Gains:
- Ability to grasp why originality of judgment is inalienable and artificial intelligence can only be a sharpening partner
- Ability to recognize the forms of hallucination and separately verify the 10 percent fabrication within the 90 percent truth.
- Ability to maintain intellectual honesty through the principles of not hiding ambiguity, presenting the contribution as accurate, and testing one's own mind.
We have come to the deepest unit of this module. The only thing valuable in philosophy is the reasoning itself—an original distinction, a bold thesis, a subtle justification. In this unit we will cover three interconnected themes: originality of reasoning (when a thought is truly the product of one's own reasoning, not a copy of someone else's or a machine's ready-made output), hallucination (where an AI produces non-existent information, source or quotation as fact), and intellectual honesty (being honest with oneself and others in the production of knowledge and ideas; stating clearly what one does not know, what one has received from whom, and what is unclear). These three are the ethical core of the use of AI in humanities.
Previous units taught single skills; This unit establishes the mental discipline that underlies it all. Because tools change, but intellectual honesty does not.
Why is originality of judgment inalienable?
In philosophy, it is not enough for a text to "give the right answer"; How you arrived at that answer, what distinctions you made, and what objections you anticipated and met are of real value. AI produces the average of millions of texts; this average seems fluid, reasonable, and often accurate—but it is not original. Original thinking begins right where you break away from the average: a distinction no one else makes, a bold challenge to a conventional view, a new connection between two concepts.
You can use AI as a thinking partner — let it generate objections, show you your blind spot, help you clarify a concept. But if you delegate the thinking itself—thesis, justification, conclusion—you are left with an average text and your contribution is zero. Moreover, this is a violation of intellectual honesty: presenting someone else's (in this case a machine's) product as your own.
Tip: Ask yourself in each session the difference between “Let the AI think about it for me” and “Let the AI sharpen my thinking.” The first kills originality; The second strengthens. Concrete criterion: Did the thesis and justification that emerged at the end of the session come from your mind or was it copied from the screen?
Recognizing and warding off hallucination
The hallucination is not a mistake of the AI, but a consequence of the way it works: the model does not "know" the truth, it produces statistically probable text. So even when he doesn't have an answer to a question, he fills the gap with a believable fabrication. Typical forms of hallucination in human domains:
- Fabricated source: non-existent book, article, fake DOI.
- Fake quotation: a sentence in quotation marks that does not belong to the real author.
- Misattribution: attribution of a true view to the wrong philosopher.
- Made-up date/fact: a date, event, or statistic that does not exist.
- Confident ambiguity: presenting an issue that is actually controversial as if it were absolutely true.
The antidote is the verification discipline you learned throughout this module: link to source, independently verify, check verbatim. But an additional reflex: Don't mistake the AI's confident tone for a sign of accuracy. AI looks most confident even when it's making up the most.
Caution: The most insidious type of hallucination is 10% fabrication embedded in a 90% true text. Since most of the text is true, the reader relaxes and thinks that the made-up part is also true. That's why the "generally seems correct" test is not enough; Verify each specific source, quote, and fact individually.
Three rules of intellectual honesty
1. Hiding uncertainty. Don't present as certain something you don't know or can't verify. Saying “this point is debatable” or “I could not verify this” is honesty, not weakness.
2. Show contribution correctly. If you have used AI and your organization requires disclosure, state this transparently. Do not take someone else's or the machine's contribution as your own.
3. Test it with your own mind. No matter how fluent an outcome is, do not accept it without going through your own judgment. The final authority is your critical mind, not the machine.
three mini cases
Case 1 — Average noticed. A PhD student wrote part of it by “brainstorming” it with AI, but used the output largely as is. His advisor said the text contained no unique distinctions but was an average "that anyone could tell." The student again used AI as a discussion partner: he received objections, but he established the thesis and justification. The second draft contained an original contribution.
Case 2 — 10% fabrication. A researcher would use a 10-source, streamlined and largely accurate literature summary that AI produces. The overall text was solid, so it almost passed verification. Still, he checked every source: 1 in 10 was fabricated and one quote was taken out of context. Just because most of the text was true didn't make that 10% innocent. Systematic checking caught a single error.
Case 3 — Honest uncertainty. A student was unable to justify a point in a seminar text. Instead of dismissing it by saying "AI said so," he clearly wrote, "this claim is not clear in the sources, it needs to be verified." The counselor considered this a sign of maturity, not weakness. Intellectual honesty was more valuable than false certainty.
Weak prompt / Strong prompt
Weak prompt:
Write an original and in-depth philosophical analysis on this topic.
This demand is contradictory: demanding originality from AI precisely kills originality. The result is a text that appears "original" but is average and possibly apocryphal.
Powerful prompt:
Your role: a socratic thinking partner. DO NOT WRITE FOR ME on this topic and DO NOT GIVE ME A THESIS. Instead, sharpen my thinking: 1) Ask for 3 weak points in my position. 2) Point out 2 distinctions I may have missed. 3) Give a counterexample so I can test my thesis. If you claim a source or fact, mark it as "must be verified"; do not make up if you are not sure. I will write the thesis and the text. My position: [insert your own position]
This prompt keeps the thinking in you, turns the AI into a sharpening partner, and prompts anti-hallucination marking.
Four copyable templates
1) Socratic partner (keeps you thinking):
Don't answer me; ask questions Test my position below with 5 sharp Socratic questions: uncover my assumptions, skipped steps, and counterexamples. Producing a thesis. My location: [location]
2) Hallucination marking:
Answer below, but label each factual claim, source, and quote with its level of confidence: [CERTAIN] / [MUST BE VERIFIED] / [NOT SURE]. Do not fabricate any source or quote you are unsure of. Say "I don't know" instead. Question: [question]
3) Average-specificity diagnosis:
The text below is mine. Which sections are average, which "everyone can say", and which contain a unique distinction/claim? Mark the remaining places so I can make them unique. Adding content; diagnose only.Text: [text]
4) Uncertainty honesty:
Find claims in the text below that are presented as accurate but are actually disputed or unsubstantiated. For each, suggest how it could be expressed more honestly. Text: [text]
Authenticity/honesty table
behavior
originality
honesty
Conclusion
Having the thesis written by AI
None
violation
worthless, risky
Brainstorming with AI, writing the thesis yourself
Yes
Honest (declare)
precious
Using AI source without verifying it
—
Breach risk
Danger of fabrication
State the uncertainty clearly
—
high
reliable
Common mistakes
- Asking for originality from AI. It is a contradictory demand; originality arises only from your judgment.
- Settle for "generally true". 10% fabrication hides within 90% truth; Verify each specific claim separately.
- Mistaking the confident tone for accuracy. AI looks most confident when it's making up the most.
- Hiding uncertainty. False certainty is more harmful than honest reservation and destroys trust.
- Concealing the contribution. Failure to disclose the use of AI is against corporate policy and integrity.
In summary
Originality of judgment, hallucination, and intellectual honesty are the ethical core of the use of AI in humanities. The thinking itself is inalienable: AI may be a sharpening partner, but the thesis, rationale, and conclusion are yours; Otherwise, what remains is an average and unadulterated text. Hallucination is the result of the way the AI works; Do not mistake the confident tone for truth and catch the 10% fabrication within the 90% truth by separately verifying it. Three rules of intellectual honesty: hide uncertainty, present contribution accurately, test with your own mind. Tools change; This discipline does not change.
Application task
- Write a philosophical position of yours (in your own words, briefly).
- Get just questions from the AI with the “Socratic partner” template; do not take any thesis.
- Strengthen your position by writing your own answers to these questions.
- Have an AI output generated with a “hallucination marking” template and actually verify claims labeled [VERIFIED].
- Test your own text with the "average-originality diagnosis" template and make a section unique.
checklist
- [ ] I produced the thesis, justification and conclusion with my own reasoning.
- [ ] I used AI as a partner to sharpen thinking, not as a substitute.
- [ ] I have individually verified each specific source, quote, and fact.
- [ ] I trusted the evidence, not the AI's confident tone.
- [ ] I have honestly flagged vague/unsubstantiated claims.
- [ ] I have declared the use of AI according to corporate policy.