A I

After having a chanced encounter and an extended conversation with a data mining entrepreneur yesterday, (who is doing a city based data collation for building analytics for commercial potential ) I really wonder hasn't our technological advancement in AI stuck in the dependancy of ML/deeplearning on neural network protocols and making it redundant too fast !!

Over the last few years in my conversations with many data/mining experts and scientists, I found this particular attempt in their conversations to hide or side step the complicated alienation issue taking place between the ML and Deeplearning protocols. Also even as these experts claim the incoherence of theoretical rootedness in deep learning process for their emphasis on empirical approach for the neural network system, the resent developments in this sector only shows their application of 60s and 70s theoretical developments in structuralism and postmodernism in linguistics in their methods.

Their confusion looks like emanate from the assumption that
1. a decision is a "statistical average" of memory nodes/ associations, a constraint in neural network model
2. Almost near religious dependancy on causality, a structural necessity again of neural network model.

Considering these two issues were dealt extensively in quantum mechanics during the early twentieth century and had rejected both these premises with statistical inference (an near. impossibility in neural network system) and rejection of static causality( intelligence as memory association, a structural necessity for the hierarchy of information in neural networks), it is time that we look for alternate models that can offer equity, justice and values, the 'reasonable' intelligence of inference and interpretation. Otherwise the story of the data mining business shaping around the world will become a big waste land. Worse the information will stop at 'recognition' and ML will remain a stranger in that lost land.