03 · What You Need to Know
Using Generative AI to Prepare Accurate Participant Information Sheets
What Is a Participant Information Sheet?
A participant information sheet is a document intended to help prospective research participants understand the study before deciding whether to participate.
Depending on the research design and governing framework, it may explain the study's purpose, procedures, expected duration, foreseeable risks, anticipated benefits, confidentiality arrangements, voluntary participation, and relevant contact information.
The information sheet may accompany a separate consent form or form part of an integrated consent document.
Its content and required structure depend on the applicable institutional ethics procedures.
For example, some universities provide approved participant information sheet templates with prescribed headings and statements.
Researchers should follow those requirements rather than assume that a generic AI-generated template satisfies every institution's standards.
What Can Generative AI Actually Help With?
AI can assist with several writing and review activities when the researcher supplies accurate information about the study.
| AI-Assisted Task |
Potential Contribution |
Necessary Researcher Check |
| Organizing the document |
Suggests headings and a logical sequence. |
Compare with the required institutional template. |
| Explaining research procedures |
Converts methodological descriptions into participant-facing language. |
Verify the actual activities and expected time commitment. |
| Simplifying technical terminology |
Suggests more accessible explanations. |
Ensure that the simplified wording preserves the intended meaning. |
| Identifying possible omissions |
Flags topics that may need clarification. |
Determine which requirements actually apply. |
| Reviewing internal consistency |
Highlights potentially conflicting statements. |
Check each statement against the approved protocol. |
| Preparing alternative wording |
Generates different explanations for particular participant groups. |
Evaluate accuracy, appropriateness, and comprehension. |
These activities can improve drafting efficiency, but they do not replace the researcher's responsibility for the content or the institution's approval process.
Why AI Should Not Invent the Research Procedures
Generative AI may produce a plausible information sheet even when the researcher provides only a brief study description.
For example, a prompt stating that the study involves interviews with university instructors may lead the model to assume that interviews last 30 minutes, take place online, and involve audio recording.
None of these details necessarily follows from the research title.
The researcher may actually plan 60-minute face-to-face interviews without recording.
Such discrepancies matter because participants rely on the information sheet to understand what participation involves.
Researchers should provide the verified procedures and instruct AI to identify missing information rather than fill gaps with plausible assumptions.
Distinguish Drafting Assistance From Ethical Determination
AI can suggest wording, but it cannot independently establish that the proposed research procedures satisfy the applicable ethics requirements.
AI-assisted drafting
Preparing language that communicates the researcher's verified procedures and arrangements.
Ethical and institutional determination
Evaluating whether the research procedures and participant information satisfy applicable requirements.
For example, AI may suggest including a statement that participants can withdraw their data at any time.
However, whether that statement is accurate depends on the research design, applicable rights, and limitations associated with data analysis or publication.
The researcher should verify the actual withdrawal conditions rather than accept a generic statement because it sounds ethically appropriate.
What Information Should Researchers Prepare Before Prompting AI?
The quality of the draft depends substantially on the information supplied.
Before using AI, researchers should establish the actual study procedures and relevant institutional requirements.
| Information to Prepare |
Illustrative Detail |
| Research purpose |
Investigating instructors' experiences with institutional AI policies. |
| Participant population |
University instructors meeting specified eligibility criteria. |
| Participation procedures |
One semi-structured interview lasting approximately 45 minutes. |
| Recording arrangements |
Audio recording with the applicable participant agreement. |
| Foreseeable risks |
Possible discomfort when discussing professional experiences. |
| Potential benefits |
No guaranteed direct benefit, with possible contribution to research knowledge. |
| Confidentiality |
Approved access restrictions and data management procedures. |
| Retention |
Verified institutional retention period or applicable criteria. |
| Withdrawal |
Actual procedures and limitations approved for the study. |
| Contact information |
Verified research team and ethics contact details. |
The details are illustrative. Researchers should replace them with the actual conditions of their study.
Where information has not yet been established, the AI system should be instructed to mark it as unresolved rather than invent an answer.
Why Generic Consent Language Can Be Misleading
AI-generated participant information sheets often contain familiar statements about confidentiality, voluntary participation, and withdrawal.
These statements may be appropriate in some studies but inaccurate in others.
For example, the phrase "There are no risks associated with participation" may be misleading when interviews involve sensitive professional experiences.
Similarly, the statement "All responses will remain completely anonymous" may be inaccurate when the research team knows participants' identities or can identify them through contextual details.
Researchers should distinguish conventional consent wording from verified information about their particular study.
A professionally written document is not necessarily a correct document.
Can AI Help Explain Research Risks?
AI may help researchers express identified risks in language that participants can understand.
For example, a study involving instructors' experiences with institutional policies may present possible discomfort or confidentiality concerns.
AI could suggest a clearer explanation of these risks.
However, the model should not independently determine that a study is minimal risk or that no foreseeable harms exist.
Risk assessment requires examination of the actual procedures, population, context, and safeguards.
The research ethics committee or other authorized institutional authority may need to review the assessment.
AI should assist with communicating the established risk information rather than deciding which risks are ethically acceptable.
Can AI Help Explain Confidentiality?
Yes, but confidentiality statements require particular care.
For example, AI may generate a sentence promising that only the principal investigator will access the research data.
However, the actual study may involve research assistants, transcription personnel, institutional administrators, or authorized external processors.
Researchers should verify who may access the information and under what conditions.
They should also distinguish anonymity from confidentiality.
Information can be treated confidentially while participants remain identifiable to the research team.
AI-generated assurances should never be accepted merely because they sound reassuring.
How Should AI Use Be Explained in the Information Sheet?
Where generative AI is materially involved in participant interactions or information processing, the participant information sheet may need to explain that role.
For example, a study may use an authorized AI service to assist with preliminary organization of interview excerpts.
The information sheet should describe the actual processing and relevant safeguards rather than merely state that AI will be used.
Researchers should also avoid claiming that AI-generated interpretations are automatically accurate or that the provider retains no information without verified evidence.
The related question of whether participants should be told about AI use depends on the nature of the proposed activity.
Where disclosure is appropriate, researchers should consider how much AI processing information participants need.
Should Researchers Use an Institutional Template?
Where the university or research ethics committee provides an approved participant information sheet template, researchers should generally use that document as the structural starting point.
For example, an institutional template may contain required headings, standard statements, or contact information that should not be removed.
AI can assist with drafting study-specific content within those requirements.
However, researchers should not ask AI to replace the template with a generic document merely because the generated version appears more polished.
Institutional requirements may reflect particular ethical, legal, and administrative obligations.
The appropriate workflow is to preserve mandatory elements and use AI only where drafting assistance is permitted.
How Can Researchers Prevent AI From Inventing Information?
A useful strategy is to restrict the model to supplied facts and require explicit identification of missing information.
For example, the researcher can provide a structured description of the study and instruct AI not to infer interview duration, compensation, retention periods, or withdrawal conditions.
The model can instead insert placeholders for information that remains unresolved.
However, prompting alone does not guarantee that every generated statement will be accurate.
The researcher must compare the draft against the source documents and verify material claims before submission.
What Is a Useful Prompt for Drafting a Participant Information Sheet?
A useful prompt should identify the required institutional structure and restrict AI to verified research information.
Illustrative AI Prompt
Drafting a Participant Information Sheet From Verified Study Details
"Help me draft a participant information sheet for a research study. I will provide the research protocol, participant population, approved procedures, foreseeable risks, anticipated benefits, confidentiality arrangements, data retention conditions, withdrawal procedures, and institutional template requirements. Follow the required headings and write in clear language suitable for prospective participants. Use only information explicitly supported by the supplied materials. Do not invent interview durations, compensation, risk classifications, privacy guarantees, withdrawal rights, retention periods, or ethics approval. Where required information is missing, insert a clearly labeled placeholder and list the questions that the researcher must resolve. Preserve the substantive meaning of the approved procedures."
This prompt can reduce unsupported assumptions, but the resulting document still requires verification and institutional review.
Researchers should also ensure that any confidential protocol information is submitted only through an appropriately authorized AI environment.
Can AI Check Whether the Information Sheet Is Complete?
AI may help identify topics commonly addressed in participant information materials.
For example, it might notice that a draft explains interview procedures but does not describe recording arrangements or withdrawal conditions.
However, identifying a potentially missing section does not establish that the section is legally required in every study.
The researcher should compare AI suggestions with the institutional template and applicable consent requirements.
AI may also fail to identify important omissions or mistakenly classify optional information as mandatory.
Completeness should therefore be determined through source-based review rather than reliance on the model's checklist alone.
Can AI Simplify the Information Sheet Without Changing Its Meaning?
Generative AI can suggest simpler sentence structures, explain technical terms, and reduce unnecessary complexity.
However, simplification may unintentionally remove important qualifications.
For example, the statement "You may request withdrawal of your identifiable interview data until the dataset has been irreversibly anonymized" should not be simplified into "You can delete your data whenever you want."
The shorter sentence changes the participant's actual rights.
Researchers should therefore compare revised wording with the original meaning and verify that limitations remain accurately expressed.
The issue is explored further in simplifying informed consent language using AI.
Can AI Adapt Information Sheets for Different Participant Groups?
AI may help adjust language for participants with different educational backgrounds, language preferences, or familiarity with research terminology.
For example, a participant information sheet intended for university instructors may use different explanations from one intended for secondary school students.
However, adapting language should not change the underlying procedures, risks, or rights.
Where participants include minors or individuals requiring additional protections, the appropriate consent and assent procedures must be established under the applicable framework.
AI should not independently determine those requirements.
Researchers should evaluate whether the adapted materials are understandable and appropriate for the intended population.
Can AI Translate Participant Information Sheets?
Generative AI may assist with preparing preliminary translations, but informed consent translation requires careful verification.
For example, an English-language information sheet may need to be translated into Filipino or another language used by the participant population.
Terms concerning voluntary participation, confidentiality, withdrawal, and potential risks may not have straightforward equivalents across languages.
A fluent translation can still alter the legal or ethical meaning of a statement.
Researchers should use appropriately qualified language reviewers and follow institutional translation requirements.
Where necessary, comprehension testing may help determine whether the translated material communicates the intended information.
Can AI Assess the Readability of Participant Information Sheets?
AI can identify long sentences, technical vocabulary, and potentially confusing passages.
It may also suggest revisions intended to improve readability.
However, readability is not equivalent to comprehension.
A document may contain short sentences and familiar vocabulary while still presenting confusing withdrawal conditions or data-processing arrangements.
Likewise, an AI-generated readability judgment may not accurately predict how the intended participant population will understand the document.
Researchers should distinguish readability assessment from evidence that participants comprehend the information.
Should Researchers Pilot Test Participant Information Sheets?
Where appropriate, researchers may benefit from asking individuals similar to the intended participants to review draft information materials.
For example, reviewers might be asked to explain what participation involves, whether interviews will be recorded, and under what circumstances their information may be withdrawn.
Such feedback can identify misunderstandings that are not apparent from grammatical or readability analysis.
However, the appropriate form of review depends on the study, participant population, and institutional requirements.
AI-generated predictions about comprehension should not be treated as equivalent to feedback from actual readers.
Who Is Responsible for Errors in AI-Drafted Information Sheets?
The research team remains responsible for the accuracy of the participant information materials it submits and uses.
For example, if AI incorrectly states that all data will be deleted immediately after the study, the researcher should identify and correct the statement before participants receive it.
The use of AI does not transfer responsibility for the content to the provider.
Institutional ethics review may identify additional issues, but it does not eliminate the researcher's obligation to describe procedures accurately.
The responsibility is especially important when AI-generated language concerns participant rights, confidentiality, or potential harms.
The broader question of responsibility for the accuracy of AI-simplified consent language applies equally to AI-drafted information sheets.
What If AI Produces a More Professional-Looking Document Than the Institutional Template?
Visual polish and fluent wording do not establish ethical completeness.
For example, AI may generate a concise, elegant information sheet while omitting mandatory institutional statements or qualifications concerning data withdrawal.
The researcher should prioritize accuracy, compliance, and participant understanding over stylistic preferences.
Where the institutional template permits adjustments, AI may help improve readability and organization.
However, mandatory information should not be removed simply because the resulting document looks cleaner.
The objective is an accurate and understandable information sheet, not merely an attractive one.
