Adaptive Systems Using Generative AI
1 OVERVIEW
Generative AI technology is making big waves . It ’ s not just about using data to make decisions anymore ; it ’ s about generating new data from scratch . Generative AI opens multiple possibilities and use cases can be identified for each stage in the value chain of any industry .
This paper explores the possibility of utilizing Generative AI to build adaptive systems for industrial context . The article draws lessons from industries such as telecom , mobile phones and automotive on how software centric concepts have helped impart resilience , adaptiveness and support to new business models . We look at specific examples of applications of Generative AI to support adaption in robotics and multi-agent systems . We then explore how Generative AI , along with digital twins , can be used to build adaptive systems .
2 INTRODUCTION
Adaptive systems can be defined as systems that can adjust their behavior and / or structure in response to changes in environment or its own components . Examples of adaptive systems can best be taken from biological systems such as cells and micro-organisms . Over millions of years , cells and micro-organisms have survived harsh changes in the environment and their own structures by adapting and being resilient . For our purpose , we will call these self-adaptive systems .
We will define adaptive systems as systems that can adjust their behavior and / or structure in response to changes in environment or components based on instructions from an operator ( employee of the enterprise ) or a customer . These human-in-middle adaptive systems still need to adhere to the secure governance policies established by the enterprise to ensure the adaptive systems operate within the boundaries of intended purpose .
Studying behavior of complex biological structures such as an ant colony 1 , self-adaptive systems have the following characteristics 2 :
• The system is comprised of multiple heterogeneous agents . Each agent has well defined objective and behavior . Each agent learns and follows the objective for the agent .
• The agents interact . The interactions are defined by purpose . The agents exhibit adaptive behavior and learn / adapt from experience or influence .
• Exhibit behavior of emergence where the whole is more than sum of parts and gives nonlinear benefits .
1 https :// hbr . org / 2011 / 09 / embracing-complexity
2 https :// www . sciencedirect . com / science / article / pii / B9780128037263000079 Journal of Innovation 71