Theory paper · Framework
Protecting the quality of leadership decisions in the age of artificial intelligence
| Author | Francesco Saverio Canepa |
|---|---|
| Type | Theory / framework paper — version 1.0 (2026) |
| Status | Public OSF project · osf.io/6n8tg · DOI 10.17605/OSF.IO/6N8TG · CC-BY 4.0 |
| Companion | Calibrated Dissent · OSF |
Leadership teams working with intensive AI assistance face a paradox: more analysis does not mean better judgment. AI systems optimised for user satisfaction tend to flatter — they sift the evidence for support of the thesis the decision-maker already favours, inflate confidence, and, when many organisations use the same tools, homogenise reasoning. The result is not augmented analysis: it is a silent erosion of independent judgment, dressed up as rigour.
This paper proposes a construct to name and protect what is at stake: cognitive sovereignty — the collective capacity of a board, a committee or a deliberative function to keep its judgment independent, calibrated and falsifiable even when AI tools push the other way. It integrates four established research streams — sycophancy in RLHF models, automation over-reliance, deskilling, and algorithmic monoculture — into four operational dimensions: independence of judgment, resistance to homogenisation, calibration of uncertainty, and traceability of the burden of proof.
The proposed protective mechanism is the cognitive contradictor: a system whose mandate is not to assist reasoning but to challenge it. Four invariant principles separate it from a sophisticated validator — structural adversariality, fail-closed on flattery, a fixed output schema, and separation of provenance from scoring — and from them the paper derives a governance framework for boards: mandatory challenge above a materiality threshold, a prospective deliberative ledger, separation between who decides and who challenges, and calibration verified over time.
What the paper does NOT claim
It is a theoretical proposal to be operationalised, not a validated result. It does not claim that the contradictor improves decisions, nor that the proposed dimensions are already measured empirically. The public N = 10 pilot in the companion work is explicitly non-evidence of efficacy. This caution is deliberate: the quality of judgment is not asserted — it is made verifiable.
Theoretical companion to Calibrated Dissent (Canepa, 2026), pre-registered on OSF, where the measurement protocol and reference implementation, CounterBrain, are documented.
Read the full paper
Full version (PDF) and reference project.
