Determine how to elicit and interpret evidence of research understanding

Determine how institutions should elicit and interpret evidence of authors' understanding of research manuscripts in a valid, fair, and defensible manner.

Background

The paper situates greCAPTCHA within broader research-integrity policies that require human authors to remain accountable for submitted work. Those policies establish a need for evidence of understanding but do not specify how such evidence should be obtained or how it should support judgments about authorship, verification, or responsibility.

The unresolved problem concerns both assessment design and interpretation: institutions must determine what questions to ask, how those questions should relate to an author's declared contributions, and what conclusions can legitimately be drawn from the resulting responses. The paper's prototype explores one configuration of question families and grading procedures, but leaves the general method for eliciting and interpreting such evidence open.

References

The previous policies establish a need for evidence about authors' understanding, but leave open how to elicit and interpret that evidence.

— greCAPTCHA: Assessing Understanding as Evidence of Research Authorship Under Generative AI  (2609.20481 - Payan et al., 17 Sep 2026) in Section 2, subsection “Designing Assessments of the Capacity to Verify”

How these sources should be combined, and which forms of elicitation are appropriate for different learning objectives, remains an open problem for AI-mediated assessment.

— Same Performance, Different Process: Epistemic Ownership in AI-Mediated Education  (2610.02731 - Fábrega, 2 Oct 2026) in Section 6, Discussion and final remarks