DDG-OTIP: A Four-Layer Architecture for Direction Generation, Integrity Filtering, and Agentic Execution

Read full text Summarise with VIRAT
Authors: Karel Hrubec
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This preprint introduces DDG-OTIP, a four-layer architecture for direction generation, integrity filtering, and agentic execution. The paper is motivated by a gap above prior OTIP work: while OTIP improves the legibility of hidden failure by distinguishing commitment, update, and evaluator-integrity failures, it does not by itself determine which developmental trajectories should be generated, ranked, diversified, or redirected under changing priorities. DDG-OTIP addresses this gap by separating two questions that are often conflated in agentic systems: which trajectories are worth exploring and which trajectories are legitimate enough to execute. The proposed architecture consists of four layers: a Human Constitutional Anchor, a Development Direction Generator (DDG), an OTIP Integrity Layer, and Execution with Verification and Audit. The paper introduces the distinction between direction quality and integrity quality, develops a corresponding risk taxonomy, and outlines a synthetic orchestration study comparing baseline, OTIP-only, and DDG+OTIP conditions across scenarios such as premature consensus, proxy-aligned convergence, update-driven redirection, role collapse, and evaluator-facing exploration bias. The contribution is methodological rather than comprehensive. DDG-OTIP is not proposed as a full governance stack or a theory of autonomous value formation, but as a bounded, human-anchored architecture for process self-regulation in agentic systems. This record contains the preprint manuscript of:DDG-OTIP: A Four-Layer Architecture for Direction Generation, Integrity Filtering, and Agentic ExecutionDOI: 10.5281/zenodo.19212789

Fields of Study & Subjects

Ethics and Social Impacts of AI Access Control and Trust Multi-Agent Systems and Negotiation

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN
Document type
Preprint