The Challenges and Opportunities in Large Language Models: Navigating the perils of stochastic and scholastic parrots in artificial understanding and common sense

Abstract

Understanding is a complex process requiring contextual cues, common-sense reasoning and reflection on our own biases. Large language models (LLMs) have shown great promise in analysing text, but they still struggle with common-sense reasoning. This chapter aims to go beyond exploring the potential creation of autonomous AI with common sense and instead focus on the interconnected human and technical components of this technology. It examines the intricate links between tools and human existence within the context of AI. By delving into the interplay between machine capabilities and human interpretive processes, the chapter will unravel the concept of understanding from its semantic perspective emphasising the world-making power of language. © 2024 selection and editorial matter, Martin W. Bauer and Bernard Schiele; individual chapters, the contributors