Editor: Jason Williams
Named Authors (Reviewed and edited the automated summary)
- Jason Williams, firstname.lastname@example.org, Cold Spring Harbor Laboratory, 0000-0003-3049-2010;
- Alex Francette, email@example.com, University of Pittsburgh, 0000-0003-1145-5847;
- Patricia M. Palagi, firstname.lastname@example.org, SIB Swiss Institute of Bioinformatics, 0000-0001-9062-6303;
- Alexander Botzki, email@example.com, VIB, 0000-0001-6691-4233
Recently, LifeSciTrainers.org held two community calls to discuss the potential of using ChatGPT, a large-scale AI language model, as a tool for training and teaching bioinformatics. The calls brought together experts, educators, and practitioners in the field to share their thoughts and experiences on the benefits and challenges of integrating ChatGPT into their teaching methods. This blog post summarizes the major points from these discussions.
Benefits of ChatGPT in Bioinformatics Education
- Enhanced learning experience: ChatGPT could potentially provide a more engaging and interactive learning experience for students. By serving as a smart assistant, the AI model can answer queries, assist with debugging, and provide real-time feedback on assignments and projects. But these answers are known to have factual errors. Until the system improves, ChatGPT must be used with caution. Some customized large language models (LLMs) could be limited to reliable information and can also cite sources.
- Accelerated learning: With the help of ChatGPT, students might be able to more quickly access relevant information, find solutions to problems, and learn new concepts. This can lead to a faster learning process and a deeper understanding of complex topics. Some participants noted that students find it easier and more “friendly” to use ChatGPT than discussion forums where users are less patient. Students also remarked feeling like they could ask as many questions as they wished with ChatGPT but would feel embarrassed to ask a human.
- Personalized learning: ChatGPT could cater to individual learning needs by providing personalized assistance and support. This can help address the diverse learning styles and backgrounds of students in bioinformatics courses.
- Reduction of mundane tasks: ChatGPT can automate repetitive and routine tasks, such as answering basic questions, which frees up time for educators to focus on more complex and challenging aspects of teaching. This could also be a boone for bioinformatics service providers who spend large amounts of time answering the same routine questions. It also might provide an easier path for educators to focus on pedagogy more than content.
Challenges and Concerns
- Dependency on AI assistance: Over-reliance on ChatGPT may lead to students becoming dependent on AI assistance, which could hinder their ability to solve problems independently and develop critical thinking skills.
- Academic dishonesty: The ease of access to information provided by ChatGPT could encourage students to misuse the tool for academic dishonesty, such as plagiarism or cheating on assignments and exams. Workshop attendees generally want to attend, so the danger here is more about self-deception – fooling yourself into thinking you have learned content.
- Limited understanding of underlying concepts: While ChatGPT can provide solutions to problems, it may not always help students understand the underlying concepts and reasoning behind the solutions. This could lead to a shallow understanding of the subject matter and hinder the capacity to learn.
- Keeping up with evolving AI models: As AI models like ChatGPT continue to evolve and improve, it becomes increasingly important for educators to stay up-to-date with these advancements and adapt their teaching methods accordingly.
Key Discussion Points
- Striking a balance: To maximize the benefits of ChatGPT while mitigating the potential drawbacks, educators should strike a balance between providing AI assistance and encouraging independent problem-solving. Critical thinking and the ability to recover from errors should remain essential components of the learning process.
- Separating concepts from implementation: Educators should focus on teaching fundamental concepts and mental models while using ChatGPT as a tool to support learning. For example, students should understand the basics of programming constructs, such as for loops, before relying on ChatGPT to generate code.
- Authoritative sources: Participants discussed the possibility of training ChatGPT on specific authoritative sources, such as course materials or official documentation, to ensure that the information provided by the AI model is accurate and relevant.
- Rethinking documentation and assessment: Educators should consider new ways of writing documentation that are more compatible with AI models like ChatGPT. Additionally, they should explore innovative approaches to assessing critical thinking and understanding, given that students may have access to AI tools during exams and assessments.
- Future implications: As AI models like ChatGPT continue to advance, the bioinformatics community should carefully consider the long-term consequences of relying on AI assistance in education. This includes questioning the necessity of teaching certain skills or concepts that may become obsolete in the future.
The discussions during the two community calls highlight the potential of ChatGPT to change bioinformatics education. The integration of ChatGPT into bioinformatics training presents both opportunities and challenges. On one hand, it can enhance the learning experience, save time, and make resources more accessible. On the other, it raises concerns about academic integrity, equity, and the quality of learning outcomes. As the technology continues to evolve, it is essential for educators and trainers to engage in ongoing discussions and collaborate on best practices to maximize the benefits of ChatGPT while mitigating its potential drawbacks. By fostering a strong community and staying connected, bioinformatics educators can navigate the challenges and embrace the opportunities offered by this powerful AI tool.
Discussants present for the above calls (as self-reported on the attendance sign in)
- Jason Williams, Cold Spring Harbor Laboratory
- Sateesh Peri
- Adrien Melquiond, University Medical Center Utrecht – The Netherlands
- Greg LoRe – SPEAR Consultants
- Sondra LoRe – SPEAR Consultants & Chatt State
- Patricia Palagi – SIB, Switzerland
- Carson Andorf, USDA-ARS
- Jacqueline Campbell, USDA-ARS
- Josefin Kenrick, Karolinska
- Chris Gates, University of Michigan
- Alexander Botzki, VIB, BE
- Jenny Drnevich, University of Illinois Urbana-Champaign
- Dan Chitwood, Michigan State University
- Radhika Khetani, Harvard University
- Beth Cimini, Broad Institute
- Marta Lloret Llinares, EMBL-EBI, United Kingdom
- Meeta Mistry, Harvard University
- Robert VanBuren, Michigan State University
- Alex Francette, University of Pittsburgh
- Lisanna Paladin, EMBL, Germany
- Alejandra Rougon, UNAM, Mexico
- Maria Doyle, Bioconductor, University of Limerick, Ireland
- Ajay Mishra, EMBL-EBI, United Kingdom
- Andrew Severin, Iowa State University, USA
- Sam Payne, Brigham Young University, USA
- Richard Barker, Blue Marble Space Institute of Science, USA
- Joel Nitta, Chiba University, Japan
- Negin Valizadegan, UIUC, HPCBio
- Jessica Holmes, University of Illinois Urbana-Champaign, USA
- Chris Fields, UIUC, HPCBio
- Sonika Tyagi, RMIT University, Australia
- Karin Verspoor, RMIT University, Australia
- Brian Ballsun-Stanton, Macquarie University, Australia
- Paul Harrison, Monash University, Australia
- Christina Hall, Australian BioCommons
- Johan Gustafsson, Australian BioCommons
- Rowland Mosbergen, WEHI
- Adele Barugahare, Monash University, Australia
- Jingbo Wang, National Computational Infrastructure, Australia
- Jessica Chung, Melbourne Bioinformatics, Australia
- Erin Becker, The Carpentries, California, USA
- Andrew Lonsdale, Peter MacCallum Cancer Centre, Australia
- Yashpal Ramakrishnaiah, RMIT University, Australia
LifeSciTrainers.org a global online community of practice for anyone and everyone who does short-format training (workshops, boot camps, short-courses, etc.) in the life sciences. The vision of this group is to improve life science by ensuring scientists and educators have the latest skills and knowledge they need to succeed.The website and online forum (Slack) is a place to share resources, advice, and conversation – all in the service of improving our teaching and our careers. Membership is open to all trainers who serve researchers and educators in the life sciences. Join at https://lifescitrainers.org/
Jason Williams, Alex Francette, Patricia M. Palagi, & Alexander Botzki. (2023). LifeSciTrainers.org Community Calls Discussion Summary – What will bioinformatics training and learning look like in the age of ChatGPT and AI? (v1.0). Zenodo. https://doi.org/10.5281/zenodo.7936329