Project in the spotlight: Avatalk
/
Project in the spotlight: Avatalk
Within theTetra project Avatalkfunded byVLAIOresearchHuman scale, DAE Research and AI Labof AI can add value for practicing challenging conversations such as evaluations, dismissals, or where there is a conflict between multiple people.
To address the needs from our own training programs and the job market, we will develop an application during the course of this research project in which users can converse via text, audio, and video with an AI-driven persona based on realistic scenarios.
Main section
The application
Application
Dashboard
The heart of our application is the dashboard. Here, users can find an overview of the conversation partners available to them.
Admins can also create new conversation partners to set up new scenarios and exercises for their staff/students.
After selecting a conversation partner, the user receives an overview of the situation and the outcome they need to achieve during the conversation. After reviewing this information, they can either start a chat to begin the conversation or initiate an audio call.
Chat
When you start a chat conversation, you enter a typical "chat window" where you can converse with the AI persona.
It is also possible to review the context of this conversation again by clicking on "info about the conversation" or viewing the entire system prompt via the "system prompt" option.
Once the intended results of the conversation have been achieved, the conversation can be concluded and an automatic evaluation will be prepared.
Voice Call
It is also possible to have an audio conversation with the AI persona by clicking "start call."
Then you end up in a Microsoft Teams-like environment where you can conduct your conversation. For the rest, the concept is the same; once you close the conversation, a transcript of this conversation will be sent to the dashboard for feedback analysis.
Bottom section
Next steps
The next step is itimplementing the avatar layeras well asto further roll out the voice modeto the test users.
Once the tool has been thoroughly tested, we will also start working on concrete cases provided by our advisory group to make the impact of this application measurable.
Do you want to stay informed? FollowDAE Research, Human scaleinAI LabopLinkedIn.