Human-AI Interaction

Artificial intelligence (AI) has been introduced to many realms of human activity. AI already profoundly impacts our society, and will only increase in importance in the future. On a daily basis, people now interact with machines, computer systems, and robots that can perceive and learn from their environments, make adaptive decisions, and even take actions in the physical world without much human intervention. While AI technologies are expected by many to benefit the human race and improve the human condition, others are concerned about potential negative consequences of AI. For example, intelligent machines may replace humans (e.g., job loss) or incompetent AI systems may cause fatal errors (e.g., accidents by self-driving cars). A human-centered approach to AI is thus needed in which we develop AI technologies with a clear understanding of how the advantages and limitations of AI are complementary to – or clash against ­– the strengths and weaknesses of human cognition. This course is designed to help students, our next-generation scientists and engineers, to comprehend and anticipate the current and future impacts of AI on human-technology interactions and human-technology relations, and to be familiarized with principles and tools that can help to engineer AI systems that benefit human well-being.

The course is roughly divided into four modules. Module 1 teaches the fundaments of AI and human-AI interaction, including the basics of machine learning (ML) and a historical perspective of human-computer interaction. Module 2 continues with a human-centered approach of ML and explains how to make state-of-art ML models interpretable and interactive (e.g., human-in-the-loop ML). Module 3 discusses the principles of designing and evaluating interfaces for human-AI interactions and how to apply these principles to practice. Module 4 focuses on the social and societal aspects of human-AI interaction (e.g., fairness in ML, data privacy and ethics, social perception of embodied AI, etc.) and concludes the course with a future outlook. Teaching methods in this course include lectures, interactive debate and discussion, and group assignments.

Additional information test
There will be a group assignment where students apply knowledge learned from the lectures to a practical project of designing and evaluating an AI-infused user interface (during Module 2 and 3). Finally, there will be an exam at the end of the course with multiple choices as well as open-ended questions.

For PDEng students from the User-AI Interaction program, the end exam will be replaced with a design/engineering assignment where they apply the knowledge learned from the course to one of their ongoing projects.

After this course, the students are able to:

• Critically assess the state of art in AI applications, including its strengths and limitations, benefits and risks, from a human-centered perspective
• Apply the principles and techniques of explainable and interactive AI to practical problems
• Design and evaluate AI-infused systems in view of human values
• Analyze the causes of biases in AI algorithms and formulate solutions to address them
• Discuss and debate on the social and societal issues on human-AI interaction and human-AI relation

You must meet one of the following collections of requirements

• Collection 1
• Completed Final examination Bsc program
• Collection 2
• Completed Pre-Master

• Reading materials will be provided by the course lecturers and will be available on Canvas. They mainly consist of published papers, book chapters, and blog posts written by experts in the field.