Artificial Intelligence-Assisted Design of High-Temperature High-Entropy Alloys: Progress, Challenges, and Perspectives

High-temperature high-entropy alloys (HTHEAs) provide a large compositional space for tuning phase constitution, microstructure, elevated-temperature strength, creep resistance, and oxidation behavior. Their design is difficult for the same reason: performance depends on processing history, microstructure, temperature, loading mode, exposure time, and environment as well as nominal chemistry. Artificial-intelligence (AI) and machine-learning (ML) methods are increasingly used to navigate this coupled space. This structured narrative review is organized around service-relevant high-temperature evidence rather than algorithm families, covering phase and microstructure stability, elevated-temperature strength, creep, oxidation, and the extent to which AI-selected alloys have been experimentally validated under the intended conditions. Evidence is assessed from internal statistical validation and physics-based filtering to alloy synthesis, direct high-temperature testing, extended service-relevant exposure, and iterative experiment–model feedback. Prospective studies now include tests of model-selected refractory high-entropy alloys (RHEAs) at 800–1200 ∘C, a 149 h creep-rupture validation of an ML-designed high-entropy superalloy (HESA) at 1023 K and 480 MPa, and active-learning campaigns coupled directly to 1000 ∘C oxidation experiments. Across these examples, the limiting factors are concrete: many records lack exposure time, atmosphere, loading mode, measured composition, or post-test microstructure; compression data still outweigh tensile and creep evidence; and few AI-selected alloys have been followed through long-duration service-relevant tests.