Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Converging Paradigms: The Synergy of Symbolic and Connectionist AI in LLM-Empowered Autonomous Agents

Haoyi Xiong, Zhiyuan Wang, Xuhong Li, Jiang Bian, Zeke Xie, Shahid Mumtaz, Anwer Al-Dulaimi, Laura E. Barnes

TL;DR

The study argues that LLM-empowered Autonomous Agents (LAAs) embody this paradigm convergence, by utilizing LLMs for text-based knowledge modeling and representation, and integrate neuro-symbolic AI principles, showcasing enhanced reasoning and decision-making capabilities.

Abstract

This article explores the convergence of connectionist and symbolic artificial intelligence (AI), from historical debates to contemporary advancements. Traditionally considered distinct paradigms, connectionist AI focuses on neural networks, while symbolic AI emphasizes symbolic representation and logic. Recent advancements in large language models (LLMs), exemplified by ChatGPT and GPT-4, highlight the potential of connectionist architectures in handling human language as a form of symbols. The study argues that LLM-empowered Autonomous Agents (LAAs) embody this paradigm convergence. By utilizing LLMs for text-based knowledge modeling and representation, LAAs integrate neuro-symbolic AI principles, showcasing enhanced reasoning and decision-making capabilities. Comparing LAAs with Knowledge Graphs within the neuro-symbolic AI theme highlights the unique strengths of LAAs in mimicking human-like reasoning processes, scaling effectively with large datasets, and leveraging in-context samples without explicit re-training. The research underscores promising avenues in neuro-vector-symbolic integration, instructional encoding, and implicit reasoning, aimed at further enhancing LAA capabilities. By exploring the progression of neuro-symbolic AI and proposing future research trajectories, this work advances the understanding and development of AI technologies.

Converging Paradigms: The Synergy of Symbolic and Connectionist AI in LLM-Empowered Autonomous Agents

TL;DR

The study argues that LLM-empowered Autonomous Agents (LAAs) embody this paradigm convergence, by utilizing LLMs for text-based knowledge modeling and representation, and integrate neuro-symbolic AI principles, showcasing enhanced reasoning and decision-making capabilities.

Abstract

This article explores the convergence of connectionist and symbolic artificial intelligence (AI), from historical debates to contemporary advancements. Traditionally considered distinct paradigms, connectionist AI focuses on neural networks, while symbolic AI emphasizes symbolic representation and logic. Recent advancements in large language models (LLMs), exemplified by ChatGPT and GPT-4, highlight the potential of connectionist architectures in handling human language as a form of symbols. The study argues that LLM-empowered Autonomous Agents (LAAs) embody this paradigm convergence. By utilizing LLMs for text-based knowledge modeling and representation, LAAs integrate neuro-symbolic AI principles, showcasing enhanced reasoning and decision-making capabilities. Comparing LAAs with Knowledge Graphs within the neuro-symbolic AI theme highlights the unique strengths of LAAs in mimicking human-like reasoning processes, scaling effectively with large datasets, and leveraging in-context samples without explicit re-training. The research underscores promising avenues in neuro-vector-symbolic integration, instructional encoding, and implicit reasoning, aimed at further enhancing LAA capabilities. By exploring the progression of neuro-symbolic AI and proposing future research trajectories, this work advances the understanding and development of AI technologies.
Paper Structure (19 sections, 4 figures)

This paper contains 19 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Connectionism vs. Symbolism: a Historical Debate on AI
  4. Knowledge Graphs: An Early Neuro-symbolic Attempt
  5. LLMs: Recent Connectionist AI Advancements
  6. LLM-empowered Autonomous Agents: The Convergence of Symbolism and Connectionism
  7. Autonomous Agents: Classic and LLM-empowered
  8. Design and Implementation of LAAs
  9. Rethink LAAs from the Perspective of Neuro-symbolic AI
  10. Symbolic Modeling and Neural Representation
  11. Search-based Decision Making by Generation
  12. Case-based Reasoning through In-context Learning
  13. Neuro-symbolic Integration Driven by Emergent Abilities
  14. Discussions and Future Directions
  15. Comparative Analysis: LAAs versus KGs
  16. ...and 4 more sections

Figures (4)

  • Figure 1: Elements of LLM-empowered Autonomous Agents (LAAs): Large Language Models (Neural Sub-System), Agentic Workflows (Symbolic Sub-System), and External Tools
  • Figure 2: Exploring the Evolution of Artificial Intelligence: A Timeline of Key Innovations and Milestones. It starts from the birth of symbolic and connectionist AI in the 1950s, through key milestones like the AI debates of the 1980s and the advancement in machine learning in the 1990s. This figure highlights significant developments such as the impact of AlexNet on image recognition, the transformation in NLP by models like BERT and GPT, and the rise of generative AI, culminating in the use of LLMs and Agents for autonomous decision-making in the 2020s.
  • Figure 3: Large Language Models and Their Agentic Abilities. The X-axis shows the release dates, and the Y-axis represents the LLM Agent Benchmark Score liu2023agentbench. Bubble size indicates the number of parameters (in billions). An asterisk (*) denotes estimated parameter counts when the official release is not available.
  • Figure 4: An Illustrative Example of Program-Proof-of-Thoughts (P$^2$oT) for Mathematical Proof Verification