In a world where communication is key, the question arises: how can we best train this essential ability? An answer comes from DialogueTrainer, a platform that develops conversation models to practice various dialogues, from performance reviews to conversations in customer service to delivering bad news. What makes this project so special? We spoke with Johan Jeuring, lead applicant and professor at Utrecht University.
'DialogueTrainer' develops conversation models that are used to teach how to conduct certain conversations, such as assessing performance or giving feedback. The conversations are conducted with a virtual character in a virtual environment (also called DialogueTrainer). The conversation models are used to develop scenarios in which a learner takes different steps through a conversation with the virtual character. At each step, the learner is given a choice between different options, and the emotion and reaction of the virtual character depends on this choice.
The motivation for this project was the stormy developments around large language models (LLMs) in recent years, and the goal of both DialogueTrainer and Utrecht University (UU) to practice communication skills not only with a choice of multiple options, but also to include text, sound and video in the practice scenarios. The research project came out of a long-standing collaboration between DialogueTrainer and Utrecht University (UU). Johan Jeuring, the lead applicant and professor at UU, spearheaded the development of the DialogueTrainer application, which was developed at UU from a collaboration with multiple faculties and fields, such as medicine, psychology, pharmacy, and veterinary medicine.
Large Language Models, or LLMs for short, are a type of machine learning model that is trained on huge amounts of text data. As a result, they can be used to generate human-like text, perform various language tasks and even better understand human language. Source: neuroflash.com
In practice, the collaboration looks as follows: DialogueTrainer provides a platform on which experiments can be conducted, and collects data that can be further analyzed. UU provides research expertise around software technology, evaluation of experiments, and analysis of data. UU also provides experimentation opportunities with students learning communication skills.
We hope to develop new concepts that can be widely used in the creative industry
The goal of this project is twofold: first, it seeks to use LLMs and dialogue corpora to analyze and improve the quality of dialogue scenarios. Second, the partners want to extend DialogueTrainer to better facilitate and utilize multimodal input. Johan Jeuring emphasizes the importance of this expansion: "We hope to develop new concepts that can be widely used in training and the creative industry."
A bold step toward a future where scenario generation and understanding of different forms of communication are seamless. We will keep an eye on the "DialogueTrainer" project, and will be happy to inform you of developments and updates!
