<p>Manufacturing engineers face decisions for which typical analytical methods are unsuitable. Material cost frequently conflicts with environmental impact, and the imperative of a circular economy introduces a further challenge that standard evaluation processes have historically disregarded. Most multi‑criteria decision‑making (MCDM) systems assume static criteria and complete information, yet real‑world procurement is rarely the case. Teams default to familiar materials not because those are optimal, but because searching supplier lists and technical literature takes time – a cost that most decision tools ignore. This research responds directly to that oversight. Before MCDM is applied, generative AI scans technical literature, supplier catalogues, and patent databases to surface non‑dominant material candidates that a formal step‑by‑step assessment would miss. Once a candidate set is established, MCDM evaluates cost and emissions, with circularity treated as a secondary consideration rather than a co‑equal criterion. This sequencing alters which trade‑offs become visible. The framework does not deliver a single definitive solution but rather a justifiable decision pathway that can be defended under scrutiny. It assumes that ethical and social equity constraints are set by external stakeholders outside the technical decision process. The paper also relates firm‑level choices to urban sustainability metrics in line with the UN U4SSC objective. By specifying where AI enters the decision cycle and where it does not, the paper offers a conceptual architecture for material selection under genuine uncertainty.</p>