Qualitative Contributions of Dsaive - Ai-Driven Hierarchical Literature Review On Sustainable Fuels

While quantitative metrics often validate the accuracy of artificial intelligence (AI) in systematic literature reviews (SLRs), they fail to capture its qualitative utility as a decision-support tool. This article presents a qualitative analysis of the Dynamic Systematic AI Vector Engine (DSAiVE), a framework designed to synthesize multidisciplinary research on sustainable fuels. Through the analysis of a corpus of 478 engineering article abstracts and 118 expert validations, we assess the framework's ability to move beyond simple binary keyword filtering. Results indicate that while AI-generated scores correlate with human judgment on core topics, the AI captures a "long tail" of contextual relevance often discarded by human reviewers. However, user feedback reveals a critical dichotomy: high trust in the AI’s ability to summarize provided content, contrasted with significant hallucinations when generating external "State of the Art" references. This validates the theoretical view of LLMs as "stochastic parrots" that require strict structural constraints to function in high-stakes domains. This paper settles a solid starting point on the path to build a reliable information support provider for complex decision-making, arguing that AI should not be viewed as an oracle, but as a "wide-angle" semantic engine guided by researcher-defined anchors. Starting from a narrow goal of synthesising information on sustainable fuels, it lays the groundwork for a controlled and analysable Core AI engine.