Skip to main content Skip to navigation

The why, what, and how of AI-based coding in scientific research

Project Overview

Generative AI, especially through the use of large language models (LLMs), is revolutionizing education by enhancing the coding landscape, making it more intuitive and accessible for students and researchers. These models support a range of coding tasks, including understanding, generating, debugging, optimizing, translating, and learning programming languages, thus streamlining the educational process in computer science and related fields. However, the integration of AI tools into educational workflows requires careful consideration, emphasizing the necessity for targeted training to ensure effective use. Additionally, ethical considerations play a crucial role in this transformation, highlighting the importance of maintaining foundational coding knowledge to prevent over-reliance on AI technologies. By striking a balance between leveraging AI capabilities and fostering essential coding skills, educators can enhance learning outcomes and prepare students for a future increasingly influenced by artificial intelligence.

Key Applications

LLMs for coding assistance

Context: Biomedical sciences and social sciences researchers seeking to improve coding skills and efficiency.

Implementation: Researchers use LLMs to generate code, understand existing code, and debug issues through a structured five-step workflow.

Outcomes: Increased productivity, enhanced learning, reduced coding time, and democratization of coding skills.

Challenges: Limitations in understanding complex coding tasks, the potential for AI-generated errors, and the need for foundational programming knowledge.

Implementation Barriers

Technical

LLMs may produce hallucinated or incorrect outputs and perform unevenly across programming languages. There is a need for better training datasets and verification of AI outputs through testing.

Proposed Solutions: Develop better training datasets and encourage the verification of AI outputs through testing.

Educational

There is a lack of systematic training for researchers in coding and AI tool usage, necessitating targeted training programs focusing on AI integration into coding practices.

Proposed Solutions: Implement targeted training programs focusing on AI integration into coding practices.

Ethical

Concerns exist around equitable access to AI tools and transparency in AI usage in research. Establishing clear guidelines on AI tool usage and documentation in academic settings is essential.

Proposed Solutions: Establish clear guidelines on AI tool usage and documentation in academic settings.

Project Team

Tonghe Zhuang

Researcher

Zhicheng Lin

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Tonghe Zhuang, Zhicheng Lin

Source Publication: View Original PaperLink opens in a new window

Project Contact: Dr. Jianhua Yang

LLM Model Version: gpt-4o-mini-2024-07-18

Analysis Provider: Openai

Let us know you agree to cookies