The combination of symbolic representation and machine learning can help to represent patterns in images and other data, but we are still far from forming the kind of wide world of computable knowledge that will form the language of the built environment. Currently, there is a loop between what is being picked up with machine learning and what is being picked up with more precise computational methods in natural language understanding, which is leading to new developments in the field.
In this episode, the guest discusses the idea of using raw data to extract hierarchies of abstractions and the challenges in developing learning systems that can form those hierarchies. They delve into the Wolfram language and Wolfram Alpha as representing the dream of what AI is meant to be, a full-scale computational language.
The integration of symbolic representation and statistical machine learning in identifying patterns in language and image processing. It serves as a tool to explore the world of computable knowledge.
The speaker discusses the current loop between picking up information through natural language processing with machine learning and more precise computational methods in natural language understanding by converting queries into a computational language.