In the field of AI we are at the technological level of ancient Greece

22 октября, 2021 от Kinok Выкл

In ancient Greece, they knew how to create complex beautiful structures and surrounding philosophical theories. However, calculate the trajectory of the abandoned stone (not to mention the trajectory of the earth’s movement) could not be Aristotle himself. After all, to calculate these trajectories you need to know specific mathematical formulas based on Newton and Kepler laws.

Today, science does not have mathematically described laws, in accordance with which biological intelligence works. And therefore, people can make a smart column with a virtual voice assistant today. Ah, similar to not even human, and at least mouse, — there is no alas.

The problem is that our intellect is the result of hundreds of millions of years of evolution, which created it with one sole purpose

Those. Biological intelligence is always embodied in a particular body and is needed to control reasonable behavior, adjusting it to the morphology of the body and adapting to the environment.

Consequently, it is possible to understand the mathematics of the functioning of biological intelligence, only studying the dynamics of body adaptation to the environmental conditions.

It is not embodied, but inexpensively (in machine learning algorithms there is no physical body, which needs to be survived and reproduced). And therefore, AI, strictly speaking, incorrectly compare with biological intelligence.

Recently, I wrote about the development of Deepmind concept of reasonable matter, allowing to avoid the problem of absence from body intelligence. However, this concept, such as the «Reasonable Ocean» from the novel «Solaris», is too unearthly, «for the entire earthly evolution, there was not a single disembodied reasonable creature. And therefore, the traditional approach is the medium, the body, intellect, — it seems still promising.


1) The complexity of the environment contributes to the development of morphological intelligence, quantitatively determined by the ability of morphology to facilitate the solution of new tasks.

2) Evolution quickly chooses morphology that learn faster, thereby allowing the behavior learned during the life of the ancestors to be transmitted for use in the life of their descendants.


But the main thing is, it will be possible to explore the occurrence of morphological intelligence through the evolution of morphologies leading to more physically stable and energy-efficient behavior facilitating training and control.

That’s what it looks like.

Additional details on the topic Look in the channel on the tag closer

Thanks for watching! Put husky I.