Why English? | Applying GUIDE-LLM to Behavioural Science Through Model Identity, Prompts, Human Validation, Bias, Ethics and Reproducibility

ORIENTATION

What this guide owns

Current fact check

GUIDE-LLM was published in Nature Human Behaviour 10, pages 1182–1186, in 2026 (DOI 10.1038/s41562-026-02492-7; PubMed 42265331). The official checklist has 14 core items plus optional items and identifies the current living-document version as v1.0 (December 2025).

Canonical boundary

This article owns reporting of LLM use in behavioural and social science. DEAL owns pathway-based technical development reporting, TRIPOD-LLM owns health research and GUIDE-LLM remains distinct from guidelines about generative-AI disclosure in medical manuscripts.

Worked example

Every chapter returns to a fictional behavioural-science study comparing human and LLM judgements about cooperative choices in short public scenarios. It is fictional, low-stakes and educational; it does not report a real school, clinical decision or guaranteed outcome.

Warm up with eduKate’s guides to explaining scientific evidence and asking better questions in class.

CHAPTER 1 OF 18

1. Name the LLM Role in the Research

Back to contents

Why this move matters

State whether the model annotates data, simulates responses, delivers an intervention, generates stimuli or is itself evaluated.

Name what the LLM does in behavioural science

State whether the model annotates data, simulates responses, delivers an intervention, generates stimuli or is itself evaluated. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The model produces comparison judgements; it is not described as a human participant. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

A clear role can still be scientifically invalid or ethically unsuitable. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Name the LLM Role in the Research”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 2 OF 18

2. Distinguish Human-in-the-Loop From Fully Automated Work

Back to contents

Why this move matters

Report where people design, monitor, validate, override or interpret outputs and where automation proceeds without review.

Name what the LLM does in behavioural science

Report where people design, monitor, validate, override or interpret outputs and where automation proceeds without review. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Researchers inspect generated stimuli and validate a sample of judgements before analysis. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Human involvement does not automatically make an automated process safe. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Distinguish Human-in-the-Loop From Fully Automated Work”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 3 OF 18

3. Define the Behavioural Construct Independently

Back to contents

Why this move matters

Explain cooperation, trust, emotion or another construct before mapping model text to a score.

Name what the LLM does in behavioural science

Explain cooperation, trust, emotion or another construct before mapping model text to a score. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Cooperation is defined through choices in prespecified scenarios, not by agreeable tone. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

An operational definition can still capture only part of the construct. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Define the Behavioural Construct Independently”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 4 OF 18

4. Justify Why an LLM Is an Appropriate Method

Back to contents

Why this move matters

Compare with surveys, experiments, human coding or simpler text methods and state the evidence the model can and cannot provide.

Name what the LLM does in behavioural science

Compare with surveys, experiments, human coding or simpler text methods and state the evidence the model can and cannot provide. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The LLM comparison explores method sensitivity rather than replacing participant recruitment. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Lower cost is not evidence of construct validity. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Justify Why an LLM Is an Appropriate Method”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 5 OF 18

5. Report Model Name, Provider, Size and Version

Back to contents

Why this move matters

Give the exact model identity, provider, parameter size when known, version or ID, source link and date of access.

Name what the LLM does in behavioural science

Give the exact model identity, provider, parameter size when known, version or ID, source link and date of access. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The report records a dated API model and a local checkpoint hash. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Commercial services may change without a visible new name. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Report Model Name, Provider, Size and Version”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 6 OF 18

6. Describe API, Web or Local Access and Session Mode

Back to contents

Why this move matters

State interface, account context, separate calls or chat, state retention, tools and hidden system constraints when known.

Name what the LLM does in behavioural science

State interface, account context, separate calls or chat, state retention, tools and hidden system constraints when known. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Each scenario uses a fresh API call with no previous conversation. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

A fresh session cannot remove pretraining exposure. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Describe API, Web or Local Access and Session Mode”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 7 OF 18

7. Preserve Parameters, Runs and Randomness

Back to contents

Why this move matters

Report temperature, token limits, sampling, seeds, repetitions and aggregation across outputs.

Name what the LLM does in behavioural science

Report temperature, token limits, sampling, seeds, repetitions and aggregation across outputs. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Every scenario runs five times and the study reports variation as well as the modal choice. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Repeated generations are not independent human participants. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Preserve Parameters, Runs and Randomness”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 8 OF 18

8. Describe Fine-Tuning and Customisation

Back to contents

Why this move matters

Report adapters, retrieval, examples, external tools and training data used to change or condition the model.

Name what the LLM does in behavioural science

Report adapters, retrieval, examples, external tools and training data used to change or condition the model. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The study uses no fine-tuning and reports that negative fact explicitly. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

No project-specific tuning does not mean the base model is neutral. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Describe Fine-Tuning and Customisation”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 9 OF 18

9. Publish Exact Prompts and System Instructions

Back to contents

Why this move matters

Provide system and user prompts, examples, ordering, delimiters and output schemas with stable versions.

Name what the LLM does in behavioural science

Provide system and user prompts, examples, ordering, delimiters and output schemas with stable versions. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The full cooperation scenario template and response format appear in an appendix. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Exact prompts may still reproduce differently after a service update. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Publish Exact Prompts and System Instructions”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 10 OF 18

10. Protect Personal and Sensitive Data

Back to contents

Why this move matters

State whether inputs include personal data, de-identification, consent, retention, approved environments and response to accidental disclosure.

Name what the LLM does in behavioural science

State whether inputs include personal data, de-identification, consent, retention, approved environments and response to accidental disclosure. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Only public fictional scenarios enter the model service. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Synthetic data can still echo stereotypes or reveal source patterns. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Protect Personal and Sensitive Data”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 11 OF 18

11. Validate Outputs With Humans and External Evidence

Back to contents

Why this move matters

Describe reference samples, reviewer expertise, blinding, agreement, error categories and how validation affected the analysis.

Name what the LLM does in behavioural science

Describe reference samples, reviewer expertise, blinding, agreement, error categories and how validation affected the analysis. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: A diverse human panel rates whether the model’s labels match scenario definitions. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Human agreement is not proof that a model represents populations. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Validate Outputs With Humans and External Evidence”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 12 OF 18

12. Report Filtering, Reformatting and Post-Processing

Back to contents

Why this move matters

Document parsers, exclusions, translations, coding, aggregation and manual edits between raw output and analysed variable.

Name what the LLM does in behavioural science

Document parsers, exclusions, translations, coding, aggregation and manual edits between raw output and analysed variable. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: Malformed responses remain counted and a versioned parser extracts the choice. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Post-processing can create the result it appears merely to clean. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Report Filtering, Reformatting and Post-Processing”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 13 OF 18

13. Assess Leakage, Bias and Systematic Difference

Back to contents

Why this move matters

Check benchmark exposure, demographic stereotypes, language effects and whether output differences could distort conclusions.

Name what the LLM does in behavioural science

Check benchmark exposure, demographic stereotypes, language effects and whether output differences could distort conclusions. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The study varies names and dialect cues, then reports unstable and biased patterns without treating them as human norms. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

A finite audit cannot prove absence of bias. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Assess Leakage, Bias and Systematic Difference”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 14 OF 18

14. Share Code, Calls and Computational Context

Back to contents

Why this move matters

Release code or notebooks for model calls, prompt files, parsing, package versions and enough logs to rebuild the workflow.

Name what the LLM does in behavioural science

Release code or notebooks for model calls, prompt files, parsing, package versions and enough logs to rebuild the workflow. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: A repository contains scripts, hashes and redacted response records. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Open scripts cannot freeze a proprietary endpoint. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Share Code, Calls and Computational Context”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 15 OF 18

15. Disclose Funding, Support and Relationships

Back to contents

Why this move matters

Report funders, provider support, credits, access arrangements and relevant personal or commercial relationships.

Name what the LLM does in behavioural science

Report funders, provider support, credits, access arrangements and relevant personal or commercial relationships. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The paper states who paid for API use and whether the provider reviewed the work. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Disclosure does not prove independence. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Disclose Funding, Support and Relationships”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 16 OF 18

16. State Ethical Implications and Representation Limits

Back to contents

Why this move matters

Discuss simulation claims, anthropomorphism, participant displacement, labour, environmental cost and groups whose behaviour is not represented.

Name what the LLM does in behavioural science

Discuss simulation claims, anthropomorphism, participant displacement, labour, environmental cost and groups whose behaviour is not represented. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The conclusion says the model is a text-generating system conditioned by training data, not a population sample. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

Ethical reflection cannot legitimise a harmful design by itself. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “State Ethical Implications and Representation Limits”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 17 OF 18

17. Use the Living Checklist Transparently

Back to contents

Why this move matters

Record checklist version, access date, item locations and future corrections when the living document changes.

Name what the LLM does in behavioural science

Record checklist version, access date, item locations and future corrections when the living document changes. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: The supplement states GUIDE-LLM v1.0, accessed 6 November 2026. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

A newer checklist does not invalidate every older study, but comparisons need version awareness. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Use the Living Checklist Transparently”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

CHAPTER 18 OF 18

18. Turn Documentation Into Behavioural-Science Literacy

Back to contents

Why this move matters

Practise separating generated output, measured human behaviour, theoretical interpretation and policy implication.

Name what the LLM does in behavioural science

Practise separating generated output, measured human behaviour, theoretical interpretation and policy implication. State whether the model is a research tool, simulated participant, intervention, annotation system or object of study.

Freeze the research condition

Worked example: A student compares model and human responses, then writes why disagreement is informative rather than declaring a winner. Record model identity, access route, parameters, prompts, conversation state and human involvement.

Test validity and ethics together

Check human validation, post-processing, leakage, bias, sensitive data, incentives and the limits of treating generated text as behaviour.

Preserve reproducibility without overclaiming

A classroom demonstration cannot establish population behaviour. GUIDE-LLM is a living reporting checklist. Complete documentation cannot make an LLM a representative human sample or settle ethical questions by itself.

Student, parent and career transfer

Make one evidence card for “Turn Documentation Into Behavioural-Science Literacy”: source location, plain-language meaning, limit and next question. A parent or teacher can ask for one everyday explanation and one specialist explanation; both must preserve the same conditions. This habit supports psychology, sociology, political science, communication, economics, computational social science, ethics and research methods without promising a grade, admission, funding decision or job.

PRACTICE LAB

Build a guide_llm decision trail

Student route

1. Choose one original report inside this guide’s scope.
2. Copy its objective and underline actor, action, context, evidence and intended decision.
3. Draw the specific reasoning chain used in this article.
4. Mark each step reported, inferred, contested or unknown.
5. Rewrite the conclusion so it preserves the narrowest important limitation.

Parent or teacher route

Ask the learner to explain treat an llm as a documented research condition, not a mysterious stand-in for human behaviour twice: once to a classmate and once to a method specialist. Praise a precise correction and a well-stated unknown; both show disciplined attention to evidence.

Early-career route

Create a one-page guide_llm audit with source link, date checked, scope, method logic, decisive wording, contrary evidence, limit and next action. A colleague should be able to continue without guessing.

Continue with eduKate’s guides to reading course and career information and giving useful feedback in group projects.

USEFUL NEXT READING

Connect this narrow owner to its neighbour

Read the CLIX-M explanation-evaluation guide for a related but different reader job. Return here when the question is specifically how to treat an llm as a documented research condition, not a mysterious stand-in for human behaviour. This boundary keeps one canonical owner per intent.

For English practice across research, school and career communication, visit the English at work and in public life guide.

FREQUENTLY ASKED QUESTIONS

Questions students and parents often ask

What does GUIDE-LLM stand for?

Guidelines for the Use of LLMs in Behavioural and Social Science.

How many core items are there?

Fourteen core items, plus optional items for additional context.

Can LLM outputs represent human participants?

Not automatically. Models generate text from training and prompting conditions and are not population samples.

Why report conversation state?

Earlier messages or retained state can change later outputs.

What does human validation prove?

It checks a defined task; it does not prove representativeness or ethical acceptability.

Is GUIDE-LLM fixed forever?

No. Its official site describes it as a living document and lists version 1.0 from December 2025.