Jonathan Y. H. SIM
Department of Philosophy, Faculty of Arts and Social Sciences (FASS)
Sim, J. Y. H. (2023). Teaching augmentative uses of ChatGPT and other generative AI tools [Paper presentation]. In Higher Education Campus Conference (HECC) 2023, 7 December, National University of Singapore. https://blog.nus.edu.sg/hecc2023proceedings/teaching-augmentative-uses-of-chatgpt-and-other-generative-ai-tools /
SUB-THEME
AI and Education
KEYWORDS
ChatGPT, generative AI, philosophy of technology, AI augmentation
CATEGORY
Paper Presentation
ABSTRACT
Since the rise of generative artificial intelligence (GenAI) like ChatGPT, educators have expressed concerns that students may misuse these tools by growing too reliant on them or use it to take shortcuts in their learning, thus undermining important learning objectives that we set for them.
Such concerns are not new in the history of technology. Socrates was one of the first to voice concerns about how the invention of writing would be detrimental to people’s memories:
“[Writing] will implant forgetfulness in their souls; they will cease to exercise memory because they rely on that which is written, calling things to remembrance no longer from within themselves, but by means of external marks.” (Phaedrus, 274b-277a)
Common to these complaints is the fear that new technologies will replace existing human processes—as a substitutive tool—leading to a deterioration or loss of certain human abilities. This is not the only approach to technology—we can also use these tools in an augmentative way to enhance existing human abilities and processes (Eors Szathmary et al, 2018). While we may not have memories as strong as the ancients did, writing has since augmented our thinking abilities, allowing us to easily record, recall, transmit, evaluate, analyse, and synthesise far more information than before.
This augmentative approach can also be applied to GenAI tools, like ChatGPT. 19.1% of my students (n=351) found ways to use ChatGPT as an augmentative tool rather than as a substitutive tool:
- As an idea generator or a sounding board to help develop ideas before working on an assignment
- As a learning resource to teach/explain concepts or clarify confusions
- As a tool to improve their expression
Admittedly, it can be difficult for non-savvy users to think of augmentative uses. Students are commonly exposed to substitutive applications of ChatGPT in learning, and 65.5% of students did not think skills were required to use it well.
How can educators encourage effective augmentative uses of GenAI tools? I believe there are three learning objectives we should focus on:
(1) Cultivate a collaborative mindset working with GenAI. Knowing how to talk is not the same as knowing how to work well in a team. Learners must feel comfortable and empowered working with GenAI as a collaborative partner if they are to use it as an augmentative tool. One approach is to incorporate activities that involve collaborating with GenAI. In my course, students are to work alongside ChatGPT to develop an evaluation criterion for ride-sharing services, seeking feedback from it while also evaluating its feedback.
(2) Develop critical questioning skills. Learners need to learn how to scrutinise GenAI output as the content may be inaccurate or shallow. Throughout the same tutorial, students were challenged to find flaws with ChatGPT’s suggestions, and to find areas where they can improve the quality of ChatGPT’s output. The exercise helped them to recognise that an AI’s answer is far from perfect, and that they cannot take a seemingly well-written work as the final answer. Human intervention and scrutiny is still necessary as the AI’s work is, at best, a draft suggestion.
(3) Master the art of prompting. The quality of AI output is dependent on the quality and clarity of instructions given to it. Learners need to hone their ability to articulate their requirements well. Later in the same tutorial, students were given a prompt for ChatGPT to generate a pitch. They were then tasked with identifying shortcomings to the output and to produce better prompts to overcome those issues.
After the tutorial, many students reported newfound confidence and competency in utilising ChatGPT (n=351):
Table 1
Students’ perception of ChatGPT competency before and after tutorial
| I considered myself very competent in using ChatGPT | ||
| Before Tutorial (Average 2.76) |
After Tutorial (Average 3.71) |
|
| 5 – Strongly Agree | 5.41% | 15.67% |
| 4 | 22.79% | 47.29% |
| 3 | 27.07% | 29.91% |
| 2 | 31.91% | 6.84% |
| 1–Strongly Disagree | 12.82% | 0.28% |
Table 1
Students’ perception of ChatGPT competency before and after tutorial
| The tutorial taught me how to effectively collaborate and work with an AI for work. (Average 4.19) |
The tutorial taught me how to effectively critique and evaluate AI generated output so that I don’t take the answers for granted. (Average 4.34) |
The tutorial taught me how to design better prompts to get better results. (Average 4.38) |
I believe the skills taught in Tutorial 4 are useful for me when I go out to work. (Average 4.28) |
|
| 30.77% | 40.46% | 43.87% | 38.46% | |
| 58.69% | 53.56% | 50.43% | 52.71% | |
| 9.69% | 5.70% | 5.41% | 7.41% | |
| 0.85% | 0.28% | 0.28% | 1.42% | |
| 0% | 0% | 0% | 0% |
Overall, students had positive experiences learning this new approach to AI. They felt empowered and even an optimism about their future—knowledge of using AI in an augmentative way opens doors of opportunities that seemed too distant previously. In one case, a social science major shared how he felt so empowered by the tutorial that he took on a coding internship (despite being new to coding). He used ChatGPT to learn how to code which facilitated him to handle coding projects at work. This augmentative approach not only allowed him to produce solutions but also evaluate them much faster than if he did it on his own.
I firmly believe that teaching students how to augment their learning with GenAI tools holds immense potential in empowering our students for the future.
REFERENCES
Eors Szathmary et al. (2018). Artificial or augmented intelligence? the ethical and societal implications. In J. W. Vasbinder, B. Gulyas. & J. W. H Sim (Eds.), Grand Challenges for Science in the 21st Century. World Scientific.
Plato. (1952). Phaedrus. Trans. Reginald Hackforth. Cambridge University Press.