Résumé
The discovery of high-performance thermoelectric (TE) materials for advancing green energy harvesting fromwaste heat is an urgent need in the context of looming energy crisis and climate change. The rapid advancementof machine learning (ML) has accelerated the design of thermoelectric materials, yet a persistent "gap" remainsbetween high-accuracy computational predictions and their successful experimental validation. While MLmodels frequently report impressive test scores (R2 values of 0.90–0.98) for complex TE properties (zT, powerfactor, and electrical/thermal conductivity), only a handful of these predictions have culminated in the experimental discovery of new high-zT materials. In this review, we identify and discuss that the primary obstacles arepoor model generalizability—stemming from the "small-data" problem, sampling biases in cross-validation, andinadequate structural representation—alongside the critical challenge of thermodynamic phase stability.Moreover, we argue that standard randomized validation often overestimates model performance by ignoring"hidden hierarchies" and clustering within chemical families. Finally, to bridge this gap between ML-predictionsand experimental realization, we advocate for advanced validation strategies like PCA-based sampling and asynergetic active learning loop that integrates ML "fast filters" for stability (e.g., GNoME) with high-throughputcombinatorial thin-film synthesis to rapidly map stable, high-zT compositional spaces.