Can an AI Chatbot Interview Research Participants?

Can a Chatbot Conduct an Actual Research Interview?

Suppose you are conducting qualitative research involving 50 teachers. Instead of scheduling individual interviews, could you provide participants with a chatbot that asks your interview questions, listens to their responses, and asks for clarification when necessary?

Conversational AI can already perform many of these activities. The methodological question, however, is whether the resulting conversations constitute defensible research interviews and whether the chatbot can collect the evidence your study actually needs.

Yes, but Interviewing Requires More Than Asking Questions

How AI Chatbot Interviews Work in Research

What Is an AI-Mediated Research Interview?

An AI-mediated research interview is a data collection interaction in which an artificial intelligence system performs some or all of the conversational activities ordinarily undertaken by a human interviewer.

The participant is still a real person providing information about their experiences, perceptions, or opinions. The chatbot serves as the interviewing interface, rather than as a substitute participant.

In a typical implementation, researchers prepare an interview protocol containing the study objectives, primary questions, permissible follow-up questions, and rules governing the interaction. The chatbot presents questions, processes participants' answers, and determines how to proceed according to its configuration.

Some systems use predetermined conversational rules. Others employ large language models (LLMs), which can generate responses and follow-up questions based on the preceding conversation. These approaches differ considerably in flexibility, predictability, and methodological control.

What Can an AI Interviewer Actually Do?

Depending on the system, an AI interviewer may be able to:

• Introduce the interview and explain the procedure using researcher-approved language.
• Present structured or semi-structured interview questions.
• Recognize when a response is incomplete or requires clarification.
• Ask additional questions within predetermined boundaries.
• Maintain conversational context across multiple exchanges.
• Record participant responses and relevant interaction metadata.
• Allow participants to complete interviews remotely or asynchronously.
• Identify predefined situations requiring human review or intervention.

These are potential system capabilities, not guarantees. A general-purpose chatbot does not necessarily provide reliable interview administration, secure storage, participant authentication, or research-grade records without additional configuration.

What Does Published Research Show?

AI-mediated interviewing has moved beyond hypothetical applications, although evidence concerning its performance remains context-dependent.

Wuttke and colleagues (2025) investigated AI conversational interviewing in a small-scale study involving university students randomly assigned to AI or human interviewers using identical questionnaires on political topics. Their findings suggested that AI interviewing could produce data of comparable quality within the studied conditions while offering opportunities for more scalable data collection.

Importantly, this finding should not be interpreted as proof that AI interviewers are interchangeable with human researchers across populations, interview formats, or sensitive research settings. The study evaluated a particular implementation under controlled conditions.

Gårdhus and colleagues (2026) subsequently presented AInterviewer, a platform designed to combine controlled question administration with LLM-based conversational flexibility. Their work addresses practical challenges involving question standardization, reproducibility, and data security. It also illustrates why research interviewing systems require more deliberate design than simply instructing a general chatbot to conduct an interview.

These developments establish the technical feasibility of AI-led interviewing. Whether an implementation generates credible evidence for a particular research question still requires empirical evaluation.

Structured, Semi-Structured, and In-Depth Interviews

The appropriateness of an AI interviewer partly depends on the interview format.

Structured interviews may be easier to control because question wording and sequence can be specified in advance. Semi-structured interviews introduce greater flexibility, but also more opportunities for the AI system to depart from the intended protocol.

In-depth qualitative interviewing presents additional challenges. An experienced human interviewer may notice hesitation, recognize an unexpected but significant detail, or reconsider an assumption during the conversation. An AI system can generate language resembling these interviewing behaviors without necessarily interpreting the participant's situation appropriately.

This distinction becomes especially relevant when considering whether AI and human interviewers are methodologically equivalent.

How Much Autonomy Should the Chatbot Have?

Researchers can configure AI-mediated interviews with different levels of conversational autonomy.

These are useful design distinctions rather than universally standardized categories. Hybrid systems may preserve fixed core questions while allowing carefully restricted generative probing.

For example, an interviewer might ask every participant, "What difficulties have you experienced when using AI tools for teaching?" If a participant mentions unreliable answers, the system could ask for a specific example. That follow-up may be useful, provided it does not introduce assumptions or lead the participant toward a preferred interpretation.

The methodological challenge of generating follow-up questions dynamically deserves particular attention because the chatbot's decisions can affect what evidence is ultimately collected.

What Makes an AI-Led Conversation a Credible Research Interview?

A transcript alone is insufficient evidence that a sound interview procedure occurred. The interaction must be appropriate for the study's methodological assumptions and capable of producing information relevant to its research questions.

Researchers should evaluate whether the chatbot consistently covers required topics, asks comprehensible and non-leading questions, responds appropriately to unexpected answers, and preserves participants' intended meanings.

For qualitative studies, this also involves examining whether the interviewing process permits participants to introduce perspectives that the researchers did not anticipate. An AI system that repeatedly redirects answers toward predefined categories may restrict the very experiences the study intends to explore.

Consequently, validating the AI-mediated data collection process should be part of the study design rather than an assumption made after interviews have been completed.

What About Informed Consent, Privacy, and Ethical Approval?

Using a chatbot does not eliminate the ethical responsibilities associated with human participant research. Applicable institutional review procedures, informed consent requirements, and data protection obligations continue to matter.

Participants should receive understandable information about the interview procedure, the role of AI, the handling of their responses, foreseeable risks, and the arrangements for contacting the research team. They should also know how to decline questions or withdraw in accordance with the approved protocol.

Where an external AI service processes interview content, researchers need to establish what information is transmitted, where it is stored, who can access it, whether it may be retained or used for model improvement, and which contractual or technical safeguards apply.

The U.S. Office for Human Research Protections provides guidance on informed consent and electronic consent, while the Secretary's Advisory Committee on Human Research Protections has examined issues arising from AI in human subjects research. These sources offer relevant principles, although the specific requirements applicable to a study depend on its jurisdiction and institutional context.

Researchers should also explicitly disclose when participants are interacting with an AI interviewer, rather than allowing them to assume they are speaking with a human researcher.

What Happens When the Interview Goes Beyond the Script?

A participant may ask an unexpected question, disclose information unrelated to the study, express discomfort, or describe a potentially dangerous situation.

Researchers should not assume that a chatbot can independently determine the correct response to every such event. Its fluent conversational output may create an appearance of understanding even when it has misinterpreted the situation.

Before deployment, the research team should establish escalation procedures, stopping rules, and a human contact mechanism. Studies involving sensitive personal experiences may require additional safeguards or a different interviewing approach altogether.

For research involving sensitive disclosures, consider both the possibility of inappropriate responses to participant distress and the need for a predetermined procedure when immediate human intervention may be necessary.

Using an AI Chatbot to Interview Teachers About Educational Technology

Consider this hypothetical exchange:

The follow-up question is potentially useful because it invites a concrete explanation without prescribing the participant's answer. The response provides additional context about resource constraints that might otherwise remain unexplored.

However, one successful exchange does not establish the reliability of the entire interviewing system. The researcher must examine how it performs across different responses, participants, and unexpected conversational situations.

Common Misunderstandings About AI Interviewers

When Should You Consider an AI Interviewer?

AI-mediated interviewing may be worth considering when your study involves a clearly defined interview protocol, participants who can comfortably use the interface, and questions that do not routinely require sensitive human judgment.

It may be particularly useful for geographically dispersed participants, asynchronous interviewing, or studies in which consistent administration of core questions is important. However, convenience and potential scalability should not override the study's epistemological and methodological requirements.

One useful question is whether your research treats the interview primarily as a means of obtaining responses to specified questions or as an interpretive encounter in which meaning develops through interaction. AI may be easier to justify in the former situation. In the latter, the interviewing relationship itself may be integral to the evidence.

Before committing to AI-led interviewing, document the intended system behavior, test its performance, and decide what constitutes an unacceptable deviation from the protocol. Preserve sufficient records of AI-participant interactions to support subsequent methodological scrutiny without collecting unnecessary personal information.

The objective is not to demonstrate that a chatbot can complete an interview. It is to establish that the interview procedure can answer the research question responsibly.

Before Allowing an AI Chatbot to Interview Participants

Frequently Asked Questions About AI Chatbot Interviews

AI Can Conduct the Conversation, but Researchers Must Establish Its Credibility

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.