Reference: Killian, C. M., Marttinen, R., Howley, D., Sargent, J., & Jones, E. M. (2023). "Knock, Knock... Who's There?" ChatGPT and Artificial Intelligence-Powered Large Language Models: Reflections on Potential Impacts Within Health and Physical Education Teacher Education. Journal of Teaching in Physical Education, 42(3), 385-389. https://doi.org/10.1123/jtpe.2023-0058
The abstract and section headings: The abstract, written by ChatGPT3 (and edited by the authors), argues that the emergence of artificial intelligence-powered chatbots should not be ignored. More than that it warns that ignoring these platforms may lead us to do “dumb things”. Burying our heads in the sand is not the way forwards. Instead, we need to pay attention to the issues and opportunities associated with this technology. We need to make the effort to learn about these tools and develop informed and solution-focused decisions about their future use.
The authors, through ChatGPT3, spoke of physical education teacher education, but I think this has wider implications and would advise others to read the paper if they can. The ultimate call from the paper is that we develop field-specific consensus statements to guide ethical and appropriate use of AI. Furthermore, there is a need to understand and address equitable practice and ensure everyone has access to these technologies.
The section headings are: Implications of AI-Powered LLMs for Learning, Teaching, and Research in H-PETE, Pitfalls and Potentials for Learning in H-PETE, Potentials and Pitfalls for Teaching in H-PETE, Potentials and Pitfalls for Research in H-PETE, Ethical Considerations for Using AI-Powered LLMs in Research and Teaching, Taking a Socio-critical Perspective, A Call for Consensus on the Appropriate and Ethical Use of AI-Powered LLMs.
Editor’s note, Introduction, and conclusion: The journal editors took time, at the start, to summarise the paper. They argued that ChatGPT3, and other such platforms, have quickly caught the attention of educational communities included in-service and pre-service groups. These technologies are capable of developing human-like responses to questions based on Internet searches. This creates both an opportunity and a need to respond. The authors’ hope is to lay foundations for a unified response from the profession in the form of guidelines and recommendations for effective and appropriate use.
The authors choice to use ChatGPT3 to wite their abstract isn’t a bit thing, but the readability and understandability of the abstract has big implications. These implications need to be put under the spotlight and the associated challenges need to be discussed. Ultimately, we (as a community) need to consider how we might address the emergence of the ChatGPT3’s in the world.
ChatGPT3 is an AI-powered large language model (LLM) designed to generate human-like text. Inputting prompts (like the one I used to ask ChatGPT3 to write the APA 7th Edition reference for this paper) demonstrates the considerable capacity of this technology to gather, synthesises and output sensible responses from public domain information. For many, the launch of ChatGPT was a seminal event ripe for reaction and prediction. For teachers and teacher educators in physical education, and beyond, there are implications for how teaching, learning and research are conducted. There are also implications for how we conduct ourselves and how these new technologies impact on our everyday lives. We could ignore the “knock on the door” from AI-powered LLM’s but that, to me and the authors, is a choice to try and ignore what is happening. Is that really a choice we should make? AI-powered LLM’s will affect our field and we need to strive to understand the different impacts they can have on all facets of our work.
There is real potential here to use the capabilities, functions, and potential applications of AI-powered LLM’s but there is also a need to understand how these will affect others and how they might use these tools for good and for bad. The ethics of our field are well-established – you write your own papers and do your own research – but AI-powered LLM’s have the potential to test, warp and even shatter those ethical practices. Being blind to, or playing dumb, with these technologies isn’t a choice. We can’t be luddites (“a person opposed to or fearful of technological advancements, particularly in relation to automation or machinery” - definition by chat.openai) and must strive, instead, to understand the potential implications and uses of AI-powered LLM’s.
In conclusion, Killian et al. (2023) argue that “conversations about the potentials and pitfalls of AI-powered LLM’s” are needed. We need to understand the individual users and uses of these technologies and understand the consequence of using and promoting them. We are in the early stages of truly understanding the “power and potential impact on humanity” of AI-powered LLM’s and we need to play our part in “ensuring the integrity of their use, protecting vulnerable population from inappropriate use, and promoting equitable access” (p.5). We must allow our differences (in thought, philosophy, perspective, experience, pedagogical approaches, and instructional styles) to divide us but must instead strive to unify and embrace these challenges.
Public trust in teaching, learning and research is vital and the ever-changing technological landscape – with its good intentions and actions – is being challenged by the AI-powered LLM’s. Consequently, we need to ne secure in our integrity and join voices to challenge and define the issues at play. Killian and colleagues ask that we proactively seek to develop discipline-specific guidelines for the appropriate and ethical application of LLMs.
Tables, figures and diagrams: Figure 1 showed the original ChatGPT3 abstract output.
The point of the paper: This paper has taken on the job of making the field aware of AI-powered LLM’s, which was needed. More than that, however, it has asked us to join together and develop discipline-specific guidelines for the appropriate and ethical application of LLMs. This is much needed and is certainly something I’d like to be involved in.
The main arguments: The authors argue that AI-powered LLM’s are knocking on the door of our teaching, learning, and research worlds. We can either ignore this knock and run the risk of doing “dumb things” or we can join together and answer the challenges and set the agenda for future engagement.
The importance of this paper: Firsts are always important, and this is a first to me. I have been toying with the idea of AI-powered LLM’s in physical education and am delighted to see others taking the first step. There is a long way to go but such journey start with one step and this paper signifies that one step.
The paper’s contribution to my knowledge: Firstly, I didn’t know the term AI-powered LLM’s which gives me a vocabulary. I also hadn’t thought of the need for equitable engagement with/access to LMM’s but that was immediately obvious after reading this paper. All in all, I finished the paper wiser and better informed.
Summary of the paper in one or two sentences: AI-powered LLM’s are her to stay. We can either embrace them and be part of their future or we can try and react to a landscape that is moving incredibly rapidly. Make your choice.
To the Authors: I would welcome a response to this blog. If you wish to write a response to be published on PEPRN, please email me – A.J.B.Casey@lboro.ac.uk with the final text.
About the Twenty 20 Vision Blog: For many years I wrote a weekly blog. In fact, between 2011 and 2021 I accumulated a catalogue of more than 450 blogs. But then I hit a wall. I simply ran out of energy and time. Consequently, 2022 wasn’t a great year for PEPRN. Following a modernization of the blog by Philte (thanks), and recognising my enduring desire that this break be a blip and not an end, I’m back with a new format and renewed ambition.
During 2022 I acknowledged that I needed to run PEPRN differently and over time the idea of a “Twenty 20 vision” blog emerged. This was, in part, in tribute to T20 cricket (which I love) and 20/20 vision (which I used to have) and in part due to the recognition that PEPRN needed to be easier to write and therefore sustain.
The idea, therefore, is for me to read a paper in no more than twenty minutes using an approach I’ve often recommended to my students but never actually done myself. For each paper I will do a shallow dive, reading only (a) the abstract and section headings, (b) the introduction and conclusion, and (c) the tables and diagrams. All other writing will be ignored. With this information to hand, I will then write the blog in no more than 20 minutes (thus twenty 20) using the eight headers above. Whatever emerges will then be published (after a little editing because I make a lot of grammatical errors typing that fast). The aim is to make paper reading and blog writing manageable once again whilst maintaining the integrity and usefulness of PEPRN. I hope I have achieved this, but feedback is welcomed and invited. Please let me know how I can get better at this and how the blog can better serve the community.