From Turing Models to Large Language Models: Evolution and Convergence of Symbolic and Connectionist Approaches in Artificial Intelligence
Shijie Ye · Highlights in Science Engineering and Technology · 2025
This paper provides a comprehensive investigation into the evolution of artificial intelligence (AI). It focuses on the enduring scholarly debate between symbolism and connectionism, with a particular emphasis on the Turing model and the neural network model. The study situates Large Language Models (LLMs) within this theoretical framework, emphasizing their connectionist foundations while critically examining their historical interactions with symbolic approaches. Key issues addressed include the academic controversies surrounding symbolic and connectionist methodologies, the distinctive attributes of each paradigm, and the future development trajectory of LLMs—specifically exploring whether their advancement should prioritize algorithmic innovation or data-driven scalability. The primary contribution of this paper lies in its comparative analysis of the Turing model and the neural network model, offering a nuanced perspective on the respective strengths and limitations of each approach. By elucidating the research landscape, this comparative framework seeks to foster the convergence of these paradigms, thereby advancing the development of LLMs. The findings suggest that integrating symbolic and connectionist paradigms holds significant promise for enhancing LLM capabilities, with profound implications for both academic research and technological innovation. This paper contributes to a deeper understanding of AI, providing insights that may expedite the development of more resilient and adaptable AI systems, ultimately benefiting human welfare and fostering societal advancement.