AI Agent Research Papers

Essential research for understanding AI agent development

Home
Research Papers

Research Papers on AI Agents

Explore the academic foundations of modern AI agent development

Foundation Models & Large Language Models

Seminal Papers

Language Models are Few-Shot Learners (Brown et al., 2020) - Introduces GPT-3 and demonstrates the capabilities of large language models for few-shot learning.
Training language models to follow instructions with human feedback (Ouyang et al., 2022) - Introduces RLHF and the methodology behind instruction-tuned models like ChatGPT.
GPT-4 Technical Report (OpenAI, 2023) - Details on the capabilities and limitations of GPT-4.
Sparks of Artificial General Intelligence: Early experiments with GPT-4 (Bubeck et al., 2023) - Analysis of GPT-4's capabilities from a cognitive and AGI perspective.

AI Agents & Agentic Systems

Agent Architecture

ReAct: Synergizing Reasoning and Acting in Language Models (Yao et al., 2022) - Introduces the ReAct framework for combining reasoning and action in LLM agents.
LLM Powered Autonomous Agents (Li et al., 2023) - Comprehensive survey of agent architectures powered by large language models.
A Survey of Large Language Model Agents (Wang et al., 2023) - Explores how LLMs can be used as the core controller of autonomous agents.
TaskMatrix.AI: Completing Tasks by Connecting Foundation Models with Millions of APIs (Yang et al., 2023) - Framework connecting LLMs to external tools and APIs.

Planning & Reasoning

Chain-of-Thought Prompting Elicits Reasoning in Large Language Models (Wei et al., 2022) - Introduces chain-of-thought reasoning for complex problem solving.
Tree of Thoughts: Deliberate Problem Solving with Large Language Models (Yao et al., 2023) - Framework for structured exploration of reasoning paths.
Think Before You Speak: Training Language Models With Chain of Thought Prompting (Zelikman et al., 2023) - Methods for improving reasoning capabilities in language models.
Reflexion: Language Agents with Verbal Reinforcement Learning (Shinn et al., 2023) - Framework for enabling LLM agents to learn from past experiences.

Retrieval-Augmented Generation (RAG)

RAG Techniques

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (Lewis et al., 2020) - Original RAG paper that introduced the concept.
RETRO: Retrieval-Enhanced Transformer (Borgeaud et al., 2021) - Introduces a retrieval-enhanced transformer architecture.
REPLUG: Retrieval-Augmented Black-Box Language Models (Shi et al., 2023) - Framework for retrieval augmentation with closed-source models.
In-Context RAG: Retrieval-Augmented Generation without Training (Chen et al., 2023) - Techniques for implementing RAG with zero-shot prompting.

Knowledge Integration

Augmenting Language Models with Long-Term Memory (Wu et al., 2023) - Methods for implementing persistent memory in language models.
Retrieval of Specific Knowledge Reduces Hallucination in Large Language Models (Sun et al., 2023) - Analysis of how retrieval affects factual accuracy.
RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study on Agriculture (Feng et al., 2024) - Comparative analysis of RAG versus fine-tuning for domain adaptation.

Multi-Agent Systems

Collaboration & Debate

CAMEL: Communicative Agents for "Mind" Exploration of Large Scale Language Model Society (Li et al., 2023) - Framework for multi-agent coordination and role-playing.
Multi-Agent Debate Drives Reasoning and Factuality in Language Models (Du et al., 2023) - Using debate between agents to improve reasoning.
Self-collaboration for Large Language Models (Li et al., 2023) - Methods for a single LLM to collaborate with itself in different roles.
ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate (Chan et al., 2023) - Using multi-agent systems for evaluation of LLM outputs.

Emergent Behavior

Emergent Autonomous Scientific Research Capabilities of Large Language Models (Boiko et al., 2023) - Analysis of how multi-agent systems can perform scientific research.
Generative Agents: Interactive Simulacra of Human Behavior (Park et al., 2023) - Framework for creating believable agent-based simulations.
The Role of Cooperation and Planning in LLM Multi-Agent Systems: A Critical Survey (Zhuang et al., 2024) - Survey of cooperative behaviors in multi-agent systems.

Tools & Frameworks

Tool Use & Integration

Toolformer: Language Models Can Teach Themselves to Use Tools (Schick et al., 2023) - Self-supervised approach for teaching LLMs to use tools.
ART: Automatic multi-step reasoning and tool-use for large language models (Paranjape et al., 2023) - Framework for complex multi-step reasoning with tools.
CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquing (Gou et al., 2023) - Using tools to enable LLMs to critique and improve their own outputs.
SheetCopilot: Bringing Software Agents to Spreadsheets (Song et al., 2023) - Agent framework for spreadsheet manipulation and analysis.

Agent Frameworks

LangChain: It's Not Just Prompt Engineering (Dellermann et al., 2023) - Analysis of the LangChain framework's capabilities and applications.
AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation (Wu et al., 2023) - Introduction to the AutoGen framework for multi-agent applications.
CrewAI: A Multi-Agent Framework for Complex Problem Solving with LLMs (Medeiros et al., 2023) - Framework for collaborative AI agents with specialized roles.
LlamaIndex: A Comprehensive Framework for Building LLM Applications Over Private Data (Liu et al., 2024) - Overview of the LlamaIndex framework for knowledge-intensive applications.

Safety & Alignment

Agent Alignment

Constitutional AI: Harmlessness from AI Feedback (Bai et al., 2022) - Framework for creating safer AI systems through AI feedback.
Learning to Trust Language Models Without Conditional Training (Cohen et al., 2024) - Methods for improving self-monitoring in LLM agents.
How Do We Align AI Systems with Human Values? (Yao et al., 2023) - Survey of techniques for aligning AI systems with human values.
Beyond RLHF: Alternative Approaches for Alignment (Zheng et al., 2024) - Exploration of alignment techniques beyond RLHF.

Safety Concerns

Navigating the Grey Area: Expressions of Harmfulness in Language Model Generations (Liu et al., 2023) - Analysis of harmful outputs in language models.
Beyond Task Performance: Evaluating and Reducing the Flaws of Large Language Models on Safety-Critical Thinking Tasks (Chung et al., 2023) - Framework for evaluating safety in LLM reasoning.
Lessons Learned from Building Autonomous LLM Agents (Wang et al., 2024) - Practical experiences and safety considerations when building autonomous agents.