Case for algorithmic auditing in presidential selection

THE EDITOR, Madam:

David Abrikian’s commentary ‘Constitution considerations’ published on September 12, exposes a structural vulnerability within the Constitution (Amendment) (Republic) Bill 2024. As Abrikian and Peter Espeut have rightly noted, a presidential selection mechanism that grants the prime minister ultimate tie-breaking authority, coupled with a 27-seat senate structure where 18 seats align with executive preference, effectively dismantles the two-thirds parliamentary check required to protect entrenched constitutional provisions.

While Abrikian thoughtfully suggests adopting a Singaporean-style Presidential Elections Committee (PEC) comprised of non-aligned institutional leaders, a practical dilemma immediately arises within our compact, highly politicised society: how do we guarantee that any human vetting panel remains entirely free from subtle partisan leverage or perceived bias?

I wish to propose a modern, technical enhancement to this debate: integrating an open-source, mathematically auditable artificial intelligence (AI) evaluation layer to assist the civic selection process.

Rather than substituting human sovereignty or democratic choice with automated decision-making, an AI audit framework would act as an objective, non-partisan filter during the initial vetting phase. Such a system could strengthen Jamaica’s constitutional design in three distinct ways:

1. Anonymised dossier evaluation: Custom algorithms can redact party affiliations, political lineage, and personal networks from submitted candidate profiles, forcing human vetting panels to evaluate nominees strictly on verified professional achievements, governance records, and policy vision statements.

2. Comprehensive conflict mapping: AI models can rapidly cross-reference public registries, directorships, corporate filings, and commercial transactions to surface undisclosed financial conflicts or partisan linkages far more rigorously than manual human inspection permits.

3. Objective merit scoring: Candidates can be benchmarked against pre-agreed constitutional criteria, such as a mandatory cooling-off period from active political officeholding, ensuring uniform enforcement without subjective exceptions.

To preserve public trust, the underlying algorithms, training datasets, and scoring rules must be fully open-source and subject to public cryptographic verification, eliminating concerns over algorithmic manipulation.

As Jamaica prepares to construct its post-monarchical constitutional architecture, we possess a historic opportunity to pair human civic wisdom with emerging technical safeguards. By placing algorithmic auditing at the service of human integrity, we can ensure that the president of a Jamaican republic remains an unassailable custodian of our democracy, completely removed from partisan or ‘billioneering’ capture.

DENNIS A. MINOTT