Episode Two Begins with the Question “What Do I Know?”
This is the second installment of the weekly Liberation and Civilization series. Before the main lecture, I explained that each episode requires research and a deliberate narrative structure. The central axis of this episode is the question I return to again and again: “What do I know?”
This is not primarily a question about some grand mystery outside us. It is a question about my own knowing. Is what I call knowledge actually true, or have I confused memory, hearsay, and inference with fact? In one word, this is skepticism.
I said that skepticism lies beneath philosophy, science, and Buddhism. Without this foundation, people speak nonsense without knowing whether their words are true. Buddhism begins only when we learn to examine whether we know what we think we know.
Philosophy, science, and Buddhism begin with the question “What do I know?”
Denying Rebirth Does Not Complete the Foundations of Buddhism
Before the formal lecture, I discussed Hyangbong and Professor Lee Jung-pyo. Both deny rebirth despite the many passages about rebirth and karma in Buddhist texts. I respect their willingness to see that rebirth conflicts with something central in Buddhism.
Yet one correct rejection does not complete awakening. Hyangbong speaks as if enlightenment enables an immediate answer to every Zen case, while Lee often treats the Nikayas he reads as early scripture and the Buddha’s own speech. Without examining what they know and do not know, many wrong statements can grow beside one useful insight.
I criticize even those who appear to be “on my side.” Respect for a correct point does not make every claim correct. The purpose of criticism is not hatred but to identify where metacognition remains absent.
The Assumption “The Buddha Said It” Must Be Examined First
I asked what stands behind the claim that an awakened Zen teacher can answer every koan immediately. Can one truly understand every question at once? Why should speed prove awakening? Perhaps a cultural image of the enlightened master has been accepted without reflection.
Likewise, before explaining why the Buddha spoke of the deathless, Professor Lee should ask a more basic question: did the Buddha actually say it? If a scriptural sentence is first declared authentic, a sophisticated interpretation may become no more than an elaborate explanation of an untested premise.
Those who claim awakening must apply “Do I really know this?” more rigorously, not less. Otherwise liberation from one belief can become captivity inside another.
One correct conclusion does not make every statement knowledge. Even the awakened require metacognition.
From Pyrrho’s Antiquity to AI Research in 2026
The first episode followed Pyrrho on Alexander’s campaign to Taxila, where he encountered Indian renunciants and pursued equanimity and an unshaken mind. I described him as someone who returned to Greece carrying something like a Buddhist path.
Episode two turns to a Google research report released in an online archive in May 2026 and then travels back to Socrates and sixteenth-century Montaigne. The periods differ, but the underlying question is the same: how can an intelligence recognize what it knows and does not know?
This is also a Buddhist question. What is liberation, and how do we escape delusion? AI hallucination becomes a contemporary mirror that reveals why metacognition belongs at the foundation of Buddhist practice.
Generative AI Plausibly Invents What It Does Not Know
An AI hallucination is an error in which a system presents false or nonexistent information as if it were true. It can invent academic papers, links, interviews, and historical events. The language sounds natural, making falsehood difficult to detect.
Early large language models developed as machines predicting what words and sentences plausibly follow an input. They were used to continue text adventures in which a user entered a room, met a character, or found treasure. They were generators of plausible continuation before they were treated as factual assistants.
When such a model must answer a question beyond its information, it may write fiction instead of remaining silent. In law, medicine, finance, or research, a single plausible fabrication can produce serious harm and destroy trust.
Hallucination Undermines Trust from Below
I unpacked the report’s language. To undermine is to dig beneath and weaken something. When AI occasionally gives a confident false answer, users begin to wonder whether any answer is safe to use.
The report presents metacognition as a way forward. Before answering, a system should assess whether it knows, whether its information is sufficient, and whether the proposed answer is fact or inference. It should admit ignorance and signal uncertainty.
I described this as a path or dao. The route to trust is not a promise of perfect answers but an honest assessment of how certain an answer is.
If hallucination undermines trust, metacognition rebuilds it by examining whether one knows.
Hallucination and Illusion Differ, but Humans Experience Both
Hallucination is perception without a corresponding external stimulus, as in seeing or hearing what is not there. Illusion begins with a stimulus that is then misinterpreted. Optical illusions and mishearing belong to this second family.
Magic exploits these limits of attention and vision. Something appears to vanish or suddenly emerge because perception is guided into an error. AI hallucination is not identical to human sensory experience, but human beings are vulnerable to both hallucination and illusion.
We see what was not there, hear words differently, remember altered versions, and remain certain that outdated information is true. AI error magnifies the fragility already present in human intelligence.
Humans Answer Without Knowing That They Are Wrong
I addressed viewers by name and asked whether any of them never misperceive or misunderstand. Among hundreds of viewers, almost nobody could honestly claim perfect separation of knowledge and ignorance.
Intelligence is developed inference. Like a student forced to select an answer to a multiple-choice question, a person often chooses the most plausible option from incomplete information. People, like language models, frequently produce an answer rather than admit that they do not know.
In a whispering game, a sentence changes as it passes through five or ten people. Each person believes that the message was repeated accurately, yet hearing, memory, and expectation alter it. Humans are animals of optical, auditory, and conceptual illusion.
Liberation Requires Admitting That My Answer May Also Be a Hallucination
If metacognition addresses AI hallucination, it must also address human delusion. What I confidently saw, heard, and concluded may be a plausible error resembling an AI’s fabricated answer.
Metacognition is not an exotic ability. When asked for the date of a historical battle, a person can inspect memory and say, “I do not know,” or “I vaguely remember, but I am uncertain.” This is already the mind observing itself from a slight distance.
The difference lies in how carefully and often the capacity is used. Before speaking, ask whether the information is current, whether the memory is accurate, and whether the reasoning has a gap. Without reconsideration, delusion undermines trust and blocks liberation.
My answer may be as plausibly wrong as an AI hallucination. Liberation begins when I admit that possibility.
The Premise “The Buddha Spoke of the Deathless” Comes First
Professor Lee may interpret “the deathless” in a subtle way that does not mean literal immortality. Yet I asked whether he was present to hear the Buddha speak. Nobody alive heard the historical Buddha’s voice.
What we possess is a sentence in scripture, commentary, narrative, or oral transmission. The first question is not “Why did he mean this?” but “Is this actually something the Buddha said?”
If authenticity is assumed before interpretation begins, an elegant explanation can become a beautiful reading of a false premise. Metacognition must examine the status of the source before defending an interpretation.
Why Reject Rebirth Passages but Trust the Passage About the Deathless?
Buddhist literature contains stories about the Buddha’s previous lives everywhere. Professor Lee rejects these rebirth passages because he sees them as inconsistent with the core of Buddhism.
Why, then, is the sentence “I attained the deathless” accepted as the Buddha’s speech simply because it appears in scripture? What criterion makes one passage late and the other authentic? Selective acceptance can disguise preference as scholarship.
The fact that a sentence is written in scripture is different from the fact that the historical Buddha spoke it. Without distinguishing the two, interpreters sell their own inference under the Buddha’s name.
There Is No Basis for Calling the Entire Nikaya “Early Scripture”
I strongly criticized treating the Nikayas as one uniform early canon. Textual strata exist inside the collections. Even the Sutta Nipata contains older verse alongside prose that may have been attached later.
Short verses such as parts of the Atthakavagga can represent an older layer, while extended narratives may be later constructions. Texts often expand as explanations, ornament, past-life tales, and interpretation accumulate. Pali language alone does not prove that every passage predates Mahayana literature or records the Buddha’s voice.
I directly urged Professor Lee to update his information. A scholar who ignores strata and explains all prose as early Buddhist teaching may keep spreading outdated assumptions as fact.
One book can contain old verses and later prose. The name “Nikaya” does not make every sentence early.
Compare the Length of the Chinese and Pali Milinda Texts
I used the Chinese Milinda Wangmun-gyeong and the Pali Milinda Panha as an example of textual growth. King Milinda lived well after the Buddha, and the work is not a record of the Buddha’s own speech. There is also no certainty that the king and Nagasena met as narrated.
The Chinese version is shorter, while the Pali version includes far more explanation and even past-life stories about the two figures. Rather than assuming Chinese translators removed all this material, it is plausible that the shorter form preceded later additions. Shorter never automatically means older, but expansion is an important comparative clue.
The famous attempt to reconcile not-self and rebirth through analogies becomes prominent in this later literature. The comparison disrupts the simple belief that Chinese texts are late while Pali texts are always early.
Information Ages, So Metacognition Must Include Updating Sources
Even careful reflection produces error if the input itself is outdated. I used the familiar “333” tooth-brushing rule as an example. Many learned to brush three times a day, within three minutes after eating, for three minutes. More recent advice may warn against brushing immediately while the mouth is acidic.
The point is not to replace one dental dogma with another. It is that information taught as correct in childhood can change with new observation. Faithful memory does not guarantee current validity.
The same applies to interpreting twelvefold dependent arising or the deathless as unquestionably spoken by the Buddha. Without new research and updated sources, an authority may confidently distribute information whose age and weakness remain invisible to him.
Professor Lee Can Be Wrong, and So Can I
Metacognition fails if criticism points only outward. I immediately turned the question toward myself. If Lee carries hallucination, illusion, and outdated information, does Yun Deok-man never do so? Can I speak for hours without error simply because I call myself an arahant?
No. I also make such errors. Every human misperceives, misremembers, and reasons from incomplete or outdated information. The claim of personal infallibility would itself be another delusion.
This is why we correct one another, exchange information, debate, and return to sources. Criticism need not destroy a person; it can be a shared metacognitive process for reducing delusion and restoring trust.
Professor Lee can be wrong, and so can I. Admitting our humanity makes dialogue a condition of liberation.
Certainty Can Block Both Dialogue and Liberation
I recalled arguing, “I definitely heard it,” only to discover later that I had misheard. Intensity of certainty is not accuracy of fact.
When someone believes his information is one hundred percent correct, conversation becomes impossible. New evidence is dismissed with “I saw it.” The chance of escaping delusion falls, because liberation means release from what one wrongly knows and clings to.
If we learned that gods, souls, afterlives, rebirth, karma, and cosmic retribution exist, we must still ask whether they do. Skeptical examination of inherited concepts is the beginning of Buddhism.
Metacognition Is Thought About Thought and Observation from a Distance
Metacognition is the ability to observe and regulate thought, knowledge, and action from a higher or third-person perspective. It asks whether I understand and whether my recent action was appropriate.
When completely immersed in resentment, it is difficult to ask why resentment is present. Observation requires a little distance from body, feeling, emotion, and thought. The aim is not to abolish emotion but to avoid treating immersion as the whole truth.
This connects directly to observing body, feeling, mind, and dhamma. We carefully notice how the body moves, feelings arise, the mind changes, and judgments about right and wrong form. Sati and mindful attention name this practice of self-observation.
Question, Observe, and Reflect: This Is Buddhist Practice
I repeated three words: question, observation, and reflection. Questions such as “How do I work? How do suffering and desire arise? Why do I call this good?” create the willingness to observe.
Most of us are absorbed in emotion, thought, certainty, and pursuit. Without curiosity about oneself, observation never begins. With distance, patterns become visible: under these conditions, I react in this way.
Observation then requires reflection. What conditions produced this pattern, what did I want, and why? Emotion is an accessible doorway into the interior. Asking where an emotion came from, observing it, and thinking about its operation is Buddhist practice.
Buddhist practice does not stop thought. It questions, observes body and mind, and reflects on their operation.
One Cannot Claim to Know Every Koan While Failing to See Oneself
Metacognition is the beginning and end of Buddhist practice. A person who cannot see his own bonds, emotion, or desire cannot credibly claim to have broken every koan or completed awakening.
I again asked Hyangbong whether every question can truly be understood and answered immediately. Why should speed be evidence of awakening? The cultural myth of the instant master itself requires examination.
I similarly criticized Kim Tae-wan for identifying a private experience with emptiness, not-self, seeing one’s nature, and final nirvana, then presenting his own path as the only truth. A private well becomes dangerous when it is declared to be the entire world.
If Accuracy Standards Rise Too Far, AI May Refuse Half the Questions
Returning to the report, I examined a practical difficulty in AI metacognition. A binary rule—answer only when certain, otherwise remain silent—reduces hallucination but also eliminates useful answers.
The graph described a tradeoff between utility and error. As the allowed error target becomes stricter, the proportion of questions answered falls sharply. Under a very strict target, roughly half the potential utility may disappear because the system cannot certify its own certainty.
Humans face the same problem. Speaking only with perfect certainty would silence ordinary advice and creative suggestions, while presenting every inference as fact destroys trust. A third path must honestly express uncertainty.
Faithful Uncertainty Calls a Guess a Guess
The report emphasizes faithful uncertainty. If something is well supported, say “It is so.” If it is unknown, say “I do not know.” If it is an inference, say “It seems so,” “This is my estimate,” or “This is how it appears.”
There is a major difference between commanding someone to follow an old brushing rule and saying, “I learned this rule, but I am not certain it remains correct.” The latter reveals the limit of the source and allows further checking.
Faithful uncertainty is not timid indecision. It is a technology of trust that discloses the boundary between knowledge, ignorance, and inference. By keeping uncertainty visible, humans and AI can preserve usefulness without disguising guesses as truth.
Distinguishing “It is,” “I do not know,” and “It seems so” is faithful uncertainty.
Hallucination Is a Tax Paid by Intelligence
I called hallucination and illusion a tax on intelligence. Intelligence recognizes patterns and infers what comes next from limited experience. It never possesses the entirety of reality before reaching a judgment.
An inference can be accurate and useful, but it can also fail. Humans and AI cannot use intelligence while removing every possibility of error.
Recognizing this tax creates a reason to reconsider. If I believe myself infallible, there is nothing to examine. When I know that intelligence works through inference, I inspect evidence, emotion, information, and desire before speaking.
Socrates Did Not Simply Declare “I Know Nothing”
I then moved from AI to Greek philosophy. The famous sentence “I know that I know nothing” can distort the expression associated with Socrates. In Plato’s Apology, it is closer to “I neither know nor think I know.”
The difference matters. “I know nothing” can become a total skepticism in which knowledge is impossible. “I do not think I know what I do not know” instead refuses to disguise ignorance as knowledge.
If I do not know how many ships sank in a battle, wisdom lies in admitting the missing information rather than confidently inventing a number. Socratic wisdom is metacognitive knowledge of the boundary of knowledge.
Socratic wisdom does not abandon knowledge; it refuses to think it knows what it does not know.
We Encounter Socrates Only Through Other People’s Records
Socrates left no surviving writings of his own. We meet him through Plato and other authors, whose portraits sometimes conflict. This produces the “Socratic problem”: uncertainty about the historical person behind the texts.
Even “I neither know nor think I know” is a line spoken by Socrates as represented in Plato’s work. The metacognitive lesson must therefore also govern the way we attribute it.
Plato’s Socrates says that another respected person did not know yet believed he knew, while Socrates was slightly wiser because he recognized his ignorance. Wisdom concerns not only the quantity of knowledge but awareness of its limits.
Montaigne’s Que sais-je? Carries the Tone “What Do I Really Know?”
The Socratic question reappears in Michel de Montaigne. I linked Montaigne’s death in 1592 with Yi Sun-sin’s victory at Hansando as a mnemonic joining distant historical scenes.
Montaigne inscribed Que sais-je? on a personal medal. Literally it means “What do I know?” but its living tone can be “What do I really know?” or “What on earth do I know?” It is self-critical rather than ceremonially philosophical.
The question does not abandon inquiry. It repeatedly places everything assumed to be known back on the test bench. It underlies Montaigne’s frank attention to body, feeling, daily life, death, society, and judgment.
The Essays Were an Experiment Conducted on the Self
Montaigne’s Les Essais helped shape the modern essay. The word carries the sense of an attempt, test, or trial. The book does not announce completed truth; it experiments with thought and life.
Montaigne wrote honestly about what he knew and felt, drawing on Epicurean, Stoic, and Pyrrhonian texts. He did not merely memorize ancient philosophy but tested it against his own habits and emotions.
The work is therefore an extended record of metacognition. It tested assumptions about self and society and deeply influenced modern intellectual life. Its power to provoke questions later placed it on the Catholic Index of Forbidden Books.
Printing Revived Ancient Skepticism as a Modern Question
After Gutenberg’s printing revolution, ancient texts spread rapidly across Europe. Pyrrho left no writings, but Sextus Empiricus’s outline of Pyrrhonism became widely read, alongside Epicurean and Stoic works.
Latin functioned somewhat like classical Chinese in East Asia, allowing intellectuals across regions to read common texts. Montaigne’s Latin education gave him access to antiquity and material for examining his own life.
Printing multiplied not only books but questions: how accurately do my senses receive the world, how far can I know, and what supports judgment? These questions disturbed scriptural authority, class, and kingship and helped open modernity.
Skepticism Is Not the Wordplay “You Cannot Say It Does Not Exist”
I harshly criticized the claim that because gods or rebirth cannot be proven, one also cannot say they do not exist. Used this way, skepticism protects superstition instead of challenging it.
If someone asserts gods, souls, heaven, hell, karma, or rebirth, that person must supply grounds for the assertion. One cannot make an unverifiable claim and then demand that the critic prove nonexistence.
I mocked the tactic with an invisible cockroach god whose billions of antennae monitor everyone. If the inventor can defend the story merely by asking others to prove it false, any fantasy survives. Skepticism asks the claimant, “How do you know?”
Inventing an unverifiable being and saying “You cannot prove it absent” is not skepticism; it is a shield for superstition.
Questions Break Sacred Blood, Religious Authority, and Superstition
“What do I know?” undermines authority as well as metaphysics. A pope crowning a king does not prove divine kingship, and aristocratic blood is not essentially different from ordinary blood.
Class, kingship, ownership, and religious privilege appear natural when nobody asks why. Questions—why are you king, where is the god who granted power, how do you know past karma determined rank—crack the myth.
This is why institutions that depend on unquestioned myths suppress questions and prohibit books. Read in reverse, the Index of Forbidden Books becomes a list of works that challenged superstition and opened modern thought.
Buddhist Inquiry Also Strikes Relics, Miracles, True Self, and Rebirth
Buddhism is fundamentally anti-superstitious. If relics are said to hold miraculous power, ask what a relic is and why human remains should possess transcendent force. Buddhist labels do not exempt a claim from examination.
The same applies to claims that seeing a true self makes one a Buddha, or that invisible energy, supernatural powers, afterlife, and rebirth are facts. If they cannot be observed or experienced through a shared method, how are they known?
Science, philosophy, and Buddhism meet in the cold intelligence that breaks false concepts. When someone blocks questions and demands belief, we should suspect an attempt to protect authority and fraud. As long as questions remain alive, deception cannot last forever.
Personal and Civilizational Liberation Begin by Reconsidering Knowledge
The lecture traveled from AI hallucination to human illusion, textual layers, observation of body and mind, Socrates, Montaigne, printing, and modernity. One sentence runs through every scene: “What do I know, and is what I think I know actually true?”
The question does not eliminate answers. It distinguishes fact from ignorance and inference, accepts correction, updates information, and observes how emotion and desire shape judgment. This is faithful uncertainty and living metacognition.
A disappearing mouse pointer interrupted the broadcast, which ran far beyond the planned length. I apologized to viewers for speaking for almost two hours and twenty minutes. Yet the conclusion is short: question, observe, and reflect. That is how individuals break delusion and civilization liberates itself from superstition.
This article was edited by AI solely from the supplied video captions and description. No external sources were used.