Adaptive Systems Using Generative AI
Software-Aided Systems : If we look at the examples of mobile phones , Software Driven Networks ( SDN ) and Software Driven Vehicles ( SDV ), the content of software in systems has increased steadily . The software in systems and artificial intelligence ( AI ) have provided flexibility for designers to add features and handle unknown situations discovered (“ known unknowns ” and to an extent “ unknown knowns ”).
AI and Generative AI : Artificial Intelligence combined with Generative AI is a major step towards General Purpose Intelligence . There is a general view that Generative AI is not yet Artificial General Intelligence . This implies that handling “ unknown unknowns ” is still not a solved solution . Generative AI , with the ability to understand larger problem domains and generate code , has potential to solve the “ unknown knowns .” Larger language models ( LLM ) with a vast amount of data can co-relate cross domain knowledge and generate interesting “ known ” solutions . We will explore the available frameworks in robotics and multi-agent systems to help in solving “ unknown knowns ” keeping “ human-in-middle ” towards building adaptive systems .
4 LESSONS FROM INDUSTRIES
In this section we look at how industries like telecom , mobile phone companies and automotive have adopted software driven systems to build products that are adaptive , resilient , enable business agility and provide customer delight .
4.1 TELECOM INDUSTRY – SOFTWARE DEFINED NETWORKS
In traditional telecom networks , each device ( such as a router or a switch ) has its own software and hardware that decides where to send the data packets . This makes the network hard to change and manage , especially when there are many devices and different types of traffic .
In Software Defined Networks ( SDNs ), the devices are separated into two parts : the data plane and the control plane . The data plane is responsible for forwarding the data packets , while the control plane is responsible for deciding how to forward them . The control plane is centralized in a software application called the SDN controller , which can be implemented with open standards and communicate with all the devices in the network .
By using SDN , telecom operators can create virtual networks that can span across multiple physical locations and devices . They can also dynamically adjust the network configuration and resources according to the demand and traffic patterns of different applications and customers . For example , they can allocate more bandwidth for video streaming or create secure zones for sensitive data .
74 March 2024