Towards AI epidemiology: a measurement standardisation framework for prospective risk detection
Kit Tempest-Walters
Why It Matters
What makes this one worth your time
This framework could enhance the reliability and governance of AI systems by providing a structured approach to evaluate expert-AI interactions.
A framework for standardising measurements in AI risk detection.
Summary
The paper proposes a measurement standardisation framework for assessing expert-AI interactions in risk detection, outlining its scope, empirical testing protocol, and initial reliability findings.
Key contributions
- Development of a measurement standardisation framework for expert-AI interactions.
- Specification of a protocol for empirical testing of the framework.
- Introduction of a statistical method for assessing alignment scores.
Notable insights
- The concept of 'AI epidemiology' introduces a novel perspective on risk detection based on correlated variables rather than mechanistic analysis.
- The use of a defined grammar of interaction fields and a statistical protocol for validation could standardise assessments across diverse AI applications.
Possible limitations
- Judge reliability at scale remains to be validated in future work.
Abstract
arXiv:2512.15783v4 Announce Type: replace Abstract: This paper proposes a measurement standardisation framework that compresses expert-AI interactions into structured, comparable fields for prospective risk detection in deployed AI systems, without access to model internals. This concept paper defines the framework's scope, semantically and statistically, and specifies a protocol for its empirical testing. The population-level claims it is designed to support therefore belong to a staged research programme rather than to results claimed here. Measurement standardisation underpins three claims. The first is a reliability claim: under bounded conditions, large language models can produce reliable, standardised assessments of the evidential and policy alignment of expert-AI interactions. The second is a governance claim: alignment scores give experts an immediate signal during deployment and give institutions a basis for monitoring alignment patterns across mission types, models, and domains. The third is an outcome validation claim: once measurement standardisation is established, aggregate alignment scores could be used to study associations with downstream outcomes in regulated professional settings. This introduces the possibility of an "AI epidemiology", a form of risk detection based on correlated variables instead of mechanistic analysis, inspired by epidemiological reasoning. A minimal application of the protocol to a published expert-AI corpus shows that the judge reproduces its policy and evidential alignment scores across two runs under the specified conditions. Judge reliability at scale remains to be validated in future work. The paper sets out a defined grammar of eight interaction fields, together with a statistical protocol based on paired bootstrap inference, DeLong's test for paired AUCs as a sensitivity check, a pre-specified one-sided non-inferiority margin of 0.05, and Holm-Bonferroni correction.