Biology does not understand how the brain works

Biology does not understand how the brain works

Science (biology) still treats the brain as a system of electrical, chemical, and — according to some — even magnetic impulses and uses standard math, physics, and logic in its research. It just doesn't work.

01

02

Idan Segev, one of the key scientists in the Blue Brain Project, explained why the brain — not of a human, but of a mouse — couldn’t be emulated: "We don't understand how the brain works or what values and parameters are important to it, and this is the reason why we don’t have enough computing power."

Let's agree on a simple postulate: solutions that the AI industry is offering right now have NOTHING to do with human intelligence or the human brain. Apparently, the human brain has a different computability.

04

This is exactly why AI can't "copy" the brain like airplane wings copy bird wings. Science doesn't know how the brain works. Questions should be asked of biology rather than of AI.

07

06

These AI’s fundamental shortcomings originate from the fundamental shortcomings of biology as a brain science. The brain may be thoroughly studied as a biological object, but those same methods provide little insight into how the brain actually works and what mind and thinking really are.

An attempt to create a model of connections between all 140 neurons of a nematode worm also failed — the model featuring all the connections couldn’t be described mathematically. Modeling the 128-cell nervous system of the roundworm was unsuccessful too. At the same time, we keep hearing about the imminent advent of AGI, which is just around the corner.

03

Why is that? Perhaps science simply CANNOT offer anything to help achieve a breakthrough, and therefore "ancient" information (about neural networks) is just enough to create a great engineering solution like machine learning or artificial intelligence, which, however, has nothing to do with intelligence at all — no matter what different people say about it.

13

It turns out that currently, AI is based on very sparse early information about brain function. So it seems only logical that more recent developments in this area should have brought about new AI solutions. But that hasn't been the case.

12

Walter Pitts/Warren Sturgis McCulloch continued to develop Rashevsky's ideas in the field of mathematical modeling and natural neural networks, but without learning, i.e., the network did not change over time.

10

One of the founders of mathematical biology, Nicolas Rashevsky, tried to reproduce the bases of higher nervous activity — Pavlov's conditioned reflexes — in artificial models.

09

At the same time, many prominent people believed that the topic of brain, thinking, and AI laid the foundation for further scientific and technological progress. But if the brain is so important for the "development of mankind", what do we know about it? What sciences study it and can study it further?

08

11

Many of the fathers of AI had a background in neuroscience or psychology. Hinton, Rosenblatt, and Hassabis are experimental psychologists. But at this point in time, artificial neural networks are based on very limited information about natural neural networks obtained back in the early 60's of the last century

05

In terms of what biology knows about the brain, AI has a number of fundamental shortcomings:

Modeled artificial neurons are far from biological ones, while physical processes are highly simplified.
AI learning is ensured by the backpropagation method, which is not the case in the human brain.

These principles serve as a basis for a perfectly working engineering solution. But claiming that AI copies the brain, intelligence, and thinking or, as Hinton says, "overtakes it" is erroneous.