Umělá inteligence a HCI

Course guarantor:

Lecturer:

Tutor:

Supervisor:

Department of Computer Graphics and Interaction

Synopsis:

Students will acquire practical skills in working with artificial intelligence (AI) in the field of HCI and learn to effectively use existing AI tools and models. The course focuses on the practical application of AI, not on the development of new models. Students will learn the basic principles of AI systems using deep learning in HCI and learn to work with available frameworks, pre-trained models, and their applications in interactive systems. Emphasis will be placed on the interpretation of AI outputs and the issues of explainability, interpretability, and trustworthiness from the user's perspective. Students will also understand the modeling of human visual attention and the possibilities of applying these models in HCI. The course will also include an introduction to intelligent interaction devices using AI and an understanding of their practical principles. The teaching will be oriented towards the application of theoretical concepts in real-world scenarios and a critical evaluation of the use of AI in human-technology interaction.

Requirements:

Students should have a basic understanding of user interface design and eye tracking methods. It is recommended that students have at least a basic understanding of UX design and human-computer interaction concepts before the course. The course is taught through lectures, exercises, and project work, with an emphasis on active student involvement in research and testing of designs. Students will work in teams and use the schools e-learning environment to coordinate projects and share outputs.

Syllabus of lectures:

1. AI in HCI: history, motivation, examples of AI applications in HCI

2. User interface using AI (Intelligent UI), devices for interaction using AI in computer vision

3. Dialogue systems, natural speech understanding, voice assistant, interfaces providing AI explanations for users

4. Modeling user behavior using AI methods, egocentric visual attention of the user, modeling salience maps with AI, and UX experiment with eye tracker: preparation, progress, metrics, evaluation

5. Explainable artificial intelligence focused on the user, concepts of interpretability and explainability, and different types of explanations

6. User trust in AI - overtrust or undertrust, calibration of trust, and metrics for their evaluation. User testing of an AI application with Wizard of Oz. Annotation tool for AI data preparation: UX-driven design and annotation optimization.

Syllabus of tutorials:

Goal of the exercises:

Design, prototype, and test an application that includes AI.

This is done through iterative development and also leverages AI tools.

Students work in teams of 23 to design and test an AI-based application throughout the semester.

- In Exercises 12, they propose two ideas, evaluate them using the Google PAIR methodology, and select one concept to develop.

- In Exercises 34, they create a low-fidelity prototype, including personas, brainstorming, and wireframes, and reflect on the use of AI tools in the design process.

- In Exercises 56, students analyze their prototype using mental models and develop a high-fidelity version prepared for testing.

- In Exercises 78, they conduct user testing with eye-tracking tools, compare different methods, and evaluate the results.

- In Exercises 910, they incorporate advanced concepts such as Wizard of Oz, Wizard of Errors, interpretability, and trust into their testing approach.

Throughout the course, teams submit three reports documenting their progress and present updates regularly, ending with a final presentation of their project in Exercises 11-12.

Study Objective:

The course aims to introduce students to the principles of connecting artificial intelligence (AI) and human-computer interaction (HCI) in the context of human-centered AI. Students will gain an overview of how modern AI systems are used in user interfaces and human-centered applications.

The course also focuses on understanding human interaction with devices using AI, including their usability, testing, and design recommendations. Students will also learn the basics of user modeling using AI, especially in the area of visual attention and its use in interface design.

Another goal is to understand the principles of explainable artificial intelligence (XAI) from the user's perspective, including concepts such as interpretability, transparency, and working with uncertainty in the outputs of AI systems.

The course does not emphasize the detailed technical design of AI methods, but rather their practical use in the design and evaluation of user interfaces.

Study materials:

1. Monarch, Robert Munro. Human-in-the-Loop Machine Learning: Active learning and annotation for human-centered AI. Simon and Schuster, 2021.

2. Liao, Q.V., Gruen, D. and Miller, S., 2020, April. Questioning the AI: informing design practices for explainable AI user experiences. In Proceedings of the 2020 CHI conference on human factors in computing systems (pp. 1-15).

3. PASSI, Samir; VORVOREANU, Mihaela. Overreliance on AI literature review. Microsoft Research, 2022, 339. Jg., S. 340.

Note:

-

Further information:

No time-table has been prepared for this course

The course is a part of the following study plans:

• Open Informatics - Extended Reality and Game Development (compulsory elective course)