AI as a research interlocutor
Method before tricks
Understand prompt engineering not as a collection of tricks, but as the design of an intellectual task.
The student learns to treat AI as a research partner, not as a final author or oracle.
Lessons of the module
3 lessons · pick one to read
Understanding why treating AI as a research interlocutor - and not as the final author - changes the quality and ethics of the work.
Who signs the work?
AI can write paragraphs, suggest structures and summarize readings. But delegating writing is not the same as delegating authorship. Authorship is the set of choices: the cut, the interpretation, what goes in and what stays out, and the responsibility for the result.
Keeping this clear from the start avoids a common slip-up: accepting a well-written text as if it were your thoughts. The text may be the starting point; the thought and decision remain yours.
Interlocution is productive friction
A good interlocutor doesn't deliver ready-made truth: it questions, offers counterpoints, organizes what is confusing and returns questions. It is in this friction that the research progresses.
Use AI to think against yourself: ask for objections to your hypothesis, alternative readings, what a critic would say. This is much more valuable than asking it to agree with you in elegant prose.
The risk of the oracle
Treating AI as a source of truth leads to confusing fluency with authority. Since the answer sounds safe, it's easy to stop checking.
In research, this is especially dangerous: a plausible and erroneous statement, dressed up as a citation, can contaminate an entire argument. Interlocutor, yes; oracle, never.
Reverse the request
Take a hypothesis from your research and, instead of the answer, ask the AI for 5 strong objections to it. What changes in your relationship with the tool?
See expected answer
You move from "consume response" mode to "test thinking" mode. AI becomes a critical mirror - and you realize that you're still the one who decides what to do with the objections.
AI is an interlocutor: it helps think, organizes and creates tension. Authorship - choices, interpretation and responsibility - remains human.
Deliverables
- Personal map of possible AI uses
- Researcher responsibility checklist
- First structured research prompt
Module model prompt
A starting point — adapt the [ brackets ].
Act as a critical research interlocutor. My topic is [topic]. Before answering, ask 5 questions to define the objective, corpus, theoretical approach, limits and quality criteria.