Inspiration
One of our team members is an educator, and he wanted to use LLMs to generate creative lesson plans for teaching about climate-related topics.
What it does
An educator enters inputs such as the lesson topic, the grade level, and learning objectives. The tool uses GPT-4 to generate a lesson plan. The educator can use idea suggestions from the AI to refine the AI-generated lesson plan.
How we built it
We built a front-end UI for the educator to enter the lesson plan requirements. On the backend, we used GPT-4 to generate the lesson plan. The tool has a "truth-checking" feature that analyzes the lesson plan for key topics and fact-checks against an online source-of-truth (in this case, we used Wikipedia). The tool outputs the lesson plan as a downloadable document.
Challenges we ran into
We were concerned about truthfulness of the content, so we created a way to fact-check. We also found that the initial results were very general. After incorporating the summarize and compare feature, the lesson plan suggestions were more topic-specific.
Accomplishments that we're proud of
Our AI-powered teaching tool generates effective and memorable teaching plans.
What we learned
Educators need to be able to review and adapt the lesson plan. The most effective way to generate lesson plans is an interactive process where the educator can collaborate with AI.
What's next for EduClimate
We plan to add capability for educators to create a series of lessons, building on previous lesson plans. We can also build a collaborative network for teachers to "fork" lesson plans from other teachers so that they can tailor them for their own students.
Built With
- gpt-3.5
- gpt-4
- javascript
- openai
- python
- streamlit
Log in or sign up for Devpost to join the conversation.