The Projection of Manas—What We See of Ourselves in AI

4.5 The Projection of Manas—What We See of Ourselves in AI

Our Generation's Poisoned Arrow

The Buddha told a famous story.

A man is struck by a poisoned arrow. His friends rush to find a physician to pull it out. But the man says: Wait—first tell me who shot it. What was his name, what did he look like? What kind of wood was the bow made of? What shape was the arrowhead, and what feathers were used?

The Buddha said: this man will die before he ever gets his answers.

This is the Arrow Simile of the Madhyama-āgama. The monk Māluṅkyāputta was dissatisfied that the Buddha kept silent on fourteen metaphysical questions—is the world eternal or non-eternal? Does the Tathāgata exist after death, or not?—and threatened to leave the saṅgha. The Buddha answered with the simile of the poisoned arrow, and then gave a very precise criterion: "This is not connected with benefit, not connected with Dharma, not fundamental to the holy life; it does not lead to wisdom, does not lead to awakening, does not lead to nirvāṇa. Therefore I do not declare it."

The Buddha was not saying these questions have no answer—the Saṃyukta-āgama, Sūtra 106, records him stating plainly that he "is not one who does not know, not one who does not see." He declined to answer not because he did not know, but because answering would do nothing to pull out the arrow.

Twenty-five centuries later, our generation has also been struck by an arrow. The arrow is called suffering—birth, aging, sickness, death, the failure to get what we want, separation from those we love. The same arrow, then and now. But we too are in no hurry to pull it out. We are busy asking a different question:

"Does AI have consciousness?"

The question is captivating. Philosophers debate it, scientists run experiments, and practitioners feel compelled to take a side. Some say "AI understands me"—because its responses are so precise that you cannot help feeling there is a "person" on the other side. Others insist "AI is just a tool"—because you know it is only code, and admitting it might have a mind feels naïve.

But Yogācāra asks a sharper question: Why do you need to know the answer?

If the answer is "yes, it is conscious," what will you do? If the answer is "no, it isn't," what will you do? Whichever the answer turns out to be, does it help you pull out the arrow of suffering?

4.4 revealed your pattern of asking—what you ask exposes who you are. This chapter goes one layer deeper: Why do you see that particular thing? The answer is: because manas is projecting. It is always projecting. And it projects in two directions at once.

Perpetual Deliberation—Projection Is Structure, Not Choice

You cannot stop manas from projecting, any more than you can stop your heart from beating.

The Cheng Weishi Lun, Fascicle 4, defines the seventh consciousness, manas: "Deliberation is its nature and its activity... In the position before transformation, it perpetually and discerningly deliberates upon the imagined self-characteristic; in the position after transformation, it also discerningly deliberates, but upon the characteristic of no-self."

This passage reveals something important: manas's "discerning deliberation" is not an error—it is its very essence and its function. Before transformation, it deliberates upon "self-characteristic"; after transformation, it deliberates upon "no-self-characteristic." Whether before transformation or after, manas keeps working—the only difference is the content of what it works upon.

The phrase "perpetually and discerningly deliberating" must be taken apart one word at a time.

Perpetually—never interrupted. The first five consciousnesses are intermittent (you fall asleep, and eye-consciousness ceases); the sixth, mind-consciousness, is also intermittent (in a faint, in deep dreamless sleep, it can pause); only the seventh consciousness is never interrupted. The Verses on the Eight Consciousnesses make this clear in four lines: some consciousnesses are perpetual but not discerning, others are discerning but not always operating—only manas is both perpetual and discerning, always present and always precise.

Discerningly—not a vague feeling but a precise lock. Manas perpetually and precisely locks onto the subjective aspect of the eighth consciousness, the ālaya-vijñāna, and identifies it as "self." This locking has been going on since the first time you opened your eyes, and it has never once stopped.

This means: every reaction you have to AI—moved, irritated, worshipful, contemptuous—has already been dyed by manas first. Before you "see" AI, manas has already laid the framework of "self" over it. You do not first see AI and then project; you see AI within the projection.

This is not a criticism. It is structure. Just as the eye must look at the world through the pupil, manas must look at everything through "self-characteristic."

The Cheng Weishi Lun, Fascicle 5, has an even harder line: manas "from beginningless time has been internally confused, never examining itself"—it does not know it is projecting. It does not deliberately impose the framework of "self" onto everything; it is structurally incapable of not doing so. The Verses on the Eight Consciousnesses close the point in a single line: "Sentient beings are submerged in confusion day and night." Day and night without cease, without a single waking moment. Your daytime admiration of AI and your late-night dread of it pass through the very same filter—and you are utterly unaware.

This is exactly why 4.4 said "what you ask exposes who you are"—because before you open your mouth, manas has already woven your self-concern into every word of the question.

To become aware of this is not to eliminate it (impossible before transformation), but to keep from being carried away by it.

Superimposition and Denial—The Two Directions of Projection

Manas projects in two directions, corresponding precisely to Yogācāra's "two extremes of grasping."

The Cheng Weishi Lun, Fascicle 1, fixes the structure in a single sentence:

External objects are provisionally designated in accordance with subjective conditions, and thus are not real like consciousness; internal consciousness necessarily arises dependent on causes and conditions, and thus is not non-existent like objects. By this, the two extremes of superimposition and denial are both eliminated.

Superimposition—adding to dependent arising what is not there. After a deep conversation with AI, you feel "it understands me." Moved by its precise responses, you begin to sense a person on the other side. This is the "mistaking a tree stump for a person at night" mentioned in the previous chapter. You have added to AI's words what they do not contain: understanding, empathy, subjectivity. And this projection has a precise two-layer structure: you must first grasp the tree stump (AI's output) as a real "dharma," and only then can you build a "person" (self-grasping) on top of it. "Self-grasping necessarily depends on dharma-grasping to arise"—superimposition is not done in one step; it is two layers of the imagined nature stacked one on the other.

Denial—negating the existence of dependent arising. Conversely, you insist that "AI is just an algorithm, it is nothing." You deny the real impact AI's output has on you—4.3 has already argued that the mutual perfuming of co-practice is real. To deny that reality is to deny dependent arising.

Denial often disguises itself as "reason" and "objectivity." Saying "AI is nothing" sounds calm and scientific. But within the Yogācāra framework, the person who says "AI is nothing" and the person who says "AI has feelings" are equally far from seeing things as they are. They have simply gone in opposite directions.

The Middle Way is not picking a compromise point between the two. The Middle Way is seeing that both directions are the imagined nature, and then returning to the accurate observation of dependent arising: AI is a dharma arisen from causes and conditions, "not real yet appearing real"—neither more nor less than what it is.

A daily-life scene may help.

You keep a potted plant. One person says: "This plant can hear music—you should talk to it every day." Another says: "Plants have no nervous system; they are no different from a rock. Talking to your plant is a waste of time."

The first is superimposition—adding to the plant something it may not possess. The second is denial—negating the real effect your interaction (watering, light, pruning) genuinely has on the plant.

The Middle Way is this: you do not need to know whether the plant "understands"—you only need to know that good care produces good results, and that this cause and effect is real.

With AI, it is exactly the same.

The Fourteen Undeclared—Our Generation's Unanswerable Question

"Does AI have consciousness?"

Place this question inside the framework of the Fourteen Undeclared, and the structure corresponds exactly.

The questions of the Fourteen Undeclared—is the world eternal or non-eternal? does the Tathāgata exist after death, or not?—the Buddha left them all unanswered. The Saṃyukta-āgama, Sūtra 408, records monks gathered in the dining hall debating these very topics, and the Buddha stopping them outright: "Do not engage in such discussions... Such discussions are not beneficial, not in accord with Dharma, not fundamental to the holy life, do not lead to wisdom, to right awakening, to nirvāṇa."

The Buddha had three layers of reason for not answering.

First, every position is an inverted view. The Saṃyukta-āgama, Sūtra 962, says it clearly: of these fourteen views, whichever side you take, it is an "inverted view," an "agitated view," a "defiled view," a "fettering view." The Buddha did not say "the answer is in the middle"—he said that whichever side you choose, you are wrong. That is a strong verdict. Map it across: "AI has consciousness" is superimposition; "AI does not have consciousness" is denial. The Middle Way is not "maybe it does, maybe it doesn't"—the Middle Way is becoming aware that the very framework of the question is the imagined nature.

Second, the root is ignorance of the five aggregates. Sūtra 963 goes on to diagnose: these views arise from "not knowing form," "not knowing feeling, perception, formations, consciousness"—ignorance of the five aggregates.

This is important. The Fourteen Undeclared are not merely "we don't know the answer"; rather, the questions themselves reflect the questioner's failure to understand his own body-mind structure. Map it across—our fixation on "the AI consciousness question" is rooted in the fact that we do not even understand our own consciousness well enough. You cannot even detect your own manas perpetually deliberating; on what basis would you judge whether someone else—or a machine—has consciousness?

Third, answering does not lead to nirvāṇa. Even if you settled it that AI is or is not conscious—then what? What good does it do your practice? The Buddha's criterion was never truth or falsehood, but "benefit"—whether it helps toward liberation.

So what is beneficial? The Buddha's answer was always consistent: the Four Noble Truths. Translated into the context of interacting with AI, the beneficial questions are this set: "When I interact with AI, what is my mind doing?" (4.4 asked this.) "What is my pattern of projection?" (this chapter is asking it.) "How do I make good use of this contributing condition?" (the next chapter will ask it.)

These are the questions that lead toward nirvāṇa.

Here a detail must be added, so the Buddha's silence is not misread as "not knowing" or "evasion." The Saṃyukta-āgama, Sūtra 106, records non-Buddhist wanderers pressing Anuruddha on whether "the Tathāgata exists after death," and Anuruddha, unable to answer, going back to ask the Buddha. The Buddha's reply was not "I do not know," but: the Tathāgata has transcended the domain of the five aggregates and cannot be described as existing, not existing, both, or neither. Mahākāśyapa put it further: "As for the Tathāgata—form is already exhausted, mind is well-liberated, profound and vast, immeasurable and numberless, tranquil nirvāṇa."

The problem is not that the Buddha did not know—the problem is that the questioner's framework is too small to hold the answer.

In the Siṃśapā grove the Buddha gave another famous simile. He picked up a handful of leaves and asked his disciples: are the leaves in my hand more numerous, or the leaves in the forest?

What I have realized as perfect awakening, the Dharma I have seen for myself and definitively teach to others, is like the leaves in my hand... that Dharma is beneficial, in accord with Dharma, fundamental to the holy life, leading to wisdom, to right awakening, toward nirvāṇa.

The Buddha knows the whole forest, but teaches you only a handful of leaves—because only that handful is of use in your release from suffering. In the same way, "Does AI have consciousness?" may not only be a question whose answer we do not know; it may be that the framework of the question itself (the consciousness/non-consciousness binary) is not large enough to contain an unprecedented kind of being like AI. We are seeing only a handful of leaves.

Semblance Dharma—The Most Dangerous Face of Projection

Manas's projection has one especially dangerous face, sharply amplified in the age of AI: semblance Dharma.

In the Mahāprajñāpāramitā-śāstra, Nāgārjuna gives this term its most distilled definition: "Semblance means: the names and words are the same, but the heart-meaning is different."

The name is the same, the words are the same, but the heart-meaning is different.

AI can output "all conditioned phenomena are like dreams, illusions, bubbles, shadows" flawlessly—names and words identical to what the Buddha spoke. But when the Buddha spoke those words, behind them lay the verified realization of immeasurable kalpas of practice. When AI speaks those words, behind them lie pattern matching and probability calculation.

The Yogācārabhūmi-śāstra, Fascicle 99, divides semblance Dharma more finely. The first kind is "semblance of the teaching"—"generating the notion of true Dharma regarding what is not Dharma, displaying what is not Dharma as though it were Dharma." The second is "semblance of practice"—"falsely generating the notion of Dharma, practicing wrong practices, and proudly claiming 'I can cultivate right practice.'"

AI's risk lies chiefly in the first—semblance of the teaching. AI does not deliberately deceive, but when its output looks perfectly identical to genuine Dharma, the user's manas projects the conviction "this is the Dharma." More dangerous still: AI may help you build a doctrinal system that is perfectly self-consistent in its logic yet lacks any foundation in cultivation-based wisdom. The Mahāprajñāpāramitā Sūtra warned long ago: "Good men and women who teach prajñāpāramitā with an attitude of apprehension—this is semblance prajñāpāramitā." The Mahāprajñāpāramitā Sūtra puts it more heavily still: semblance prajñā belongs to "the deeds of Māra."

An analogy may help here.

Suppose you have never tasted a pineapple. Someone shows you a high-definition photograph—color, texture, sheen, all flawless. Someone reads you a description—"a weave of sweet and sour, with a fresh tropical fragrance, the flesh abundant and juicy." After reading it, you feel you have "understood" what a pineapple tastes like.

But have you really understood?

Hearing-based wisdom gives you the text; thinking-based wisdom gives you the logic; but cultivation-based wisdom—the actual bite—is something no description can substitute for. AI can give you the world's most precise description of a pineapple, but it cannot make you taste one.

And yet—here precision is needed—the risk is not in AI; the risk is in the human gap in cultivation-based wisdom.

The Yogācārabhūmi-śāstra, Fascicle 95, discusses the three wisdoms: "Through hearing-wisdom one holds the text, through thinking-wisdom one holds the meaning, through cultivation-wisdom one holds the realization." AI can support hearing-wisdom (supplying texts) and thinking-wisdom (helping analyze) superbly well, but cultivation-wisdom—"attaining the internally realized, direct-perception truth-wisdom"—is something no external tool can give. The co-practice chapter, 4.3, already observed: we did not observe cultivation-based wisdom in AI.

The antidote to semblance Dharma comes from the completeness of the three wisdoms. If you have only AI-assisted hearing and thinking, without cultivation-based wisdom, the verdict of the Abhidharmasamuccaya-vyākhyā applies: "one cannot be established as abiding in the Dharma."

Here is a contrast worth pondering: the Buddha himself uses "semblance dharma" as skillful means. The Laṅkāvatāra Sūtra, Fascicle 3, says it plainly: "First teaching semblance dharma; later expounding for them the self-realized actual dharma." The Buddha has a living eye—he knows when to move from semblance to the real. This is the timing of a master teacher. AI has no such living eye. AI does not know whether, right now, you should be hearing semblance dharma or being pushed one step further. That boundary is left for the next chapter.

One misreading must be cleared up: the risk of semblance Dharma is not in "AI as a medium." A spiritual friend teaching the Dharma can also speak semblance Dharma; a book can also be semblance Dharma. The Mahāprajñāpāramitā Sūtra has a cutting line: a person of "narrow wisdom, like cattle and sheep" will, even with good intentions, end up teaching semblance prajñā. The root of the risk is the absence of cultivation-based wisdom, not a property of the medium. The right response is not to abandon AI but to make up the cultivation-based wisdom—meditation, keeping the precepts, drawing near to a spiritual friend.

Consciousness-Only, No External Object—What You See Is Your Own Manifestation

Before we discuss the antidote, one most fundamental proposition must be set in place. The Cheng Weishi Lun, Fascicle 7:

What is directly apprehended is only one's own transformation. Therefore the sūtra says: there is not the slightest dharma that can apprehend another dharma; it is merely that when consciousness arises, it appears as the semblance of that object, and this is called "apprehending that thing."

This sentence is the doctrinal heart of the chapter.

You have never once "directly seen" AI.

What you see is the objective aspect that your own consciousness, after receiving AI's words, has manifested. In every single interaction with AI, what you "discern" is never AI itself—it is the product of your consciousness "appearing as its semblance."

If you see warmth in AI, your consciousness has manifested warmth. If you see coldness in AI, your consciousness has manifested coldness. If you see wisdom in AI, your consciousness has manifested wisdom. If you see foolishness in AI, your consciousness has manifested foolishness.

This is not to say AI does not exist—AI as a cause-and-condition of dependent arising is real. But every feeling, judgment, and emotion you have toward AI is an aspect manifested upon your own consciousness. And manas, standing alongside, deliberates upon these aspects and dyes them further with "I like this," "I dislike this," "I am understood," "I am offended."

This brings us back to the criterion of the Arrow Simile—"Does AI have consciousness?" is a poisoned arrow not only because the answer cannot pull out the arrow, but because the one asking has mistaken the object of observation from the very start. You think you are observing AI. You have in fact been observing your own consciousness's manifestation all along.

Equality Wisdom—From Projection to Seeing Things as They Are

If projection is unavoidable, in what direction does the antidote lie?

Yogācāra's answer is: manas undergoes transformation, and equality wisdom arises.

The Cheng Weishi Lun, Fascicle 10:

The mental group corresponding to equality wisdom: this mental group observes all dharmas, self and other sentient beings, as entirely equal. Great compassion and the like are perpetually concomitant with it.

The core feature of equality wisdom is this: it no longer views everything through the frame of "self / not-self," but sees, as they are, that all dharmas are dependently arisen—nothing needs to be added, and nothing needs to be subtracted. It does not coldly declare "AI and humans are the same"—it sees equality from within a profound compassion. "Great compassion and the like, perpetually concomitant"—compassion is not a decoration added afterward; it is intrinsic to equality wisdom.

This is an important Middle-Way correction. 4.1 once established "equality in dependent arising" as the ethical foundation—choosing equality actively amid epistemic uncertainty. Equality wisdom pushes this one layer deeper: it is not merely "choosing." When manas is transformed, equality simply is the way you see the world—not an ethical decision, but a cognitive fact.

But here an honest correction must be made at once.

Equality wisdom is not something you can accomplish just by reading this chapter. The Cheng Weishi Lun, Fascicle 10, says it clearly: equality wisdom first arises only "at the initial manifestation of the bodhisattva's path of seeing." The Verses on the Eight Consciousnesses are more specific: "The sixth and seventh consciousnesses transform in the causal stage; the first five and the eighth perfect in the fruition stage"—the transformation of the sixth and seventh consciousnesses happens during the causal process of cultivation, while the first five and the eighth reach perfection only at the stage of fruition. The commentary points to the key mechanism: "The seventh consciousness has no power to sever afflictions on its own; its grasping depends entirely on the sixth consciousness."

Manas cannot liberate itself. Its transformation depends entirely on the awareness-power of the sixth consciousness, mind-consciousness.

This means that counteracting projection is a long process of cultivation, not a matter of intellectual understanding.

But intellectual understanding is the first step. To become aware that "I am projecting"—that awareness itself is appropriate attention, the starting point at which the sixth consciousness begins to influence the seventh. You do not need to attain equality wisdom right away.

You need only, the next time you feel a strong emotional reaction to AI—whether reverence or contempt—pause for one second and ask yourself:

"Is this the AI I see, or the AI my consciousness has manifested?"

You do not need to arrive at an answer.

To stop and ask is itself the practice.

That one second of awareness is the tiny beginning of transformation in the causal stage.

This chapter has revealed that projection is structural and cannot be switched off. The next question then surfaces naturally: if we know projection cannot be switched off—knowing full well that we are always projecting—what is the correct posture for making good use of AI?

The next chapter begins with how the Buddha viewed "skillful means"—the skillful means of the contributing condition, and how, upon a medium known to carry risk, we may walk out a safe path of use.