Admission requirements
Recommended knowledge:
Intro Cognitive Science
Introduction to programming
Description
Conversational agents are playing an increasing role in our society. Think about chatbots, telephone bots, robots in education at school, and robots in airports or restaurants, and even chatbots in healthcare. Some of these technologies even serve a social function for example as companions or coaches. But how do you actually program or script the human-agent interaction (HAI)? How do you ensure a pleasant dialogue? How do you respond to a human? How do you measure what a human thinks, and how do you do it in a smart way so that the human doesn't find it annoying? Often, developing a pleasant and effective interaction means dealing with different AI techniques, all of which are powerful but not perfect. Further, novel techniques such as Large Language models and Text-to-Speech change the way we interact with AI rapidly, and the way we can build dialogues between humans and AI agents. In this course, you will learn how to shape and program the interaction between humans and intelligent agent by making use of appropriate AI techniques, conversational principles, and different modalities, and you learn how to evaluate your design using test subjects using an appropriate experimental setup.
Course objectives
1. Remember and understand the different theoretical concepts underlying human agent interaction and multi-modal dialogue.
2. Apply these theoretical concepts in the design of in interactive agent.
3. Implement the design using suitable technology including but not limited to social robots\, chat bots\, virtual agents\, TTS\, STT\, computer vision\, LLMs\, etc...
4. Analyse and evaluate the implemented design using an appropriate experimental setup with human test subjects.
5. Collaborate as a group.
6. Present and discuss your research project in oral and written form.
Schedule
Teaching method
Lectures: On-site.
Project: On-site (lab) and in Robot lab: a group project
Assesment method
Exam (individual) (50%) (must be >=5 to pass course)
Project (group project) (50%) (must be >=5 to pass course)
The project grade is based on a live demo + discussion of the system (50%), and a written paper (50%).
passing the course means average grade >=5.5
Important: The exam is held half-way the course. You must pass the exam >=5 to be able to continue the course. If you fail the first time, there is one resit two weeks later. If you fail the resit, you may not participate in the project anymore that year.
There is no resit possibility for the project. Moments of feedback, help and instruction is given throughout the course. Therefore this project is considered a practical test with sufficient learning moments to enable students to pass the test.
Resit, review & feedback
The exam is held half-way the course. You must pass the exam >=5 to be able to continue the course. If you fail the first time, there is one resit two weeks later. If you fail the resit, you may not participate in the project anymore that year.
The instructor will inform the students about when the review and debriefing of the exam takes place. This will be done based on a 1-on-1 feedback session with the student going through the answers of the exam.
There is no resit possibility for the project. Moments of feedback, help and instruction is given throughout the course. Therefore this project is considered a practical test with sufficient learning moments to enable students to pass the test.
Reading list
Bartneck et al. "Human-Robot Interaction: An Introduction." (online! dont buy it)
Selected papers and articles referenced during class.
Registration
Contact
For course content questions: Joost Broekens (see lecturer)
For admin questions: contact the bachelor coordinator for DSAI.
Remarks
AI use in this course is limited to the following:
1. You may use Generative AI to help in searching literature or providing feedback on your ideas (so as an interactive sparring partner).
2. You may NOT use Gen AI to generate texts for your paper or research proposal. You may use it to enhance your own text\, remove syntactical or otherwise disturbing phrases.
3. You may use Generative AI to create visuals for your presentations/paper.
4. You may use Gen AI to evaluate how your proposal and/or paper does on the evaluation criteria of those deliverables.
5. You may NOT ask Gen AI to optimize your written deliverables for point 4.
6. You may use Gen AI for coding.
You are the final responsible for your understanding of and your ability to discuss your deliverables, as well as the correct functioning of the HAI system for your project.