LIFTOFF
ZERO

QuickAI

Natural Language Processing (Reasoning)

There are several problems in AI that are called “AI complete”. This is to say, if you solve for one of them, then you will have solved them all. And one of these is the domain space of Natural Language Processing (NLP, also Natural Language Reasoning), which is the vector by which we approach the goal of True AI. The question is, does a machine have the capacity to actually understand what it’s dealing with, when it works with the language of flesh and blood beings like us?

The clue to understanding is to define what logic is, and saying that when we understand something, then we are saying we can see its logic. If we take logic to mean the way that things make sense, we are using the word correctly, even if we use it in comprehending seemingly illogical structures. What we’re saying is that if a human being can make sense of it, there is some kind of logic at work. And if we can get the machine to perceive that kind of logic, the machine will actually understand when we talk to it.

Theoretica.info

This is a new way of looking at... everything. What is in the nature of information that lends it meaning? What is meaning itself? How to define cause and effect? In a form of something like a “contact metaphysics”, we posit such ideas as, “meaning is a change in state”, “causation is measurement”, and “the satisfaction of a pattern is also a pattern”. It is a practical metaphysics, in other words, dealing with phenomena we so take for granted there is little thought in what they might be, but crucial when we consider how we perceive the way the world works. To understand such things is to be able to put it in concrete form, into bits and bytes, into the code that makes our AI function.

For more information, check out our project page at Theoretica.info.