Welcome to Research — building the evidence base

[vdmeu]

This category is for researchers helping to build the empirical and theoretical foundation for the framework.

The open questions Cupel cannot answer alone:

1. Signal validity: which trust signal types actually correlate with the ability to catch AI errors in specific domains? The multi-signal model assumes diversity improves reliability — is that empirically supported?
2. Gaming resistance: what attack strategies are feasible against each signal type, and how does the multi-signal model respond to them?
3. Below the abstraction layer: can competence at this level be reliably distinguished from surface-level performance? What are the reliable indicators?
4. Consent and privacy: what disclosure architectures preserve both individual privacy rights and organisational need-to-know?
5. Pipeline collapse: what is the evidence for the apprenticeship pipeline contraction claim, and what are its implications for future supervisory capacity?

These are tracked as open issues in the repository. Research that engages with any of them is directly useful.

If you are conducting relevant research and want to coordinate, open a thread here or email contact@cupel.foundation.