Why perception is necessary for modeling human-like thinking_post_1

Company

Company

Why perception is necessary
for modeling human-like thinking

02

It is by perceiving reality in this most complex way that humankind developed the elements of language (letters, words, etc.). The human use of those elements culminated in the advent of texts. And it is from those human-made elements and texts that LLMs learn. For us, it goes without saying that LLMs have made better use of this innate ability than classical science did.

By non-scientific we mean that this kind of computability was not derived and/or invented and then "wired" into the brain through science, like it happened with math and linguistics. The computability we're referring to is something natural and specifically intrinsic to the human brain. This tool is so powerful that even an essentially mindless (unthinking) AI has managed to take advantage of it, while science, on the contrary, has slowed it down.

05

08

In the T-Bit project, we’re going to try and describe different aspects of individual scientific disciplines, AI included, which, in one way or another, study the issues of human thinking, mind, and consciousness, and show how TAPe explains them and even, we believe, makes discoveries and sets directions for further research of those complex concepts.

The Theory of Active Perception (TAPe) does just that — uses perception, as humans do, to obtain those meaningful primary attributes. TAPe-based methods using those primary meaningful attributes do not rely on modern mathematics, logic, or linguistics. Even more than that, TAPe-based methods deal with a different computability which is much more efficient than existing AI solutions through being closer to the human one.

07

Language is not a product of human science, but rather of evolution (or the Universe, God, and so on, you name it) which was deployed in the human brain, mind, and thinking in the form of something completely non-scientific and unknown to science — computability and its constituent elements.

04

But do language and texts suffice to describe a world picture? We doubt it. We ought not to forget about images, videos, sounds, smells, and other things that are very difficult to describe with language alone. This is why perception is necessary, at least as a starting point to obtain more meaningful primary attributes describing reality.

06

03

AI, with its ability to run through huge data sets, discovered the solution "on its own" and started yielding results that meet today's challenges. Still, 99.99% of the task, which was about creating a language and its primary elements, has actually been "completed" by humans.

01

AI success in language models specifically is understood as building non-scientific ways to learn a language. These ways, however, are based on the innate human capacity for language, one of the most important aspects of human perception as such.