Projects

Human speakers increasingly communicate with conversational agents, such as chatbots, for a variety of reasons including coaching and health-related purposes. Such conversational agents use dialog systems to process the textual input provided by the human interaction partner and generate appropriate textual responses. While chatbots have a number of highly attractive properties (low cost, highly adaptable, facilitating personalised interactions), trust in, and engagement with, these agents is currently remarkably low, primarily because they communicate through a non-human-like language structure and lack an ‘understanding’ of the ongoing interaction with their human conversation partner. Consequently, communication with current conversational agents is often regarded as non-personalized and boring, thereby hampering their proposed benefits for educational and coaching purposes. We argue that by ​combining insights from social sciences and humanities, we can address the limitations of current chatbots. The ​main aim of this project is to create and field-test chatbots designed to develop and foster long-term engaging and human-like conversations with humans for two technologically and linguistically diverse domains (smoking cessation and promotion of safe sex, respectively). Additionally, we will study if and how people build up a social relationship with such more advanced chatbots, and what the ethical implications of communication with such health chatbots are.

This project explores the possibilities for the development of an “understanding virtual coach” for people with amnesia. This virtual coach will take the form of a ‘conversational agent’, a piece of software that can interact socially with the patient with the aid of artificial intelligence. Unique within this project is that this coach will maintain a model of the mental state (wishes, needs, emotions) of the patient, based on an automatic analysis of conversations in combination with various sensory data. It is expected that this will improve the self-reliance of people suffering from amnesia, allowing them to participate qualitatively in society for longer. As a result, the project may provide a solution to the capacity problems created by the increasing shortage of healthcare personnel.

The goal of this project is to develop a serious game focusing on sustainability. Within the game, science on sustainability is communicated in an interesting and engaging way by letting the user experience different future climate scenarios in a text-based adventure game. The game will show them how their surroundings change when the earth continues to warm up. Moreover, the user will be able to experience how their own behavior can positively and negatively affect their surroundings. Several modules will be developed focusing on different target groups (e.g., children, house owners), and different contexts (e.g., scenarios for specific regions in the Netherlands). Additionally, users will learn how science can contribute to tackling the climate crisis, through insights from several disciplines (e.g. biology, physics, behavioural science). Additionally, we will explore the possibilities of implementing the game in our educational programs.

A healthy diet can help keep the body and mind healthy and extend the quality of life. A healthy diet and lifestyle are even more important for elderly people with care needs. Seniors often experience a decreased appetite, resulting in malnutrition. INtoEAT aims at development and implementation of innovative digital tools that can improve healthy eating habits and lifestyle in elderly as a nutrition aid. Specifically, a smart tray, a digital food coach, and a lifestyle app will be used for enabling personalised (remote) diet and lifestyle supervision and coaching, facilitating elderly to live longer independently and at home. Nutritionists also benefit a lot from the information in the app and can make better adjustments to patients’ diets. This, combined with intensive guidance from a nutrition coach, supports lifestyle and eating behaviour improvement.

Our society can benefit tremendously from artificial intelligence (AI). AI advances important goals, such as efficiency and economic growth. AI can help to predict where car accidents will occur, who will commit fraud, who can repay a mortgage, and so on. However, AI-driven decisions about people can also have unfair or discriminatory effects. For example, courts in the US used an AI system to predict whether prisoners will commit crime again. In 2016, researchers showed that the system was ‘biased against blacks’. The general public (consumers, minorities, employees) is most impacted by AI. Yet, their voice is surprisingly absent from discussions about human-centered AI. This research is one of the first to assess what the Dutch public thinks about AI-driven decision-making. The project’s overarching question is: What are the public’s knowledge of AI-driven decision-making, and their opinions about the promises and threats of AI-driven decision-making? To answer this question, we adopt a multimethod approach, using focus groups as well as a large scale survey. The research is conducted by an interdisciplinary team combining insights from computer science, AI, law, communication science, psychology, and science and technology studies.

Virtual reality (VR) provides many promising opportunities with regard to training of communication skills, as it provides a medium where users can safely practice their skills by engaging in social interactions with Intelligent Virtual Agents through verbal and non-verbal modalities. This way, VR allows users to practice what to say when, for instance to learn how to respond to a nervous interlocutor, or how to communicate professionally under increasing levels of stress. The ability for people to practice their skills is of great importance, as the transfer of ‘knowing how to do something’ to ‘actually doing it’ is difficult, even when one knows what he/she is expected to do. In this project, led by the Simulation Crew, we explore the potential of Intelligent Virtual Agents for training of employees of the Dutch Railways (NS).

In crisis situations, policemen and other emergency workers often face difficult decision making problems. Even though these people usually have clear instructions about how to act, they often have difficulties in making appropriate decisions, due to a combination of factors, including time pressure and heavy emotions. The main goal of the current project is to develop an intelligent system that is able to analyse human decision making processes in stressful circumstances, and analyse the causes of incorrect decisions. In addition, the system will be incorporated in an electronic training environment for emergency workers. In this environment, trainees will be placed in a virtual emergency scenario, in which they have to make difficult decisions, while negative emotions are induced. During the scenario, various existing techniques will be applied to measure their mental state (i.e., stress level and anxiety). This information will then be used as input for the intelligent system, to determine why they made certain incorrect decisions and teach them how to improve this.

Intelligent Virtual Agents (IVAs) become widely used for numerous applications, varying from healthcare decision support to communication training. In such applications, IVAs play various roles in which they interact with users, for instance as an instructor or teammate. Interestingly, in the vast majority of these cases, IVAs are friendly and supportive. Instead, the area of IVAs with a ‘negative’ attitude towards users (i.e., ‘virtual bad guys’) has been heavily under-researched. However, believability, a common problem in the design of IVAs, poses a particular challenge for ‘virtual bad guys’. Many effective applications require that users feel indeed seriously threatened or stressed by the IVA. However, as a general problem, IVAs are non-consequential, i.e., are unable to apply serious sanctions to users. Accordingly, users also perceive and categorize IVAs are virtual beings. These factors plausibly shape -and skew- how users respond to IVAs. So how to build virtual bad guys that are taken seriously? The present project tackles this question by designing and experimentally examining the effects of a technologically advanced IVA that is able to seriously threaten users. The study tests effects on stress and anxiety, and assesses the role of believability. Ethical boundaries and implications represent an important part of this project, too.

Cultural Awareness refers to the capacity of human beings to adapt to a different cultural environment and context. In the military domain, Cultural Awareness is increasingly being recognized as a crucial skill, for which adequate training is required. The aim of the Cultura project is to develop and validate a simulation-based Cultural Awareness training system for military staff. The system is based on high fidelity virtual environments, and dialogue-based interaction with believable virtual characters. Skills that are addressed include aggression de-escalation, information acquisition, and negotiation. The effectiveness of the system is studied in an evaluation experiment with military students.

Supermarkets, jewelry stores, and gas stations are prototypical examples of enterprises that are regularly confronted with robberies. During a robbery, the behaviour of the employees that are being robbed has an important impact on the outcome of the incident. Hence, employees of such enterprises need dedicated training to prepare them for robbery incidents. To this end, the current project focuses on the development of a simulation-based Robbery Awareness training system. The system will be based on high fidelity virtual environments, and dialogue-based interaction with believable virtual characters. As the decisions made by the trainee will have a direct impact on the behaviour of the virtual robbers, the system provides run-time feedback, thereby enhancing the learning effectiveness. The system will be validated based on user studies with employees from small enterprises.