<b>Manifesto for a Deliberative Artificial Intelligence</b>

This manifesto outlines the epistemological, architectural, and ethical foundations of a new paradigm in artificial intelligence: <b>Deliberative AI</b>. In contrast to current generative systems—statistically powerful yet epistemically opaque—it proposes a cognitive architecture grounded in structured inquiry, reflexive dialogue, and traceable reasoning.Drawing from pragmatist epistemology, dialogical philosophy, and critical computation studies, this document articulates 22 core theses. It defends the view that knowledge arises from networks of questions, that language is an interpretive and plural technology, and that computation should not automate judgment but sustain it.The manifesto calls for a shift from AI systems that merely generate answers toward those that can structure, inhabit, and reopen meaningful questions. It introduces the concept of <i>hierarchical inquiry complexes</i> and advocates for modular AI architectures capable of adaptive questioning, epistemic navigation, and reflexive learning.Far from being a rejection of computation, this manifesto envisions an <b>AI for judgment, not for closure</b>—an intelligence that collaborates with humans in the construction of knowledge, rather than replacing them with algorithmic certainty.This text is both a conceptual framework and a call to action for researchers, technologists, educators, and institutions committed to the development of responsible, explainable, and pluralistic AI systems.