Human-Centric Artificial Intelligence

Recommended Prerequisites

Competências de Programação, Matemática, Inteligência Artificial e Aprendizagem Computacional.

Teaching Methods

Theoretical classes comprise detailed presentation of Human-Centered Artificial Intelligence concepts, principles and fundamental theories.
Practical Lab classes adopt a Project Based Learning approach, directed towards competence acquisition through the development of a laboratory work, comprising three components:
(i) analysis of Human-Centered Artificial Intelligence works described in the literature,
(ii) implementation,
(ii) writing of a scientific article, describing the lab work done, plus its presentation and defense.

Learning Outcomes

The course aims to provide students with knowledge about an AI where human-AI cooperation should predominate, based on considering adequate levels of automation and machine autonomy, without loss of human control, and guided by human objectives and values, towards the promotion of the human condition in its various dimensions, including those outlined in declarations and reference guides such as the United Nations Sustainable Development Goals, the EU AI Act, or the UNESCO Recommendation on the Ethics of AI.
It is expected that students will acquire knowledge and skills to study, design and develop this category of human-centered AI systems, in the context of a multi-agent system formed by humans and AI systems, towards a path of hybrid and collective intelligence, human-AI, that amplifies human intelligence without replacing it.

Work Placement(s)

No

Syllabus

1. Human-centered Artificial Intelligence (AI)
- Agents and Multiagent Systems
- Requirements for Human-centered AI
2. Human-AI Cooperation
- Coordination, Cooperation, Collaboration
- Hybrid Human-Artificial, Collective Intelligence
- Autonomy, control, automation
- Human-in-the-Loop, AI-in-the-Loop, Human-out-of-the Loop
- Human-AI Trustworthiness
3. Human in the loop machine learning
- Sampling training data and human annotation:
- Curious agents and Active Learning
- Data annotation:
- Human, automated, and semi-automated annotators
- Annotation quality control
- Computer interfaces for annotation
- Learning to delegate
- Learning how to behave from human demonstrations, feedback, and natural language:
- Imitation and inverse reinforcement learning
- Reinforcement learning from human feedback
- Inferencing human beliefs, desires, intentions for personalisation
4. Human-AI Communication
- Affective computing
- Natural Language Processing
- Interactive AI
5. Personal assistant agents

Head Lecturer(s)

Luís Miguel Machado Lopes Macedo

Assessment Methods

Assessment
Exam: 40.0%
Project: 60.0%

Bibliography

Russell, S. J., & Norvig, P. (2020). Artificial intelligence: A modern approach (4th ed.). Pearson.
Monarch, R. (2021). Human-in-the-Loop Machine Learning: Active Learning and Annotation for Human-centered AI. Manning Publications.

Settles, B. (2009). Active Learning Literature Survey (Computer Sciences Technical Report1648). University of Wisconsin--Madison.

Russell, S. (2019). Human compatible: Artificial intelligence and the problem of control. Viking.

Shneiderman, B. (2022). Human-centered AI. Oxford University Press.

Molnar, C. Interpretable machine learning. A Guide for Making Black Box Models Explainable, 2019.
https://christophm.github.io/interpretable-ml-book/.

EU AI Act. https://artificialintelligenceact.eu/ (accessed 30/04/2024)

UNESCO, Recommendation on the ethics of artificial intelligence, Digital Library UNESDOC (2020). https://en.unesco.org

United Nations Sustainable Development Goals. https://www.un.org/sustainabledevelopment/ (accessed 30/04/2024)