AI Research Summary
Before Human ReviewBased on abstract · Full text not reviewed
This abstract argues that the unit of evaluation for health chatbots should not be the model output, but rather the 'constituted response system' including the LLM, source selection, retrieval, interface, action routing, and subsequent data return. The authors call this 'synthetic gatekeeping' and discuss that even when the content is the same, the actual executability of the pathway may differ. This is an abstract-based summary and the full text was not reviewed.
Key Points from the Abstract
- It proposes that the unit of evaluation is the constituted response system rather than the model output.
- It argues that content matching does not guarantee equivalence in pathway matching or behavioral feasibility.
- It proposes component comparison, organizational disclosure, and proportional governance review.
Relevance to AI Law and Policy
It is relevant to regulatory and audit frameworks for evaluating information provision, appointment prompting, and organizational bias in medical chatbots.
Limitations to Consider
- This is a theoretical paper presented on the basis of the abstract alone; the full text was not reviewed.
- Empirical evidence and effect size cannot be sufficiently confirmed from the abstract alone.
- Any unspecified limitations cannot be determined from the abstract.
Please compare with the original text before citing or using it in your assessment.
Abstract Preview
Two health chatbots can give the same clinically appropriate advice and still constitute different interventions:…Read more in the abstract and original text Research Topic
Original text and source
Synthetic gatekeeping in health chatbots: configured answer systems as the unit of AI evaluation and governance for medical information
The Korean title on this page was translated by AI.- Database
- Crossref