This article is part of the Iris Ethics Knowledge Base; you can find a directory of all articles here.
Overview
Research designs that use AI to conduct data collection, including conversational AI systems (chatbots) and agentic interview systems that adapt their questions based on participant responses, raise ethical considerations relating to consent, participant welfare, and the disclosed nature of the research relationship. These considerations are addressed through the National Statement's general provisions on consent (NS 2.2), limited disclosure and deception (NS 2.3.1-2.3.4), and beneficence (NS 1.6-1.9); the National Statement does not contain AI-specific requirements.
Disclosure that the interviewer is AI
Participants must be clearly and directly informed that they are interacting with an AI system rather than a human researcher. This follows from the general requirement that consent be based on sufficient information and adequate understanding of the research (NS 2.2.1-2.2.2); a participant who believes they are speaking with a person is not making an informed decision about the interaction they are actually having.
Where the AI nature of an interviewer is not disclosed, this constitutes active concealment or deception for the purposes of the National Statement's limited disclosure provisions. Such a design may only be approved where it meets the additional conditions in NS 2.3.2, including that participants are not exposed to increased risk of harm as a result of the concealment, and that a full explanation is subsequently made available to participants. Research involving undisclosed AI interaction can only be reviewed and approved by an HREC (NS 2.3.4); it is not eligible for lower risk review or exemption.
Participant welfare
AI interview systems cannot respond to participant distress in the way a trained human researcher can. Beneficence requires that research be designed to minimise risks of harm or discomfort to participants (NS 1.6-1.7); where the research involves sensitive topics, a mechanism must be in place for participants experiencing distress to be connected with human support, whether through the AI system's own escalation pathway or a standing referral arrangement. The absence of a human researcher during data collection does not reduce this obligation.
Data and algorithmic considerations
Agentic AI systems that adapt their line of questioning based on participant responses can introduce variability into the data collection process that affects the reliability of the resulting data. Because merit requires that methods be appropriate to the research aims (NS 1.1(b)), the AI system's behaviour, including the boundaries within which it can vary its questions, must be sufficiently defined and documented for a reviewer to assess the research design.
AI-mediated interactions may also generate data beyond what participants anticipate, including inferences drawn from conversational patterns, tone, or response timing. This must be addressed in the data management plan (NS 3.1.44), and participants should be told, in plain language, what data the interaction will generate and how it will be used (NS 2.2.6(f)).
Data security and third-party platforms
Where a chatbot or agentic interview platform is provided by a third party, researchers should confirm its data handling arrangements, including where data is processed and stored, before any participant data is collected. This is a data management plan requirement (NS 3.1.44) and should be disclosed to participants as part of the consent process.
What reviewers look for
Reviewers look for clear evidence that participants are told, at the point of interaction, that they are engaging with an AI system; a defined escalation pathway for participant distress; and documentation of the AI system's scope of variation sufficient to assess the reliability of the resulting data.
Related resources
This page was last updated on 16 September 2026.
Was this helpful? We'd appreciate your feedback!
This article makes reference to the National Statement on Ethical Conduct in Human Research 2025. It is intended as practical guidance and does not constitute legal advice.
AI Disclosure: Resources in this library were prepared using Large Language Models with input from Iris Ethics staff who guided the scope and design. Subsequent revisions and final versions were developed and approved by Iris Ethics staff.