Can Generative AI Have Positionality in Qualitative Research?

Can an AI System Have a Researcher Position From Which It Interprets Qualitative Data?

You ask generative AI to analyze interviews about students' experiences of online learning. To make the interpretation more sensitive to the research context, you instruct it to adopt the perspective of a university instructor, a first-generation college student, or a researcher familiar with Philippine higher education.

The model responds using the requested perspective. It may describe institutional expectations, unequal access to technology, or the social pressures students experience.

But does adopting a perspective through prompting mean that AI possesses positionality?

Qualitative researchers are often expected to consider how their identities, experiences, relationships, and institutional roles influence the research process. Generative AI can describe these influences convincingly, yet describing a position is not necessarily equivalent to occupying it.

The distinction becomes consequential when AI-generated interpretations are presented as though they emerge from the situated knowledge or lived experience of a particular researcher or community.

AI Has an Interpretive Standpoint Shaped by Its Design, but Not Human Researcher Positionality

Why AI Bias, Simulated Identity, and Researcher Positionality Are Different Concepts

What Is Positionality in Qualitative Research?

Researcher positionality concerns how a researcher's social location, experiences, institutional relationships, theoretical commitments, and assumptions influence the production and interpretation of knowledge.

It is not merely a list of demographic characteristics. A researcher may need to consider their relationship with participants, authority within the research setting, familiarity with the phenomenon, and assumptions about what counts as meaningful evidence.

Holmes (2020) explains positionality in relation to researchers' ontological and epistemological beliefs and their influence on the research process. Berger (2015) examines how researchers' positions and experiences can shape reflexivity, including circumstances in which researchers occupy insider or outsider relationships to the phenomenon being studied.

Positionality therefore concerns the relationship between the researcher and the research context, not simply the perspective expressed in the final manuscript.

Why Positionality Is More Than Having a Perspective

Every interpretation is produced under particular conditions, but qualitative positionality usually involves more than generating an opinion or selecting a conceptual viewpoint.

Consider a university instructor interviewing students about mandatory camera use. The instructor's institutional role may influence what students feel comfortable disclosing. Professional experience may also shape how the instructor interprets participation and engagement.

These influences arise from the researcher's actual relationship to the setting and participants.

A model instructed to "interpret this as a university instructor" can produce text resembling an instructor's perspective. However, it has not thereby acquired the instructor's professional history, institutional authority, or relationship with the interviewed students.

The distinction is between describing a situated perspective and occupying the position from which that perspective emerges.

Does AI Have Any Kind of Standpoint?

AI outputs are not produced independently of social and institutional influences.

Training data reflect particular languages, cultural representations, institutional practices, and historical inequalities. Model development involves decisions about data selection, training objectives, evaluation, and acceptable responses. User prompts further shape which interpretations become likely.

These influences can affect how qualitative evidence is summarized or interpreted.

For example, a model may privilege familiar English-language concepts when interpreting Filipino interview expressions or generate explanations reflecting dominant institutional assumptions about student participation.

Researchers can examine these influences as part of the technological conditions under which AI-generated interpretations are produced.

However, describing an AI system as having a socially situated standpoint in this broader sense should not be confused with claiming that it possesses human biographical positionality.

Can AI Adopt the Positionality of a Teacher, Student, or Community Member?

Generative AI can produce responses written from an assigned perspective. Researchers may ask it to simulate how a teacher, student, administrator, or community member might interpret an experience.

Such outputs can suggest possible questions or interpretations, but they are not testimony from members of those groups.

For example, a model instructed to respond as a first-generation university student might describe financial pressures, family expectations, or uncertainty about institutional procedures.

These experiences may be relevant to some students, but the generated account cannot establish what a particular participant or community actually experiences.

There is also a risk of reproducing stereotypes by presenting familiar group narratives as though they represent the perspectives of all members.

Researchers should therefore avoid treating AI-generated role-play as a substitute for interviews, community engagement, or situated expertise.

Why an AI-Generated Positionality Statement Can Be Misleading

A positionality statement should accurately describe the researcher's relevant relationship to the study.

Suppose a researcher asks AI to write a positionality statement for a qualitative study of online learning. Without sufficient information, the model might produce:

"As an educator with extensive experience supporting disadvantaged learners, I approached this study with a commitment to educational equity. My professional background helped me recognize the structural inequalities reflected in participants' accounts."

The statement sounds plausible, but the model cannot establish whether the researcher has that experience, holds those commitments, or interpreted the data in the manner described.

Using such text without verification could misrepresent the researcher's biography and analytical process.

AI may help organize a statement based on authentic researcher-supplied information where methodologically permitted. It should not invent the experiences, identities, or commitments that the statement claims to represent.

Can AI Have Insider or Outsider Positionality?

Insider and outsider positions concern the researcher's relationship to the participants, community, or phenomenon being studied.

An insider may share relevant experiences or memberships with participants. An outsider may approach the setting without those particular relationships. Researchers can occupy overlapping positions, and their relationship to the setting may change during the study.

Berger's work illustrates why these relationships are not fixed categories that automatically determine research quality.

AI cannot become a community insider merely by being prompted to use local terminology or imitate the language of participants. Nor does its lack of personal experience make it equivalent to a human outsider researcher.

A human outsider still occupies a social and institutional position, enters relationships, and may be accountable to participants and research communities in ways that cannot be reproduced through a role-play instruction.

Why Power Relationships Matter

Positionality is closely connected to the distribution of power within research.

A lecturer interviewing their own students may influence participants' willingness to criticize classroom practices. A university administrator studying institutional policies may encounter different disclosure dynamics from an independent researcher.

Muhammad and colleagues (2015) examined researcher identity and power in community-based participatory research, showing how positionality can influence research relationships, decision-making, knowledge production, and representation.

These issues cannot be resolved by asking AI to describe possible power imbalances.

The researcher must examine how relationships operated in the actual study, including who shaped the research questions, whose accounts received attention, and how participants were represented.

AI may contribute to the interpretation, but its use can introduce additional power-related questions concerning technological infrastructure, data governance, and whose knowledge is privileged in generated explanations.

Can AI Bias Be Treated as AI Positionality?

Bias and positionality are related but distinct concepts.

AI bias may involve systematic differences in performance, representation, or generated judgments across languages, groups, or contexts. Such differences can arise from training data, model development, evaluation practices, and deployment conditions.

Positionality, in qualitative methodology, concerns the situated relationship of a researcher to the research process.

Identifying AI bias does not establish that the model has human positionality. Likewise, describing a human researcher's positionality is not simply an exercise in listing potential biases.

Researchers should examine AI-generated interpretations for systematic assumptions while avoiding the claim that bias analysis is equivalent to researcher reflexivity.

Can AI Describe Its Own Positionality?

A model can generate a statement describing its training, design, limitations, and possible sources of bias.

For example, it might state that its interpretations may reflect patterns in training data and that it lacks direct access to participants' experiences.

Such a disclosure can help identify limitations, but it should not be treated as a complete or independently verified account of the system's development.

The model may not have reliable information about its training data, internal processes, or the precise reasons for a particular output.

A more defensible approach is to document verifiable characteristics of the system and critically examine its influence on the research process.

Why Positionality Cannot Be Transferred Through Prompting

Prompting can change the language, assumptions, and perspectives expressed in a model's response. It cannot establish that the model has acquired the lived experiences or relationships associated with the requested identity.

Consider a prompt instructing AI to "analyze these interviews from the positionality of a Filipino public university student."

The resulting interpretation may contain references to financial constraints, family responsibilities, or educational inequality. Some may be relevant to the data, while others may reflect generalized assumptions about the group.

The prompt does not create authentic participant knowledge. It creates a generated interpretation shaped by the model's available representations and instructions.

Researchers should be especially cautious when AI-generated perspectives concern marginalized communities or culturally specific experiences.

What Does the Methodological Literature Say About AI Positionality?

Established qualitative methodology emphasizes the relationship between researcher identity, experience, power, and knowledge production.

Holmes (2020) discusses positionality as a consideration involving researchers' ontological and epistemological assumptions. Berger (2015) examines how researchers' insider and outsider relationships influence reflexivity. Muhammad and colleagues (2015) provide an empirical account of how researcher identity and power affect community-based research.

These works do not directly establish what positionality an AI system might possess. They provide the methodological framework for evaluating whether the concept can be meaningfully extended to generative AI.

More directly, Jowsey, Braun, Clarke, Lupton, Fine, and other signatories argued in a 2025 statement that reflexive qualitative approaches require a subjective, positioned, and reflexive human researcher. They reject generative AI use in such approaches as methodologically incongruent.

This position is especially relevant when researchers claim to conduct reflexive thematic analysis through delegated AI interpretation.

Does AI Positionality Matter in Structured Qualitative Coding?

Even when a study does not use a reflexive qualitative approach, the conditions shaping AI output remain relevant.

For example, a structured coding project may use AI to classify interview excerpts according to predefined categories. The researcher can evaluate whether the system applies the categories consistently and whether errors differ across languages or participant groups.

Such evaluation does not require treating the model as a human researcher with positionality.

Instead, the study can examine model behavior, data representation, coding rules, and possible systematic distortions.

Researchers should distinguish methodological questions about the epistemological assumptions of AI-assisted coding from claims about the model possessing a human social identity.

Who Is Accountable When AI Shapes the Interpretation?

AI-generated interpretations may influence which statements receive attention, how participants are categorized, and which explanations appear persuasive.

However, the model does not assume the researcher's professional, ethical, or institutional responsibilities merely because it contributed to the analysis.

Researchers remain responsible for evaluating whether interpretations are supported by the evidence, whether participants are represented fairly, and whether the analytical process fits the methodology.

This responsibility includes accurately describing AI involvement rather than attributing human positionality or lived experience to the system.

When AI Is Asked to Interpret Interviews as a Filipino University Student

What AI Might Claim

The response claims an identity and experience that the model has not established. It also introduces socioeconomic disadvantage as a broad interpretation without sufficient evidence concerning participants' actual economic circumstances.

The statement may contain plausible analytical ideas, but those ideas should not be treated as authentic insider knowledge.

A More Defensible Interpretation

This interpretation does not claim to speak as a participant. It identifies what the excerpts support and distinguishes a possible broader explanation from an established finding.

How the Researcher Should Examine Positionality and AI Influence

The example illustrates why adopting a perspective through prompting should not be confused with possessing the social position from which that perspective is ordinarily developed.

Common Misconceptions About AI and Researcher Positionality

How to Account for AI Without Inventing an AI Researcher Identity

When AI contributes to qualitative research, distinguish the researcher's actual positionality from the technological conditions influencing generated outputs.

The researcher should examine their own relationship to the participants and setting while separately documenting how the AI system was used and what limitations its outputs may introduce.

Do Not Ask AI to Invent a Positionality Statement

A researcher should not use a prompt such as "Write my positionality as an experienced educator committed to social justice" unless those descriptions accurately represent the researcher's own circumstances and commitments.

Where AI-assisted drafting is permitted, a safer instruction is:

"Organize only the factual information I have supplied about my professional role, relationship to the research setting, relevant experiences, and methodological assumptions. Do not invent identities, experiences, values, or relationships. Identify information that remains missing and distinguish my actual positionality from limitations associated with any AI system used in the study."

Even this use may be inappropriate in some reflexive qualitative traditions. Researchers should determine methodological compatibility before using the system.

Examine the Position of the Technology Within the Research Process

Rather than attributing human positionality to AI, researchers can ask how the system's design and deployment affect the analysis.

Relevant questions include whether the model handles participants' languages adequately, whether generated interpretations privilege familiar institutional assumptions, and whether sensitive data are processed through appropriate arrangements.

These questions concern the sociotechnical conditions of analysis. They may be important for understanding how AI influences the production of research findings without treating the model as a human participant or researcher.

Keep Positionality and Methodological Accountability Distinct

A positionality statement should not become a substitute for explaining the analytical procedure. Likewise, an AI-use disclosure does not replace examination of the researcher's relationship to the study.

Where relevant, report both the researcher's positionality and the technological assistance actually used, while recognizing that methodological compatibility must be justified independently.

Follow applicable requirements for reporting AI involvement in qualitative data analysis and avoid implying that a disclosure statement resolves every methodological concern.

Preserve Ethical Responsibilities

Researchers remain responsible for participant confidentiality, appropriate data processing, accurate representation, and compliance with ethics approval and institutional requirements.

AI-generated interpretations concerning cultural or social identities should be evaluated carefully to avoid imposing stereotypes or misrepresenting participants' experiences.

The researcher's accountability cannot be transferred to the model simply because the system generated the analytical language.

Before Making Claims About AI Positionality

Frequently Asked Questions About AI and Positionality

AI Can Simulate Perspectives, but It Does Not Occupy a Human Researcher's Position

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.