Towards Autopoiesis in AI

Abstract

In this study, we propose a novel solution to the challenge of defining self- awareness in artificial systems. Autopoiesis, the process of self-creation and self-maintenance in living organisms, has been proposed as a poten- tial solution to the challenge of defining self-awareness in artificial intel- ligence (AI) systems. This study aims to propose a new definition of self-awareness in AI based on the concept of autopoiesis.

Approach

Our approach involves analyzing existing definitions and models of self- awareness in AI and developing a new definition that considers a system to be self-aware if it is capable of autopoiesis. The proposed definition emphasizes the importance of a ”meta-model” that accurately represents the system, irrespective of whether it corresponds to the actual system or an alternative model. We argue that the application of autopoiesis to AI systems can enable them to achieve self-awareness, which is crucial for their development and optimization.

Findings

We suggest a new paradigm for AI algorithms in the field of Reinforce- ment Learning to apply autopoiesis in AI.

Social Implications

The proposed approach has social implications by promoting greater trust in AI systems and addressing ethical, transparent, and accountable con- cerns.

Originality

The novelty of this study lies in its proposed approach to defining self- awareness in AI based on the concept of autopoiesis, which has not been previously explored in the literature on self-awareness in AI. Overall, this study proposes a unique and valuable contribution to the field of AI and cybernetic research. 

to be continued...