Ethical, Legal and Societal aspects of AI and Autonomous Systems (3hp)

Course type:

• AI track: mandatory
• AS track: elective
• Joint curriculum: mandatory

Time: Given yearly, Spring

Teachers:

Virginia Dignum, Leila Methnani, Lilly Jiang, Peter Ericsson, Mattias Brännström, Nina Khairova  (UMU)

Sandra Friberg, Anna-Sara  Lind, and Magnus Strand (UU)

Examiner: Nina Khairova  (UMU)

N/A

LO1. Summarize and identify of the Ethical, Legal, Societal (ELS) challenges that arise upon the development, deployment, and usage of intelligent systems.

LO2. Understand the legal and regulatory context of development and use of AI systems.

LO3. Analyse the decisions made during a system’s lifecycle and their relationship to individual and organisational accountability and responsibility.

LO4. Relate socio-technical mechanisms, within the scope of the Fairness-Accountability-Transparency research community, for the effective socially-beneficial use and governance of AI systems recognise the drawbacks and benefits of core mechanisms.

LO5. Explain and apply the core non-technical skills for the responsible design of Ethical, Legal, Social, AI systems.

By completing the above outcomes, the student will have a fundamental understanding of how intelligent systems influence—and are influenced by—our societies and of the socio-ethical responsibilities they have as developers and users of such tools.

(LO1-5 refer to Learning Outcomes, see above)

The course introduces the fundamental aspects of AI ethics by providing a holistic multidisciplinary view of the discipline. The course structure is such as to introduce students to the impact AI systems have on societies and individuals (LO1) and ongoing state-of-the-art discussions related to ELS aspects of AI (LO2). This introduction will be followed by a critical discussion of where accountability and responsibility lie for ethical, legal, and social impacts of AI systems, considering decision points throughout the development and deployment pipeline (LO3). With this knowledge in mind, students will be introduced to socio-technical approaches for the governance, monitoring and control of intelligent systems as tools for incorporating constraints into intelligent system design (LO4). Finally, learners apply these skills on a simulated responsible AI design problem (LO5). The course is, for simplicity, divided into following thematic modules:

Introduction to AI Ethics: Establishes the motivation behind the field of AI ethics by using real-world use cases related to algorithmic biases, generation of disinformation, and attempts to escape accountability. (LO1)

Introduction to Legal and Regulatory aspects (LO2)

• Guidelines and agencies landscape
• Standarisation initiatives
• Legal constraints and implications
• Legal landscape (AI never operates in a lawless world)

Responsible Development and use of AI

• Responsibility In Design: processes that go around the development, deployment, and usage of a system (e.g. process standards, traceability of decisions, etc) (LO3)
• Responsibility For Design: stakeholders and power distribution identification and analysis; understanding how to balance conflicting ELS requirements (LO3)
• Responsibility By Design (FACTT): system behaviour; e.g. checking and mitigating unwanted biases, ensuring transparency, developing fallback (LO4)
• Responsibility For Designers: the codes of conduct, chain of responsibility, and critical individual decisions that can be made. (LO5).

Each module will be delivered to the students through a combination of teacher and student-led activities. Lectures will provide the theoretical underpinning needed for the practical component. Students will conduct problem-based learning activities where they will discuss—and come up with solutions to—real-world problems related to the field.

Main Text:

Dignum V. (2019). Responsible Artificial Intelligence: How to Develop and Use AI in a Responsible Way, Springer.

Recommended Papers:

Bender E. M., Gebru T., McMillan-Major A., & Shmitchell S. (2021) On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜. In FaccT’21. https://doi.org/10.1145/3442188.3445922

Bryson, J. J., & Theodorou, A. (2019). How Society Can Maintain Human-Centric Artificial Intelligence. In M. Toivonen-Noro, E. Saari, H. Melkas, & M. Hasu (Eds.), Human-Centered Digitalization and Services (pp. 305–323). https://doi.org/10.1007/978-981-13-7725-9_16

Hildebrandt M. (2018). Algorithmic regulation and the rule of law. Philosophical Transactions of Royal Society A. https://doi.org/10.1098/rsta.2017.0355

Theodorou, A., & Dignum, V. (2020). Towards ethical and socio-legal governance in AI. Nature Machine Intelligence, 2(1), 10–12. https://doi.org/10.1038/s42256-019-0136-y

Fiesler, C., Garrett, N., & Beard, N. (2020, February). What do We teach when We teach tech ethics? A syllabi analysis. In Proceedings of the 51st ACM Technical Symposium on Computer Science Education (pp. 289-295). https://dl.acm.org/doi/pdf/10.1145/3328778.3366825

Further relevant reading:

Coeckelbergh M. (2020). AI Ethics, MIT-Press.

Gunkel D. (2020). An Introduction to Communication and Artificial Intelligence, Willie.

Hildebrandt M. (2020). Law for Computer Scientists and Other Folk, Oxford University Press.

O’Neil C. (2016). Weapons of Math Destruction, Crown Books.

Pasquale F. (2014). The Black Box Society: The Secret Algorithms That Control Money and Information, Belknap Press.

To pass the students must attend the whole course and do the assignments.

Assignment: Critically evaluate the ELS aspects of their projects, applying the RAIN tool.