Great point. However, I assume that you are aware but ML really is moving towards more general purpose AI. For example, the Turing Award winners for Deep Lea

Great point. However, I assume that you are aware but ML really is moving towards more general purpose AI. For example, the Turing Award winners for Deep Learning (Hinton/LeCun/Bengio) have all listed the numerous shortcomings of today's narrow AI and public declared an earnest interest in more human-like capabilities. And started serious research programs attacking those shortcomings. With many of the current generation of researchers quite engaged with them. There is progress on self-supervised learning, transfer learning, multimodal learning, more powerful models such as the transformer, extracting more accurate and fundamental latent spaces, model-based, online learning, more modular systems, 3d reconstruction from images and even videos, J. Tenenbaum's clear functional explanation beginning to be addressed with the new neural network tools, etc. Most of the results are not at the present moment inspiring enough to base a novel around. But there are many promising threads.