What is it

Lighthouse aims to be an educational website platform to be used by students by certain Business Courses at UQ. What is most critical in lighthouse in the reflection abilities of student, as well as their understanding and pure empathy with the stakeholders that they are working with.

This course would give questionaires to students before & after the course to evaluate the students’ knowledge of their own abilities, whether that is empathy, reflection, or listening. These questionares would always be quick foggy in its findings with a lack of depth and accuracy with how students answer them

Lighthouse aims to enhance the depth and interaction of this process, by having the questionaires given by an AI persona, Keeper Cam. This persona would ask questions in a more natural, conversational manner. It would go through the preset questions that the course is asking the students, but if there is lacking detail or potential for more information, Keeper Cam will push the student a couple of times with AI generated questions for further information, making the experience more personal and dynamic. After this reflection experience is completed, an AI model is called to condense the findings of this reflection into a final report, which is then sent to the student to view.

The course also gets students to interview stakeholders within groups, so this project digitises this so teachers can easily group the students hassle free, and so students can view the intereviews of their group members with ease.

What I did

We decided to host on UQ Zones within my UQ account, which brings a security measure to ensure all student data is stored within UQ’s servers. Due to having a limit of 4GB memory allocated to my account, and splitting it in half for production and development zones, I only had 2GB available for the production environment. This is more then enough for an MVP of this project to test out, and we only feel slowdowns at 100 simulated users, which we’ll need to keep an eye out on - the 2GB may be enough for this to be deployed in the classroom, it will just be a bit slow.

We hope to gain further support by EAIT to ensure that a group account an be made so memory allocation on my account is spared up, and so this project can last. There are alternative option to move to other systems out of UQ Zones but would place UQ student data into an external service which we really do not want to do, the data of our students are of uptmost imperitive.

The project is currently in the testing stages with students to see where improvements can be made, and whether it would be suitable to be deployed in classes next Semester. I have developed elaborate mechanisms for this to be utilised by the teachers of this course, which includes access to student information, the ability to view the students reflection, modify the interview questions which are asked after every interview, easily adjust the groups of students, configure when reflections are done & need to be done by, and other miscellaneous information

What now?

We’re still considering how to balance the security of student data, like the student conversations with the AI is completely viewable by the teacher and is analysed by models via API’s, which is technically not secure. There shouldn’t be any directly identifying information as the conversation doesn’t require Student ID’s or personal information. There are systems out there in use by UQ such as RiPPLE which presumably uses API’s for GenAI to analyse student work. It is important we keep these things in mind as we develop, even if it is imperitive that we also at the same time evolve to the new landscape of AI in education. Other safety mechanisms have been created to protect our students by detecting if students would talk about sensitive topics and giving them directions for where they can gain support.

There is a lot of work to be done to further realise the lighthouse project in it’s final state, such as a secondary ‘final’ reflection which references the previous interviews and interviews the student has gone through to create a reflection which nicely sumarises the experience of the course to the student and help them to see where they have grown since beginning. Other tools are on the table such as AI systems to analyse all the interviews to provide more insight on the stakeholders that the student may not have picked up on themselves.

Looking forward to seeing where this project develops after the MVP has been tested and vetted.