I think it's been accelerated by the review community being overwelmed and the lack of experienced researchers with the combination of classic ML, deep learning, transformers, and DSP backgrounds -- a rare breed but sorely needed.