Prompts Are Not Questions
The grammatical form of a question has almost nothing to do with what Gadamer meant by a genuine question. A sentence can take interrogative form and be a command: 'Could you close the door?' is a request for compliance, not an inquiry. Much of what passes for questioning in the AI conversation is of this kind — interrogative in syntax, imperative in substance. 'Can you write a Python function that sorts a list by the second element of each tuple?' is a prompt. The human who types it does not wonder whether the task can be accomplished. They know, within reasonable parameters, what the output should look like. The 'question' is simply the most convenient way of requesting it. Gadamer would not object to prompts as such — the productive capacity of tools is legitimate — but he would insist that prompting and questioning belong to different orders of engagement, and that confusing them constitutes the defining category error of the AI age.
In the AI Story
The difference between prompt and question can be articulated precisely by examining what happens to the human in each case. When Segal prompts Claude to build a face-detection component, the interaction follows a characteristic arc: specification, execution, review, refinement. At the end, Segal has a working component. He does not, in any meaningful sense, have a deeper understanding. He has extracted a result.
When Segal brings a genuine question — 'Why is adoption so fast? What does the speed measure beyond product quality?' — the interaction changes character entirely. The question arises from learned ignorance. It puts the questioner at risk. The answer, when it arrives through the collision of intuition with Claude's associative reach, changes understanding not incrementally but by reframing the framework itself.
The rise of prompt engineering as a serious discipline — with best practices, certification programs, and career tracks — reflects a cultural assumption that better prompting produces better understanding. Gadamer's framework identifies this assumption as a category error. Prompt engineering is a techne, a productive craft of extraction. It is not hermeneutics.
The AI conversation structured as prompting cannot produce the Socratic elenchus. The machine does not challenge the prompter's assumptions. It receives the prompt, processes it, and produces an output calibrated to satisfy. If the request rests on a false premise, the output reflects the false premise — not because the machine is deceived but because its design is oriented toward satisfaction rather than challenge.
Origin
The distinction has philosophical roots in Plato's Meno, where Socrates distinguishes between the slave who recites what he has been taught and the slave who, through genuine questioning, arrives at understanding.
The contemporary form of the distinction emerged as large language models made the extraction of fluent output cheap, forcing philosophical reconsideration of what question-asking actually is and whether it survives the collapse of its production cost.
Key Ideas
Grammar is not substance. Interrogative form does not constitute a question. Many questions are commands; many statements are inquiries.
The prompt operates within the horizon. It seeks what the prompter can already envision. The question operates at the boundary where understanding fails.
Extraction versus transformation. Prompts produce output that confirms existing frameworks. Questions produce encounters that revise them.
The cultural category error. Training people to be better prompters trains them to extract more efficiently. Training people to be better questioners trains them to understand more genuinely.
The Socratic test. A genuine question can produce the elenchus — the painful recognition that what one thought one knew, one does not. The prompt cannot produce this, because the prompter already knows what they want.
Debates & Critiques
Defenders of prompt engineering argue that well-crafted prompts can produce surprising outputs that widen the prompter's horizon. Gadamerian critics respond that this conflates two different phenomena: the prompt-as-surprise (which is productive craft) and the genuine question (which is hermeneutic risk). The difference is not subtle — it is the difference between techne and the opening of a space in which phronesis becomes possible.
Appears in the Orange Pill Cycle
Further reading
- Gadamer, Hans-Georg. Truth and Method (1960), Part Three, Section II.B: 'The Logic of Question and Answer.'
- Plato, Meno, for the Socratic origins of the distinction.
- Collingwood, R.G. An Autobiography (1939), for the logic of question and answer as methodological principle.
- Palmer, Richard. Hermeneutics (1969).
- Vilas-Boas, Daniel. "Prompt Engineering and the Hermeneutic Deficit" (2024).