In a recent study conducted by researchers from West Virginia University (WVU), the potential of the ChatGPTCode Interpreter plugin has been highlighted for educational settings. This innovative tool has been lauded for making coding more accessible to students in STEM fields. However, while Code Interpreter showcases promise in education, researchers have identified limitations that impede its utilization by scientists working with biological data in bioinformatics.
Bridging educational gaps
The Code Interpreter plugin, an extension of the popular artificial intelligence chatbot ChatGPT, has garnered attention since its release in December 2022. Researchers from WVU have recognized its significance in educational settings, especially in STEM disciplines. Gangqing “Michael” Hu, an assistant professor in the Department of Microbiology, Immunology and Cell Biology at the WVU School of Medicine and director of the Bioinformatics Core, emphasized that Code Interpreter makes coding more accessible to students without a science background.
This fosters curiosity and sparks an interest in data analysis, encouraging students to delve into the programming world. Moreover, the tool’s cost-effectiveness has been noted as a bonus for educational institutions.
Limitations in bioinformatics
Despite its potential in education, the Code Interpreter plugin falls short in meeting the demands of scientists working in bioinformatics. Bioinformatics, the convergence of computer science and biology, relies heavily on accurate coding, software programs, and internet access to interpret complex biological data for medical advancements.
Hu and his team subjected Code Interpreter to rigorous testing, revealing its strengths and shortcomings. While the tool aids users in distinguishing accurate responses from unreliable ones, it lacks certain technical features necessary for bioinformatics. Hu explained that while ChatGPT is known for its capabilities, it struggles to provide verifiable references to support its answers. In this regard, the Code Interpreter plugin stands out as it incorporates the actual code as a source or citation, minimizing the risk of presenting fictitious information.
Overcoming challenges for bioinformatics
The study’s findings underscore the need for specific enhancements tailored to bioinformatics. Suggestions for improvements include facilitating internet access to download genome data, incorporating software relevant to bioinformatics, expanding storage capacity, and supporting additional programming languages. Additionally, privacy and security measures are imperative to adhere to regulations such as HIPAA.
However, the Code Interpreter plugin faces limitations in data analysis. It currently supports only Python programming, with a limited selection of software packages dedicated to bioinformatics. Accessibility to internet data and the capability to handle large files are areas where the tool falls short. Hu noted that the tool’s file capacity is insufficient for the gigabyte-level files typically encountered in bioinformatics work. Moreover, its lack of support for parallel processing further hampers its performance with large datasets.
A glimpse into the future
Hu remains optimistic about the evolution of the Code Interpreter plugin. Despite its current limitations, he plans to integrate it into his teaching curriculum to aid students in comprehending data visualization. He anticipates that future upgrades will address the shortcomings, making the tool applicable to a broader range of bioinformatics coding tasks. This aligns with his commitment to staying at the forefront of AI programming advancements and discovering innovative applications.
In a related study earlier this year, Hu pioneered the OPTIMAL approach (Optimization of Prompts Through Iterative Mentoring and Assessment) to prepare high school and college students for effective interaction with AI chatbots, such as ChatGPT.
The ChatGPT Code Interpreter plugin demonstrates its potential to revolutionize coding education in STEM fields. Although it has yet to meet bioinformatics requirements fully, researchers remain hopeful that ongoing developments will bridge this gap. As AI technology advances, the realm of possibilities for its application is vast and promising.