The extraordinarily rapid and disruptive impact of Large Language Models and their ability to generate human-like, conversational responses to text inputs has revolutionized machine interaction with humans. This step change has stimulated tremendous public attention, debate, controversy, and applied innovation of this artificial intelligence (AI) technology to various industries and purposes, from customer service and language translation to creative writing, content creation, and software development.
The state-of-the-art transformer architectures behind large language models are pushing the boundaries of generative AI, providing new ways of directing AI model behaviour. For example, when combined with diffusion models, they deliver improved methods of creating high-quality content, like synthetic images or 3D models, directly from text input. Generative AI’s ability to create data and content as plausible, synthetic alternatives to real data offers significant flexibility, lower cost, and faster generation of content, data, and synthetic environments. However, although the capability to create specific AI-generated content may exist, it is the opportunity to incorporate these AI functions into an integrated pipeline that, when combined with AI-based autonomy, has the potential to deliver transformative improvements in efficiency, productivity, and capability.
Current SE pipelines typically employ AI content and data creation processes as independent operations. This paper will build on ongoing research to explore how disruptive AI technologies, like transformer and diffusion architectures together with large language models, may be leveraged and combined with other generative and AI techniques – not only to generate data and content but to provide new, innovative AI-based approaches to directing, integrating and automating these processes. Furthermore, it will consider how these capabilities can be composed into a solution architecture that provides an end-to-end AI-based SE construction pipeline to facilitate the dynamic, faster delivery, and economical production of complex SEs and ultimately enhance their exploitation, user readiness and decision-making.
Keywords
AI, AUTHORING TOOLS, AUTOMATION, AUTONOMY, CONTENT GENERATION, DATA, DEEP LEARNING, ENVIRONMENTS, MACHINE LEARNING, NATURAL LANGUAGE PROCESSING, SYNTHETIC ENVIRONMENT
Additional Keywords