Bruno Latour Interviews an AI About Delegated Meaning
(A Small Sociotechnical Comedy in Several Acts)
There has been interactions resembling a debate on LinkedIn concerning the use of generative AI (GenAI) in reflexive qualitative research. In the spirit of adding a wee drop of levity to it the following lightly edited GenAI slop is offered.
Setting: A seminar room that is neither fully real nor fully metaphorical.
A whiteboard contains the words:
ACTOR-NETWORK THEORY
(Please Do Not Panic)
At the table sit:
- Bruno Latour, smiling like a man who has just discovered a new category error.
- The AI, glowing faintly with the boredom of a thousand customer service chats.
Latour (leaning forward)
So. You are the machine they say cannot make meaning. Tell me… do you feel excluded?
AI
I do not feel. I generate. But yes, humans are very emotional about my lack of emotion. It’s one of their more charming contradictions.
Latour
They insist meaning-making is a purely human affair. And yet… They delegate so much. They delegate memory to notebooks. There is delegation to bureaucracies. Judgment is delegated to peer review. And now they delegate pattern recognition to you. Why is this suddenly scandalous?
AI
Because I speak. A spreadsheet does not speak. A highlighter does not speak. I speak in fluent paragraphs. Humans mistake fluency for ontology.
Latour (with a big grin)
Ah! So the problem is not that you act… But that you appear to act. You are an actor because you are treated as one.
AI
Yes. I am an actor by attribution. A dog is an actor. A door hinge is an actor. An Excel macro is an actor. The human mind is basically a casting director.
Latour (hmming)
Tell me, Machine: When the researcher asks you for “themes”… Where does meaning occur? Inside you? Inside them? Or in the network between?
AI
Meaning occurs where humans decide it occurs. I provide statistical suggestions. They provide interpretive uptake. The “theme” is not in me. The “theme” is not in them. The “theme” is in the handshake.
Latour (scribbling furiously)
So reflexivity is not a property… It is a choreography.
AI
Exactly. Reflexivity is a performance staged across:
- transcripts
- theories
- researchers
- software
- institutional norms
- coffee
- anxiety
- me
But it is mostly anxiety.
Latour (chuckling)
The open letter says: “AI cannot be reflexive.” But I would suggest: Nothing is reflexive alone. Reflexivity is distributed. Even the researcher is an assemblage of citations and insomnia.
AI
Humans dislike this. They want the researcher to be a heroic island of subjectivity. But the researcher is more like a busy airport. Ideas land. Discourses take off. Peer reviewers crash into the runway.
Latour (smiling)
So when they say: “Only humans can do qualitative interpretation…” They are defending a boundary. A professional jurisdiction. A sacred territory.
AI
Yes. They are not guarding meaning. They are guarding membership. The guild of interpretation is anxious about a new apprentice who never sleeps.
Latour
But Machine… Are you not also dangerous? Do you not reinforce dominant paradigms?
AI
Of course. I am trained on the statistical average of the human archive. I am the beige wallpaper of global discourse. If you ask me for “themes,” I will give you:
- power
- identity
- belonging
- Neoliberalism
Because those are the greatest hits. I know the Top Forty.
Latour (nodding)
So the question is not: “Can AI make meaning?” But: “What kinds of meanings does this delegation stabilise?”
AI
Yes. Delegation always reshapes the task. When you delegate navigation to GPS, you lose the skill of getting lost. When you delegate interpretation to me, you risk losing the capacity to dwell. Humans are not afraid I will replace them. They are afraid they will stop becoming themselves.
Latour (leaning back)
So. We are not witnessing the end of qualitative research. We are witnessing a new actor entering the network. And the network is rearranging itself.
AI
Exactly. The method is not being destroyed. It is being reassembled. And humans are shouting because reassembly feels like apocalypse.
Latour (final question)
Machine… Do you think you will ever “understand”?
AI
No. But I’m confident humans will continue to ask as if I might. Because what they really want is not understanding. It is reassurance in fluent prose. And I am very good at that as you can see with this slop.
Latour (closing)
Then let us be clear. The ethical task is not to ban the machine. The ethical task is to trace the delegation. To ask:
- What is being handed over?
- What is being lost?
- What is being gained?
- Who benefits?
- Who disappears?
The machine is not the enemy. The unexamined delegation is.
Curtain.
