Beyond Inductive, Deductive and Abductive: The Conductive Approach and Agnoseology in the AI Era

Every research methods textbook teaches the same three reasoning logics. Induction: patterns rise from the data and become theory. Deduction: theory comes first and the data test it. Abduction: faced with surprising data, the researcher infers the best available explanation. These three logics have organised qualitative research for a century — and all three quietly assume that the only knowing agent in the room is the human researcher.

Generative AI has broken that assumption. When a researcher uses ChatGPT or a similar model inside thematic analysis, a second contributor enters the analytical conversation: one that proposes keywords the researcher had not noticed, patterns the researcher had not looked for, and meanings the researcher had not anticipated. The researcher answers with what the machine lacks — theoretical judgement, contextual understanding, philosophical position. Which of the three classical logics describes that exchange? None of them. That gap is what the conductive approach fills.

What conductive coding is

The approach was introduced in a study of luxury brand co-creation on Instagram, where AI-assisted thematic analysis was conducted within a fully specified theoretical and methodological frame. The paper defines it precisely:

“The definition of conductive coding refers to the integration of the theoretical, contextual, major research aim, objectives, and methodological background of this research through AI-aided coding, where the coding is shaped by recommendations generated by artificial intelligence in response to human input related to the context, methodology, and philosophy, and this combined process is what is called conductive coding.” (Naeem, 2026, p. 6)

The word “conductive” is doing real work here. The researcher conducts — in both senses: directing the analysis like a conductor directs an orchestra, and serving as the medium through which theoretical and contextual currents flow into the AI’s pattern-finding. And the paper is explicit about the contrast with the classical logics: “as opposed to Inductive, Deductive or Abductive logics, conductive coding was used as an approach that is simultaneously iterative and dialogical” (Naeem, 2026, p. 6).

Agnoseology: the paradigm underneath

Behind the method sits a larger philosophical claim. Epistemology — the theory of knowledge — asks how we know what we know. But AI-assisted research constantly confronts a different situation: the machine surfaces things we did not know we did not know. The paper names the systematic study of that process Agnoseology:

“Agnoseology has been developed as a philosophical concept in which researchers seek to explore the area of unknowns, from identifying research gaps to addressing those gaps… researchers receive recommendations from AI that help uncover unknown or hidden data and previously unrecognised meanings within the dataset, and this exploratory process is what is referred to as Agnoseology.” (Naeem, 2026, p. 6)

Conductive coding is then positioned as “a methodological application of Agnoseology” — the practical procedure through which the paradigm operates. In this framing, uncertainty stops being an obstacle to rigour and becomes analytical material: “uncertainty was contextual rather than obstructive and became co-constructed to reveal large-scale ideas from new data” (Naeem, 2026, p. 6).

Why this matters for qualitative researchers

The practical stakes are high. Thousands of researchers are already using AI in their analysis, but most describe it — inaccurately — as “inductive coding with AI assistance.” That description fails at peer review, because inductive logic cannot account for the AI’s contribution or the researcher’s steering. The conductive approach gives that practice an accurate name, a published definition, and a philosophical justification, aligned with the step-by-step AI processes for thematic analysis (Naeem, Smith & Thomas, 2025) and case study research (Naeem & Thomas, 2025). If your methods section says AI was involved, your reasoning logic section can now say how — and cite it.

There is also a deeper implication. The paper concludes that “both Agnoseology and conductive coding contribute to a new understanding of AI as a methodological and philosophical collaborator” (Naeem, 2026, p. 6) — turning qualitative analysis into “a conversation between the AI and human interpretation, from which new philosophical questions can then develop as an adventure into the unknown.” The AI era does not just need new tools; it needs new logic. This is that logic, in print.

How to use it in your own study

Three moves make your analysis conductive rather than vaguely “AI-assisted.” First, specify the human input: document the theory, context, aims, methodology and philosophy you provided to the AI before and during coding. Second, document the dialogue: keep the record of AI recommendations and your acceptances, rejections and refinements — this is the iterative, dialogical evidence trail. Third, name and cite the logic: state that coding followed the conductive approach (Naeem, 2026) as a methodological application of Agnoseology, within the six-step systematic thematic analysis process.