AI-Researcher - An open-source automated scientific research tool launched by the University of Hong Kong

What is AI-Researcher?

AI-Researcher is an open-source automated scientific research tool developed by the Data Science Lab at the University of Hong Kong. Based on a Large Language Model (LLM) agent, it automates the entire process from research idea to paper publication. AI-Researcher supports two modes: one where users provide a detailed description of their research idea, and the system generates an implementation strategy; and another where users provide references, and the system autonomously generates innovative ideas for implementation. The platform integrates core functions such as literature review, idea generation, algorithm design and validation, results analysis, and paper writing, supports research across multiple fields, and evaluates research quality based on an open-source benchmark suite.

Main functions of AI-Researcher

• literature reviewThe system automatically collects and analyzes existing research literature in specific fields, and obtains high-quality research resources by searching academic databases (such as arXiv, IEEE Xplore, etc.) and code platforms (such as GitHub, Hugging Face).
• Algorithm verification and optimizationIt automatically designs, executes, and analyzes the results of experiments, evaluates the performance of algorithms, and optimizes them based on feedback to ensure the effectiveness and reliability of the algorithms.
• Thesis writingIt automatically generates complete academic papers, including research background, methods, experimental results, and discussion.
• Multi-domain support and benchmarkingIt supports research in multiple fields such as computer vision, natural language processing, and data mining, and provides a standardized benchmarking framework for evaluating research quality and innovation.

The technical principles of AI-Researcher

• Multimodal data integration and processingThe system uses automated tools to collect literature, code, and datasets from academic databases and code platforms, and uses natural language processing (NLP) technology to parse and analyze the text content to extract key information.
• LLM-based intelligent agentsBased on large language models (LLM) as the core driver, high-quality text content, including research ideas, algorithm design and paper writing, is generated using pre-trained models (such as OpenAI's GPT or DeepSeek's models).
• Automated Experimentation and VerificationThe system leverages containerization technologies (such as Docker) and automated scripts to enable rapid algorithm deployment and experimental execution. It automatically designs experimental procedures, collects results, and analyzes and optimizes these results using machine learning techniques.
• Multi-level task processing and modular designIt supports two task levels: users provide detailed ideas (Level 1) and users provide only references (Level 2). The system calls different modules according to the task level to automate the entire process from idea generation to paper writing.

AI-Researcher project address

• GitHub repository:https://github.com/HKUDS/AI-Researcher

Application scenarios of AI-Researcher

• academic researchAccelerate the research process, automating the process from idea to paper, and improve research efficiency.
• Interdisciplinary researchIt integrates knowledge from multiple fields, provides innovative ideas, and promotes interdisciplinary collaboration.
• Beginner's GuideIt provides easy-to-use tools for novice researchers to quickly generate research ideas and experimental designs.
• Enterprise InnovationIt helps companies quickly explore new technologies and accelerate product development and optimization.
• Educational applicationsAs a teaching tool, it helps students understand the entire research process and provides high-quality case studies.