Slowing down of AI technology

Principles of scientific methodology

How to Increase AI profitability

Despite the revolutionary productivity of the Method of language generative neural networks, the current scale of AI application by corporations and the corresponding profit of neural network creators is tens of times less than the expectations of enthusiasts of this technology at the start of its mass distribution 2 years ago.

By now, almost all of Humanity's knowledge has already been recorded in the main neural networks.
The main technical ideas of the Method have already been used.
Therefore, the speed of development of the neural network industry is slowing down.

In the current circumstances, the main factor slowing down the AI industry, which can be minimized, is the insufficient quality of a large part of the knowledge that is entered into neural networks for training and for using them.

AI operates most effectively with knowledge in formalized formats that correspond to the classical principles of scientific methodology – http://world.kamerton.global/en/node/369

Principle One – Knowledge should be presented in a discrete format, that is, in the form of structured sets of individual short descriptions of the elementary properties of the objects being studied or designed.

Principle Two – Descriptions of projects and hypotheses should consist of three mandatory parts:
Description of the current state of the object being studied or designed and the problems that must be eliminated;
Description of the target state of the object and the criteria for achieving it;
Description of the process of transforming the object from the current state to the target state, the necessary and available resources for this process.

Principle Three – Knowledge should be presented in a discussion format and be the result of discussions of possible errors in the hypothesis or project between their proponents (authors) and opponents (critics), as well as possible errors in the arguments of the discussants.

Errors are:
incorrect words, the meanings of which do not correspond to the meaning of the descriptions and statements containing them;
logical contradictions between statements;
inconsistencies between descriptions and statements and known facts.

Formalized discussions of hypotheses and projects can be conducted in three ways: between experts, between experts and AI agents, and between AI agents.

Compliance with these classical principles requires a significantly more intense strain of consciousness.
Therefore, a significant part of experts and managers ignore and even deny these principles and violate them in their activities.
Therefore, a significant part of the knowledge that these experts produce, and which is then entered into neural networks, does not comply with the principles of scientific methodology and contains errors.
As a result, neural networks operate with erroneous knowledge and naturally compile erroneous statements from it.
That is, AI receives a lot of poorly structured, erroneous rhetoric and produces useless or even harmful text streams from it.

Thus, the usefulness of AI can be significantly increased as a result of improving the quality of information entered into it by formatting it according to the principles of scientific methodology.
Therefore, the processes of dissemination of AI technologies should be accompanied by the introduction of procedures for scientific formatting of semantic information in the activities of users.
Neural network manufacturers should include descriptions of these procedures in AI usage guidelines.
Corporations – manufacturers of AI systems should agitate the US corporate business community to implement these procedures and adhere to scientific methodology by publishing examples of their productive practice of intelligent processes.
To do this, they should implement these procedures in their internal activities – in management processes, in document formats, and in decision-making and publish information about them and their high efficiency.

Using information that is formatted according to scientific methodology will increase the profitability of the production and application of AI several times.

E. Gershman

___