For the conference on AI-Literacy for students and teachers, the Tech Labs of the Network Institute together with the VU Centre for Teaching & Learning developed an interactive avatar that came to life using a special “3D”-screen.
The conference allowed participants to discuss the possibilities and issues when using AI or LLMs in an educational setting. The main discussion themes were:
- Sustainability of using LLM
- How to keep up with the fast changing world of LLMs
- Privacy and the use of LLMs
- How do LLMs affect teaching
- What are the health effects on students when using LLMs often
To enhance the conference the Tech Labs developed a full-sized humanoid avatar, called Alisha, that allowed interactive speach. Participants were able to aks Alisha question about AI-Literacy and the conference. To create this the Tech Labs used speech-to-text, chatGPT3.5Turbo, text-to-speech, emotion expression and animation. Most of the processing was done using several Microsoft Azure services that allow a flexible use of different tools. This way it is very simple to replace the chatGPT3.5 model with any other available model on Azure without having to change the avatar program. The Tech Labs used their own emotional expression code that instructs the LLM to include emotional tags in responses. Those tags are then used by the code to animate facial expressions of the avatar. On top of that the latest AI-driven speech engine of Azure was used to insert emotions into the speech making the overall look-and-feel of the avatar more human.
To display the avatar a unique “3D”-screen was used the HoloBox by Holoconnects. This special huge screen incorporates a glass pane in front of the display and special white backlighting. Together with tweaked camera positioning and lighting in Unity, this creates the illusion that the avatar is 3-dimensional.
For more info on the conference please see here (Dutch only).
If you have any questions about this article or the services of the Tech Labs of the Network Institute, please contact us!