Prompting Tools for Active Learning: Grow into an Expert by Talking to Experts — GenAI Monthly Meetup #2
Before officially launching the EUR genAI Monthly Meetup next academic year, we had an interesting second pilot meetup! These gatherings are designed to unite teaching staff, students, and other enthusiasts from EUR to explore the world of generative AI. Each month, we come together to share knowledge, tackle challenges, and brainstorm solutions in a collaborative environment — alternating each month between teaching-focused and learning-focused use of genAI. Speakers, both external and from our community, share the latest insights (and we encourage everyone to jump in with topics to present and workshop!). The meetups offer flexible participation and are perfect for anyone eager to stay ahead in the field of genAI.
To give you a better insight into what happens during these meetings, here’s a recap of our second pilot meetup. In case you missed the recap of our first pilot meetup, no worries, you can read that one here.
The Second Monthly Meetup — What Did We Do? 🤔
In the second Monthly Meetup, Sam Saidi, member of Turing Students and instructor at LifeVersity, workshopped two intermediate prompting techniques: Chain of Thought (CoT) and Tree of Thought (ToT) — we will explain what these are later.
Three scenarios were given to the participants: (i) you are an ESE student developing a research question on a 2022 general social survey, (ii) you are an ESHCC student evaluating a research article on online dating, and (iii) you are an EUC student conducting a critical analysis of K. Mansfield’s short stories. These case studies were chosen to demonstrate the transferability of the prompting techniques.
Here’s a recap of the two prompting techniques
Chain of Thought (CoT)
We delved into prompting techniques and the input/output tree, demonstrating how to guide ChatGPT to generate insightful responses. This method is particularly useful for students working on complex cases, as it helps them structure their thinking and approach problems methodically by emulating a step-by-step approach.
Tree of Thought (ToT)
The highlight of the session was the “expert method”, where students prompt the model to engage in a multi-round conversation with emulated experts, tasked to solve more complicated issues. In this conversation, a number of virtual experts engage in multiple rounds of conversations, sharing one step of their thought process before moving on to the next step. The ChatGPT-generated experts continuously share their perspective and their assigned probabilities of their solution being viable until they have reached a logical conclusion that considers every other expert’s contribution. The number of experts, their backgrounds and character traits can be defined to obtain specific angles on a question. By visualising each step of every expert’s thinking in approaching the problem, students can further develop their critical thinking and analytical skills.
If you’re curious to experiment with the prompts we used during our session, you can access the open-source article here.
Insights from the Participants 🔍
Benefits for Students 🧑🎓✅
- Moving toward AI:human Collaborative Learning: By adjusting prompt templates to better fit the context, students become more active participants in the learning process. Students may opt for the CoT or ToT prompting technique depending on their current knowledge and the complexity of the topic.
- Multiple Voices of the Expert Method (ToT): Getting multiple perspectives from virtual experts and hearing a debate among them stimulated students to critically consider different viewpoints, providing more objective and less biased views.
Challenges for Students 🧑🎓❌
- Critical Analysis of the Output: The prompting techniques require students to use their critical thinking skills to evaluate the output and ensure it is relevant and accurate.
- Prompt Selection Conundrum: Selecting the prompting technique that will give the “best output” may be challenging.
Benefits for Teachers 👩🏽🏫✅
- Learning Goal Adjustment: Knowing what intermediate/advanced prompting techniques students have at their disposal to make their learning more active and complex can help teachers adjust the learning goals of the course.
Challenges for Teachers 👩🏽🏫❌
- Add Time to Course Prep and Class Time: To make these prompting techniques useful for students in their learning process, teachers would need to provide guidance (including examples and opportunities for practice).
Come Join Us! 😁
To participate in the meetups and receive follow-ups, we invite you to become a member of our community! Don’t miss our next meetup this upcoming September 2024, where we will explore more advanced genAI techniques with EUR students, teachers and staff.
You can join us here (EUR members only)!
