Most ML models have been obliterated by LLMs but I do hope that the vast diversity of different ML techniques remains relevant. There's so much mathematical ingenuity in all the pre-LSTM approaches. SVMs, Gaussian models, Random Forest, network algorithms, ...