AI Evaluation of the Drake-S Equation for Postmortem Survival Against Sudduth’s Evidentiary Standards

Abstract

We applied two advanced large‐language models—OpenAI’s ChatGPT-4o and GitHub Copilot—to evaluate the Drake–S Equation framework for postmortem survival of consciousness relative to eight previously specified evidentiary standards. ChatGPT-4o judged this equation approach a “Good” to “Very Good” fit on 75% of the criteria (23/32 points, 72% overall), whereas Copilot rated it a “Good” to “Very Good” fit on 88% of the criteria (26/32 points, 81% overall). Both programs flagged two consistent weaknesses: the use of legal benchmarks (e.g., Daubert standard) as an evidential standard and the apparent assumption that gaps in known confounds imply the increased odds of survival, reflecting concerns about inappropriate criteria and probabilistic fallacies. Each program suggested methodological refinements—chiefly, estimating covariation among error terms, pre-registering and calibrating scoring rubrics, and embedding Bayesian updating (e.g., priors, likelihood ratios, and credible intervals)—to strengthen the framework’s logical rigor and evidentiary justification. Although the Drake–S approach does not demonstrate ontological survival, it highlights a substantial residual (30.3%) of data that ostensibly eludes major materialist and psi‐only explanations. Rigorous study of these anomalies should therefore continue in anticipation of chance- or challenge- type discoveries that can meaningfully advance our understanding of the nature or limits of human consciousness.