Title:	Next-Generation AI Agents: Language, Learning, and Autonomy
DNr:	Berzelius-2025-242
Project Type:	LiU Berzelius
Principal Investigator:	Ziyun Li <ziyli@kth.se>
Affiliation:	Kungliga Tekniska högskolan
Duration:	2025-08-04 – 2026-03-01
Classification:	10210
Keywords:

Abstract

This project focuses on building autonomous AI agents that combine Generative AI, Large Language Models (LLMs), and Reinforcement Learning (RL). These agents will move beyond static text generation to perform goal-directed tasks through reasoning, memory, tool use, and adaptive learning. By integrating LLMs’ language capabilities with RL’s decision-making and feedback mechanisms, the agents will be able to plan, interact, and improve in dynamic environments. Key applications include research assistance, code generation, and intelligent automation.