Authors:
(1) Jinge Wang, Department of Microbiology, Immunology & Cell Biology, West Virginia University, Morgantown, WV 26506, USA;
(2) Zien Cheng, Department of Microbiology, Immunology & Cell Biology, West Virginia University, Morgantown, WV 26506, USA;
(3) Qiuming Yao, School of Computing, University of Nebraska-Lincoln, Lincoln, NE 68588, USA;
(4) Li Liu, College of Health Solutions, Arizona State University, Phoenix, AZ 85004, USA and Biodesign Institute, Arizona State University, Tempe, AZ 85281, USA;
(5) Dong Xu, Department of Electrical Engineer and Computer Science, Christopher S. Bond Life Sciences Center, University of Missouri, Columbia, MO 65211, USA;
(6) Gangqing Hu, Department of Microbiology, Immunology & Cell Biology, West Virginia University, Morgantown, WV 26506, USA ([email protected]).
Table of Links
4. Biomedical Text Mining and 4.1. Performance Assessments across typical tasks
4.2. Biological pathway mining
5.1. Human-in-the-Loop and 5.2. In-context Learning
6. Biomedical Image Understanding
7.1 Application in Applied Bioinformatics
7.2. Biomedical Database Access
7.2. Online tools for Coding with ChatGPT
7.4 Benchmarks for Bioinformatics Coding
8. Chatbots in Bioinformatics Education
9. Discussion and Future Perspectives
8. CHATBOTS IN BIOINFORMATICS EDUCATION
The potential of integrating LLMs into bioinformatics education has attracted significant discussions. ChatGPT-3.5 achieves impressive performance in addressing Python programming exercises in an entrylevel bioinformatics course[12]. Beyond mere code generation, the utility of chatbots extends to proposing analysis plans, enhancing code readability, elucidating error messages, and facilitating language translation in coding tasks[105]. The effectiveness of a chatbot's response depends on the precision of human instructions, or prompts. In this context, Shue et al.[10] introduced the OPTIMAL model, which refines prompts through iterative interactions with a chatbot, mirroring the learning curve of bioinformatics novices assisted by such technologies. To navigate this evolving educational landscape, it becomes imperative to establish guidelines that enable students to critically assess outcomes and articulate constructive feedback to the chatbot for code improvement. Error messages, as one form of such feedback, turn out to be an effective way to boost the coding efficiency of ChatGPT across various studies[10, 11, 96].
The convenience of using chatbots for coding exercises poses a risk of fostering AI overreliance, leading to a superficial understanding of the underlying concepts[12, 13, 106]. This AI reliance could undermine students' performance in summative assessments[12]. Innovative evaluation strategies, such as generating multiple-choice questions from student-submitted code to gauge their understanding[107], are needed to counteract this challenge. Such methodologies should aim to deepen students' grasp of the material, ensuring their in-depth understanding of coding concepts.
The art of crafting effective prompts emerges as a critical skill that complements traditional programming competencies. Feedback from a pilot study involving graduate students interacting with ChatGPT for coding highlights the challenges in generating impactful prompts[108]. This prompt-related psychological strain may discourage students from using the chatbot[13]. Instructors need to equip students with prompting skills known to enhance coding efficacy, such as one/few-shot by providing examples and stimulus prompting by listing function/package names[13]. In addition, the development of a repository featuring carefully crafted prompts, accompanied by quality metrics, reference code, and outcomes, could serve as a valuable resource for students to learn coding through prompting[10, 13].
In conclusion, while chatbots demonstrate potential as educational tools, their efficacy and effectiveness have not yet been systematically evaluated in classroom settings with controlled experiments. The use of chatbots should be viewed as supplementary to traditional education methodologies[10, 12, 13]. Meanwhile, new assessment methodologies are needed to measure the pedagogical value of chatbots in enhancing bioinformatics learning without diminishing the depth of understanding of concepts and analytical skills.
This paper is available on arxiv under CC BY 4.0 DEED license.
