This is AI generated summarization, which may have errors. For context, always refer to the full article.
I vividly recall my reaction when OpenAI first unveiled ChatGPT to the public. I was mind-blown. My astonishment was not rooted in the technology itself; our students have been working with transformers since the seminal paper by Vaswani et al. titled “Attention Is All You Need” was published in 2017. I was surprised by how OpenAI had operationalized the tool for public consumers. I would spend hours interacting with ChatGPT for the next few weeks to learn the platform’s capabilities and limitations.
First, some terms. ChatGPT, under the hood, operates on a large language model (LLM) developed by OpenAI. LLMs, in turn, fall under the umbrella of generative AI, a suite of artificial intelligence applications dedicated to generating new content — text, images, or other media forms — by analyzing and learning from vast datasets. Moreover, there are other LLM-driven platforms made available for public use: Google’s Bard, Microsoft’s Bing, and Anthropic’s Claude AI, among others.
Once I got the hang of “playing” with ChatGPT and exploring what it can and cannot do, I began using it for lower-level tasks at work, such as writing emails and performing tasks similar to what Grammarly does, like proofreading. And the more I interacted with this large language model (LLM), the better I understood the nuances of “prompt engineering,” which essentially is the art of asking questions or giving instructions to ChatGPT. I realized early on that ChatGPT can be a valuable assistant. —and not just for menial tasks. It excelled in bouncing off ideas for a wide range of topics; it’s like having a collaborator to brainstorm with. The way I described ChatGPT to friends was that it’s an intelligent assistant.
ChatGPT in education
Now, as a higher-education professor, I can most certainly attest to the significant impact of ChatGPT on my productivity. It has been instrumental in refining my existing course designs, preparing new course outlines, designing class sessions (lesson plans), crafting assessments, and even introducing innovative ways of delivering material in the classroom. The LLM also shined in preparing multiple and contextualized hypothetical scenarios, thereby maximizing students’ learning experience.
Thus, I have always been (cautiously) optimistic about how ChatGPT can augment us teachers, making us more effective in the classroom.
However, the sentiment among fellow teachers and educators has often differed. Particularly in the early part of this year, many colleagues in academia feared and disliked it. Others questioned its capability, noting, “It’s not yet updated with the latest information,” which was, and still is, true. Common concerns included, “Is there a real risk of AI tools like ChatGPT taking over my job as an educator in the future?” and “What exactly is ChatGPT? Is it just like using Google search?” I realized then that the issue stemmed from (1) a lack of understanding of the science of LLMs — what they can and cannot do, and (2) the apprehension about integrating new technologies into our established teaching practices.
Bridging the gap by demystifying LLMs for educators
I realized that proactive steps must be taken to bridge this widening and deepening chasm. It would be a missed opportunity if we didn’t harness the potential of large language models; they can augment our capabilities as educators.
“Why not roll out hands-on training on large language models for educators?” This thought was shared with colleagues at the Asian Institute of Management (AIM), and it garnered support from no less than our President and Dean, Jikyeong Kang. Thanks to AIM’s support and funding, we successfully organized two complimentary “Demystifying LLMs for Educators” sessions, primarily targeting early, primary, and secondary education teachers; we also had more than a handful of higher education teachers attending the program. In total, the program had its first 80 participants.
The training covered understanding artificial intelligence and the science behind LLMs, highlighting that while large language models are fluent, they are not inherently factual, and their outputs are predictive rather than deterministic. Hence, the concern regarding updated information becomes less significant, provided that ChatGPT is not used for novel research or in contexts where factual accuracy is paramount. A crucial component of the training was the importance of prompt engineering. A series of activities complemented this, focused on designing course outlines, drafting lesson plans, preparing assessments, and creating engaging discussions for the classroom.
The one-day training culminated in an open forum where teachers shared their thoughts on what they had learned. It was now evident that LLMs could significantly enhance our productivity. It was truly inspiring to see how a clear sense of empowerment emerged from understanding how the technology functions. Moreover, through the hands-on training, participants realized that the results and solutions provided by ChatGPT are not infallible. This recognition reinforced the value of their wisdom and expertise, highlighting that using ChatGPT requires thoughtful application and ethical considerations. This newfound understanding seemed to alleviate fears and encouraged the thoughtful integration of AI into educational practices.
Impact of AI on student learning
An expected discussion then emerged on the impact of AI use on student learning, focusing on integration and policy considerations. During the open discussion, the conversation shifted towards a call for clear guidelines and policies regulating the use of ChatGPT in education. I was pleased. Moreover, participants were empowered to demand a more strategic approach to integrating AI tools, ensuring tools like ChatGPT and Bard have a defined purpose in the classroom and that they reinforce program and course learning objectives. This approach includes our responsibility to educate our students about the ethical use of AI and the development of their critical thinking skills within the context of AI interactions.
Empowering educators: The way forward
As I reflect on this journey with ChatGPT and the successful “Demystifying LLMs for Educators” sessions at the Asian Institute of Management, I am filled with a sense of hope and inspiration. Seeing the transformation in understanding and attitudes among fellow educators has been profoundly rewarding. I am hopeful that this initiative, though just a small step in the grand scheme of things, has the potential to create significant ripples in the educational landscape.
Of course, I also recognize that more intentional and comprehensive efforts are needed for such initiatives to truly scale and become sustainable. One promising idea from our President is the implementation of an educating-for-educators program.
We need to continue to break down barriers, fostering a culture where educators appreciate and skillfully leverage technology to enhance our teaching. This involves more than introducing new tools; it includes ensuring that teachers are actively involved in shaping the conversation around educational technology. Empowering educators is synonymous with involving us in developing education policies that directly impact us and our students. By doing so, we ensure that these policies are grounded in the reality of classrooms and the evolving needs of our educational systems.
The journey with generative AI is more than just about embracing a tech tool; it’s about reimagining our approach to education in an AI-augmented world. From what I have seen, I am convinced that the AIM-driven initiative is just the beginning. As we continue to explore, learn, and adapt, I am optimistic that, collectively, we can harness the full potential of technology — specifically AI — to enrich our teaching practices, ultimately benefiting educators and students alike. – Rappler.com
Erika Legara is a Complex Systems Scientist at AIM.