Professional Agents -- Evolving Large Language Models into Autonomous Experts with Human-Level Competencies

TL;DR

The paper tackles how to transform LLMs into autonomous, professional-level experts by introducing Professional Agents (PAgents) with a tri-layer architecture that enables genesis, evolution, and multi-agent synergy. It defines a four-module cognitive anatomy—Role, Perception, Brain, and Action—and details multimodal perception, memory architectures, planning strategies, and cooperative agent collaboration. Core contributions include a gene-driven genesis framework, iterative self- and co-evolution mechanisms, and structured pathways for scalable multi-agent systems, all aimed at achieving continuous professional mastery and potential AGI. The work highlights practical implications for deploying trustworthy, adaptable AI across complex domains while identifying critical research challenges that must be addressed to realize safe and robust PAgents in real-world settings.

Abstract

The advent of large language models (LLMs) such as ChatGPT, PaLM, and GPT-4 has catalyzed remarkable advances in natural language processing, demonstrating human-like language fluency and reasoning capacities. This position paper introduces the concept of Professional Agents (PAgents), an application framework harnessing LLM capabilities to create autonomous agents with controllable, specialized, interactive, and professional-level competencies. We posit that PAgents can reshape professional services through continuously developed expertise. Our proposed PAgents framework entails a tri-layered architecture for genesis, evolution, and synergy: a base tool layer, a middle agent layer, and a top synergy layer. This paper aims to spur discourse on promising real-world applications of LLMs. We argue the increasing sophistication and integration of PAgents could lead to AI systems exhibiting professional mastery over complex domains, serving critical needs, and potentially achieving artificial general intelligence.

Figures (1)

• Figure 1: The standard PAgent has a general framework, i.e., a sophisticated tri-layered framework. Specifically, individual PAgent features a modular architecture consisting of role, perception, brain, and action components.