Maikel Leyva

This book addresses a fundamental epistemological limitation of contemporary artificial intelligence systems: their inability to represent and communicate uncertainty. Current AI architectures, particularly large language models, generate outputs based on probabilistic pattern matching without an intrinsic mechanism to distinguish between verified knowledge, partial knowledge, and fabricated information. This results in what the authors define as “confident machines,” capable of producing fluent yet potentially unreliable responses without signaling epistemic boundaries. To address this structural deficiency, the book introduces a practical and conceptual framework grounded in neutrosophic logic, a triadic system developed by Florentin Smarandache, in which every proposition is characterized by degrees of truth (T), indeterminacy (I), and falsity (F). The authors extend this formalism beyond mathematics, integrating it with philosophical traditions from Latin America and other non-Western epistemologies that embrace ambiguity, complementarity, and productive doubt. The work develops an operational methodology—referred to as “The Third Answer”—which enables users to critically evaluate AI-generated outputs through structured questioning, classification of uncertainty, and decision protocols. By translating abstract logical principles into accessible tools (e.g., decision frameworks, uncertainty mapping, and prompt strategies), the book provides a practical guide for professionals interacting with AI systems across domains. Ultimately, the book argues that improving AI reliability is not merely a matter of increasing data or model size, but requires a paradigm shift toward uncertainty-aware reasoning. It positions neutrosophic logic as a bridge between ancient philosophical insights and modern computational systems, offering a pathway toward more transparent, trustworthy, and epistemically responsible artificial intelligence.