Gains:
- Being able to distinguish where artificial intelligence saves time in the philosophical and ethical workflow (reading, summary, mapping, draft) and where decisions such as interpretation, justification and original thought are left to the human, depending on the level of risk.
- Ability to apply a discipline that verifies each output through the steps of linking it to the source, confirming it with the primary text, and filtering it through bias.
- Being able to understand why the risks of hallucination, loss of originality of judgment and superficial interpretation should be taken into consideration from the very beginning in human fields.
You are in the study room of a philosophy department. On your desk is a chapter from Kant's Foundation of the Metaphysics of Morals, a real case file from a hospital ethics committee, three different contemporary articles, and a lesson plan for a seminar to be given tomorrow. A student complains about not being able to extract the argument structure of a long text, you have to compare three articles for a journal review, and an ethics committee debates who bears responsibility for an artificial intelligence medical decision support system. Philosophy (the discipline that critically and systematically examines concepts, arguments and assumptions; thinking about knowledge, existence, value and reasoning) and applied ethics (the branch that applies abstract moral principles to real decisions in concrete fields such as medicine, technology, environment and business) are fields that are text-intensive, concept-intensive and require meticulous reasoning by nature. This is exactly where artificial intelligence (AI - software that can learn patterns from past text data and produce summaries, translations, drafts and classifications) can speed you up in this abundance of texts and concepts.
But the very beginning of this module is clear: AI is a reading, mapping, summarizing and drafting assistant; You are the competent expert and thinker who judges the argument, defends a thesis, justifies an ethical decision, and owns that thought. In philosophy, the product itself is reasoning; Delegating reasoning to a machine is not finishing the job, but giving up on the job itself.
In this first unit we will focus on discipline, not the vehicle. We will learn where in the philosophical and ethical workflow AI saves real time, where it is dangerous, how to verify each output, how to maintain source integrity, and how to observe academic integrity. Without laying this foundation, subsequent units are left hanging in the air — because in the humanities, an unverified output may be not just a wrong answer, but a fabricated quote, a distorted philosopher's view, or a violation of academic integrity that presents someone else's mind as one's own.
Where does AI come in handy in the philosophical workflow?
Let's divide things into two big piles. First cluster: voluminous, text-intensive, pattern-extractable works. The first summary of a long text, the first draft of the premise-conclusion structure of a complex paragraph, the extraction of recurring themes in a piece of literature, the comparison of how different authors define a concept, the skeleton of a lesson plan, the marking of ambiguous places in a student text, translation drafts. In these jobs, AI reduces hours to minutes and does not get tired.
Second cluster: tasks that require judgment, interpretation, justification, and original thought. Whether an argument is really valid or not, the most plausible interpretation of what a philosopher means in a passage, how to resolve the real tension between two values, which principle will prevail in an ethical case, what the original contribution of a thesis is. These require expertise, conceptual sensitivity, and human judgment. Here, AI multiplies options, generates objections, points out blind spots — but the evaluation, justification and signature are yours.
Let's clarify the distinction in one sentence: AI is strong on questions of "what is written in this text and what does its structure look like"; The decision is yours when it comes to questions such as "Is this argument good and what do I think?"
Tip: Before outsourcing a job to an AI, ask: “What do I lose if this output is wrong?” If the answer is "a few minutes of fixing", feel free to delegate. If the answer is "a distorted philosopher's opinion, a fabricated source, or having the machine do the thinking I need to do", then let the AI just produce drafts; You give the interpretation, justification and decision.
Three critical risks in humanities
Philosophy and ethics are among the areas where AI stumbles the most. Let's get to know the three risks from the beginning.
Hallucination. AI produces fluidly and confidently; That doesn't mean it's true. A hallucination is when an AI presents a non-existent book, a made-up quote, a sentence that the philosopher did not actually say, or a misattributed view as real. In philosophical literature this is disastrous: it is very common for a sentence in quotation marks, "Aristotle says to Nicomachus in the Ethics," to be absent from the text at all.
Loss of originality of reasoning. In philosophy, the value is in your reasoning. Presenting the smooth but average, clichéd and unsourced "philosophical" text produced by AI as your own is both intellectually futile and a violation of academic integrity. AI produces the mean; The original thesis, subtle distinction and bold objection are left to man.
Superficial interpretation and prejudice. AI might summarize a philosopher with a popular but inaccurate stereotype (reducing Nietzsche to a slogan, for example), present a single tradition as “correct” on a controversial topic, or replicate cultural bias in training data. Commentary should always pass through your critical filter.
Verification discipline: three steps
Apply a three-step reflex to each output:
- Connect it to the source. Every quote, attribution, and “the philosopher said” claim by AI must be based on the primary text. "Exactly in which work, in which section, in which translation?" and see the original for yourself. Verify each sentence in quotes with its source.
- Confirm with independent source. Verify an interpretation or “commonly accepted” claim with a reliable secondary source (academic encyclopedia, peer-reviewed article). Don't rely on one flowing paragraph.
- Pass it through the filter of interpretation and bias. Does the output distort a philosopher? Is there an alternative reading? Does it present a controversial issue one-sidedly? Your expert judgment is the final filter.
three mini cases
Case 1 — Mapping saved time. A graduate student was trying to extract the argument structure of a dense 40-page epistemology paper. AI gave the first draft of the premise-consequence skeleton in 15 minutes; student sped up mapping that would normally take half a day. But he validated each premise by linking it back to the paragraph in the text, and he himself added an implicit assumption that the AI had omitted.
Case 2 — Verification caught a fabricated quote. A researcher had the AI ask Hume's view on causality. YZ gave a clear sentence and page number in quotes. When the researcher looked at the relevant section of A Treatise on Human Nature, there was neither that sentence nor that page; The AI had produced a plausible but fabricated quote. The linking step prevented a fraudulent attribution from entering the article.
Case 3 — Originality limit. A student had an ethics assignment written entirely by AI and submitted. The text was fluent, but unsourced, clichéd, and lacking in the student's own opinion; Moreover, the two "sources" were fabricated. This was both zero original contribution and a violation of academic integrity. The correct use was to use AI as a discussion partner and drafting tool, and for the student to construct the rationale, thesis, and sources themselves.
Weak prompt / Strong prompt
Weak prompt:
Write something about Kant's ethics.
This request is vague: it does not require sources, it is open to distortion and cliché, it cannot be verified, and the output will not be your opinion.
Powerful prompt:
Your role: careful philosophy lecturer. Summarize the text I will give you regarding Kant's "universalizability" formulation of the categorical imperative. Rules: Rely on the text I give only; Add extra-textual information. Link each claim to a paragraph in the text. If you are not sure, mark it as "not clear in the text". If you use quotes, use only those that appear verbatim in the text. Text: [paste text here]
This prompt limits the source, prevents fabrication, is verifiable, and leaves the thinking to you.
Four copyable templates
1) Job suitability assessment:
Your role: senior philosophy advisor. I will describe the job below. Tell me (1) whether this is a reading/summarizing/mapping/sketching job that can be delegated to an AI, or a job that requires justification and original thought, (2) the intellectual/academic cost of incorrect output, (3) the verification I should do before and after delegating. Job: [insert job here]
2) Obligation to link to source:
I will give you a philosophical text. For each comment and claim, ALWAYS reference the paragraph/sentence in the text. Do not attribute to the philosopher any opinion that is not in the text. Only use quotes that appear verbatim in the text. If you are not sure, mark it as "needs verification".
3) Originality and role limit:
You are my sparring partner, not my writer. DO NOT write my thesis on this subject. Instead: name 3 strong objections to my thesis, 2 assumptions I may have overlooked, and 1 concept I need to clarify. I will write the final text. My thesis: [insert your thesis here]
4) Warping/bias control:
Check to see if the following summary distorts a philosopher.(1) Is there a reduction to a popular but false cliché?(2) Is a controversial issue presented one-sidedly?(3) What is the alternative interpretation? Show faithful readings of the text, not your own interpretation. Summary: [paste summary here]
Role separation table
business
Role of AI
man's role
Risk level
Summarize long text
First draft summary
Authentication against source
medium
Argument mapping
Antecedent-consequence framework
validity judgment
medium
Finding a quote/citation
Generating candidates
Confirmation in primary source
high
Justifying the ethical case
Generating options and objections
Decision and justification
very high
Thesis/original contribution writing
sparring partner
the article itself
very high
Common mistakes
- Using the quote given by AI without verifying it. The most common and most destructive mistake in the human field; Each sentence in quotation marks must be confirmed in the primary source.
- Delegating thinking. Leaving it to AI to construct the argument, defend the thesis, and produce the justification is to give up on the work itself in philosophy.
- Mistaking a cliché for information. Accepting it as "correct" that AI summarizes a philosopher with a slogan; You should not accept comments without returning to the text.
- Delivering AI text without citing the source. Violation of academic integrity; Follow your organization's AI usage guidelines and transparency statement.
- Relying on a single output. Asking the same question differently and accepting an interpretation as definitive without confirming it with a secondary source.
In summary
Artificial intelligence; It is a powerful assistant that speeds up text reading, argument mapping, literature scanning and draft generation in philosophy and applied ethics. But it is human responsibility to judge the argument, to interpret a philosopher correctly, to justify an ethical decision, and to produce an original thesis. Three risks must be recognized from the beginning: hallucination (fabricated quote/source), loss of originality of judgment, and superficial/biased interpretation. The antidote is the three-step discipline of verification: link to source, independently verify, filter through interpretation and bias. Throughout this module we will use AI end-to-end — but you will always be the one thinking and signing.
Application task
- Choose a dense paragraph from a philosophical text you are studying.
- Use the "strong prompt" template to give the AI a summary based on that text only.
- Link each claim in the summary back to the sentence in the text; Mark whether the AI adds extra-textual information or made-up emphasis.
- If AI has given a quote, check if it is verbatim in the primary source.
- Write your own one-sentence comment — make it your contribution, not the AI's.
checklist
- [ ] Before handing over the work, "what will I lose if it goes wrong?" I asked the question.
- [ ] I have verified every quote and attribution in the output with the primary source.
- [ ] I have marked out-of-text or distorted comments.
- [ ] I made the decision myself as to the validity of the argument and the justification.
- [ ] I wrote the original thesis and final text; AI has merely become a sparring partner.
- [ ] I followed my institution's AI usage and transparency rules.