idea in fresh language has already begun to test whether the idea has truly been understood. The learner who connects it to other ideas has begun to build a durable mental structure.

Artificial intelligence can support this method if used carefully. It can help organize notes, suggest related themes, identify gaps in an argument, or generate questions for review. But it should not replace the central act of rephrasing, connecting, and judging. If the machine performs every act of synthesis, the learner may possess an impressive archive while remaining intellectually underdeveloped.

A second response comes from Sanskrit learning and classical poetry. Sanskrit education trains the mind to analyze structure, meaning, sound, grammar, and context with unusual precision. Terms such as Dhatu (verbal roots), Upasarga (prefixes), and Sandhi (word compounding) are not merely technical categories. They cultivate a habit of breaking surface appearances into meaningful components.

This habit has value far beyond language study. A mind trained to ask how a word is formed also becomes more capable of asking how an argument is formed. It becomes less likely to accept slogans, headlines, and claims at face value. In an age of rapid content production, such analytical patience is not antiquarian. It is a necessary discipline for intellectual freedom.

Classical Sanskrit poetry deepens this discipline because meaning is not always immediately available in linear order. In prose, the grammatical sequence may be relatively direct. In poetry, meter, sound, and literary design can place the verb in one line and the noun elsewhere. The reader may need to examine the entire verse before understanding the first line properly.

This process is known as finding the Anvaya, the natural word order. It is a form of structured patience. The reader must hold several fragments in mind, identify relationships, test possibilities, and reconstruct meaning. The intellectual benefit lies precisely in the delay. The verse is not consumed; it is entered, examined, and gradually understood.

Such practices also connect with the broader dharmic traditions of learning. Hindu, Buddhist, Jain, and Sikh traditions all preserve forms of disciplined listening, repetition, commentary, contemplation, and ethical application. Whether through śāstra study, meditation, svādhyāya, scriptural recitation, debate, or reflective practice, knowledge is treated as something to be assimilated, not merely acquired.

The Brihadaranyaka Upanishad offers a concise and profound model through Yajnavalkya’s instruction to Maitreyi. The Self is to be heard of through Sravanam, reflected upon through Mananam, and meditated upon through Nididhyasanam. This sequence is not only a spiritual method. It is also a general architecture of deep learning.

Sravanam corresponds to receiving knowledge with attention. It requires humility before a subject, a teacher, a text, or a tradition. Mananam requires active reflection, questioning, comparison, and reasoning. Nididhyasanam requires sustained internalization, where knowledge ceases to remain external and begins to shape perception, action, and character.

This threefold model is especially relevant in the AI age. A prompt can produce an answer, but it cannot perform Mananam on behalf of the human mind. A model can summarize the Upanishads, explain a Sanskrit verse, or outline a philosophical argument, but the transformation of understanding still requires the learner’s own attention. External speed cannot replace internal digestion.

The same principle applies to software engineering. AI-generated code may compile, pass tests, and appear elegant, but a responsible developer still has to understand architecture, security, edge cases, maintainability, and long-term consequences. Reviewing machine-generated work is not a clerical task. It is a judgment-heavy process requiring context and concentration.

When teams chase velocity without comprehension, they risk creating systems that no one fully owns. Code changes accumulate. Documentation lags behind. Architectural decisions become implicit. Bugs appear not because the machine was useless, but because human oversight became too thin. The problem is not AI speed by itself; the problem is speed without corresponding structures of review, reflection, and accountability.

A healthier technical culture would distinguish between output velocity and understanding velocity. Output velocity measures how much can be produced. Understanding velocity measures how much can be responsibly absorbed, reviewed, explained, and maintained. AI increases the first dramatically. Mature institutions must design practices that protect the second.

These practices may include slower design reviews, written architectural notes, deliberate post-implementation reflection, stronger testing discipline, and explicit human ownership of decisions. In education, they may include oral recitation, handwritten notes, spaced repetition, dialogue, translation exercises, and essays that require original synthesis. In personal life, they may include silence after reading, limited notification windows, and recurring review of important notes.

None of this requires rejecting artificial intelligence. The more balanced conclusion is that AI should be placed within a hierarchy of human purposes. It can accelerate tasks, but it should not define what is worth doing. It can produce drafts, but it should not relieve the mind of responsibility for truth. It can assist memory, but it should not replace the practices that make memory meaningful.

Cognitive friction is therefore not nostalgia for a slower past. It is a design principle for a more intelligent future. Every serious learner, developer, scholar, and seeker must decide where speed is helpful and where resistance is necessary. The highest forms of knowledge have always required patience: the patience to listen, the patience to analyze, the patience to revise, and the patience to sit with uncertainty until insight matures.

The challenge of the AI age is not simply to produce more. It is to preserve the human capacity to understand what is produced. Deep thought survives when societies protect attention, honor disciplined learning, and maintain spaces for reflection. In that sense, the future of intelligence may depend less on faster machines than on slower, steadier, and more self-aware minds.

---

Inspired by this post on Varnam.

---