Skip to Content
EnergiesEnergies
  • Article
  • Open Access

16 November 2026

33 Pages

From Prediction to Assurance: An Evidence-Centered Decision Assurance Framework for Trustworthy AI in Energy Markets and Sustainable Finance

and
1
Institute of Marine and Environmental Sciences, University of Szczecin, Wąska 13, 71-415 Szczecin, Poland
2
Faculty of Data Science and Information, INTI International University, Persiaran Perdana BBN, Putra Nilai, Nilai 71800, Negeri Sembilan, Malaysia
3
Faculty of Navigation, Maritime University of Szczecin, Waly Chrobrego 1-2, 70-500 Szczecin, Poland
*
Author to whom correspondence should be addressed.

Abstract

Artificial intelligence (AI) can predict energy-market and sustainable-finance outcomes accurately without thereby being justified in making consequential decisions. This conceptual study proposes the Evidence-Centered Decision Assurance Framework (ECDAF), which treats decision materiality—the potential financial, market, consumer, environmental, and regulatory consequences of reliance on a particular AI-supported action—as the starting point for assurance. ECDAF maps this profile to authority-specific critical claims, required evidence, enforceable controls, accountable oversight, and time-dependent renewal conditions. Its contribution is a sector-oriented instantiation of established claims–evidence–authority governance, rather than a new non-compensatory rule or authority taxonomy. A worked illustration anchors automated bidding in the EPEX SPOT day-ahead auction and contrasts ECDAF’s decision-specific outputs with CI-AIGAF. Two further illustrations examine consumer-facing energy decisions and renewable-project financing, including a financially attractive proposal with unverified sustainability evidence. The examples clarify the framework’s internal decision logic; they are not independent tests of validity, deployment feasibility, or superior performance. Numerical thresholds, inter-assessor consistency, implementation costs, and prospective effectiveness require empirical calibration and evaluation.

Article Metrics

Citations

Article Access Statistics

Multiple requests from the same IP address are counted as one view.