(SOT82145) Computer-Assisted Data Analysis - Exercise

exercise

The module consists of a seminar and an exercise.

This module introduces students to the principles and practice of computer-assisted data analysis. It focuses on how empirical data can be systematically cleaned, explored, analysed, and communicated using modern analytical tools. Rather than treating software as an end in itself, the course emphasises the logic of the data analysis process: transforming raw data into interpretable evidence that supports analytical reasoning and informed decision-making.
Content:
• Introduction to data and data analysis workflows
• Data cleaning and preparation
• Exploratory data analysis
• Basic statistical analysis
• Data visualisation in R
• Principles of effective visual communication
• Dashboard creation and interactive visual analytics in Tableau
• Applied data analysis project

Held by Dalia Ali, M.Sc.

(SOT82145) Computer-Assisted Data Analysis - Seminar

seminar

The module consists of a seminar and an exercise.

Module content:
• Introduction to data and data analysis workflows
• Data cleaning and preparation
• Exploratory data analysis
• Basic statistical analysis
• Data visualisation in R
• Principles of effective visual communication
• Dashboard creation and interactive visual analytics in Tableau
• Applied data analysis project

Held by Dalia Ali, M.Sc.

(SOT86065) Machine Learning and Society (part of 6 ECTS module)

seminar

Further information (ECTS, description of assessment methods, etc.) can be found in the module description in TUMonline.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86065, SOT86066) Machine Learning and Society

lecture

This lecture sheds light into the profound impact of machine learning (ML) on society, emphasizing the ethical, societal, and technical challenges posed by advanced algorithms. It aims to equip students with a critical understanding of the implications of ML technologies and the responsibility of practitioners in this field.
1. Algorithmic Bias
Explore the concept of bias in machine learning algorithms.
Understand how bias originates from data sources and algorithmic design.
Examine case studies where algorithmic bias has impacted decision-making in areas like recruitment, criminal justice, and lending.
2. Fairness: Formal Definitions
Introduce formal definitions of fairness in the context of machine learning.
Analyze different models of fairness and their applicability in various scenarios.
Discuss the trade-offs and challenges in implementing these models in real-world applications.
3. Alignment of Large Language Models (LLMs)
Investigate the alignment challenges in large language models.
Discuss strategies for aligning LLMs with human values and ethical norms.
Evaluate the effectiveness and limitations of these alignment strategies through practical examples.
4. Red Teaming
Learn about the practice of red teaming in the context of machine learning.
Understand how adversarial approaches can identify vulnerabilities in ML systems.
Study real-life instances where red teaming has been used to improve system robustness and safety.
5. Auditing
Examine the role of auditing in ensuring the ethical use of machine learning.
Discuss methodologies for conducting effective audits of ML systems.
Assess the impact of regulatory and compliance requirements on ML auditing processes.
6. Limits of Prediction
Investigate the theoretical and practical limits of predictive modeling in machine learning.
Understand the implications of these limits on the reliability and utility of ML predictions.
Explore scenarios where over-reliance on predictive models can lead to adverse outcomes.
7. Algorithmic Harms
Define and categorize different types of algorithmic harms.
Analyze the societal impact of these harms, with a focus on vulnerable and marginalized groups.
Discuss strategies to mitigate and prevent algorithmic harms, including policy interventions and technical solutions.
8. ML in Society
Understand how ML systems are influenced by and, in turn, influence social structures, norms, and values.
Discuss case studies that illustrate the deep interconnection between ML applications and social dynamics.
9. ML as Society
Understand the collaborative and often complex interactions between these stakeholders in the development and deployment of ML systems.
Examine the roles and responsibilities of each stakeholder group, emphasizing the ethical considerations at each stage of ML development.
Explore the concept of ML systems as socio-technical systems, where technical aspects are deeply intertwined with social, ethical, and organizational dimensions.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86106) AI Applications in Non-Governmental Organizations

lecture with integated exercise

This module consists of a seminar and an exercise.
This module will explore the practical applications of Artificial Intelligence in achieving the United Nations Sustainable Development Goals. Through seminars, interactive workshops, and guest talks by NGO practitioners, students will learn how AI is used to improve operations, decision-making, and outreach in the non-profit sector. Real-life case studies by guest speakers and their experience in the field will create an opportunity for students to network with professionals in the field. In addition to theoretical learning, students will also be placed into groups where they will be tasked to solve real-world projects offered by partner non-governmental organisations (NGOs). This approach aims at allowing the application of knowledge gained, making a difference through projects that matter, and preparing students for potential careers in the non-profit sector.

Held by Dalia Ali, M.Sc.

(SOT86138) Persuasion, Cooperation, and Deception in LLM-Based Agents - Exercise

exercise

AI agents powered by large language models (LLMs) are increasingly deployed to perform tasks in digital environments, including information retrieval, decision-making, and crucially, manipulating opinions in online discourse. As these systems move from single-agent task execution toward multi-agent settings, understanding how agents interact, persuade, cooperate, and compete through natural language becomes critical. To analyse the persuasive capabilities of AI agents in an environment, this project conducts a study on how agents’ behaviors emerge from prompting strategies in an environment (The Settlers of Catan) where communication directly influences outcomes.

Held by Dalia Ali, M.Sc.

(SOT86138) Persuasion, Cooperation, and Deception in LLM-Based Agents - Seminar

seminar

AI agents powered by large language models (LLMs) are increasingly deployed to perform tasks in digital environments, including information retrieval, decision-making, and crucially, manipulating opinions in online discourse. As these systems move from single-agent task execution toward multi-agent settings, understanding how agents interact, persuade, cooperate, and compete through natural language becomes critical. To analyse the persuasive capabilities of AI agents in an environment, this project conducts a study on how agents’ behaviors emerge from prompting strategies in an environment (The Settlers of Catan) where communication directly influences outcomes.

Held by Dalia Ali, M.Sc.

PhD Kolloquium - Societal Computing

colloquium

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86074) Aligning Generative AI to Social Values (part of 6 ECTS module)

seminar

This module explores the intersection of generative AI and social values, focusing on ethical considerations, bias mitigation, and the development of AI systems that reflect diverse societal values. Students will engage with theoretical frameworks, practical tools, and case studies to understand and address the ethical challenges in generative AI. Course Content: 1. Introduction to Generative AI and Social Values Overview of generative AI technologies Importance of aligning AI with societal values Ethical principles in AI (fairness, accountability, transparency, inclusivity) 2. Ethical Frameworks and Principles Key ethical theories and their application to AI Existing frameworks for ethical AI Case studies on ethical AI failures and successes 3. Understanding and Mitigating Bias in Generative AI Types and sources of bias in AI models Methods for detecting and mitigating bias Practical exercises in bias identification and mitigation 4. Transparency and Accountability in AI Importance of transparency in AI systems Techniques for improving transparency Ensuring accountability in AI deployment 5. Auditing and Red Teaming Approaches Introduction to auditing AI systems Red teaming methodologies for stress-testing AI models Hands-on projects in auditing and red teaming generative AI 6. Aligning Large Language Models with Societal Values Challenges specific to large language models (e.g., GPT, BERT) Techniques for aligning language models with ethical standards Practical projects in fine-tuning and aligning language models 7. Balancing Innovation and Ethical Responsibility Case studies on innovation vs. ethical dilemmas in AI Strategies for responsible AI development Policy implications and regulatory considerations.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86097, SOT86074) Aligning Generative AI to Social Values

lecture

This module explores the intersection of generative AI and social values, focusing on ethical considerations, bias mitigation, and the development of AI systems that reflect diverse societal values. Students will engage with theoretical frameworks, practical tools, and case studies to understand and address the ethical challenges in generative AI. Course Content: 1. Introduction to Generative AI and Social Values Overview of generative AI technologies Importance of aligning AI with societal values Ethical principles in AI (fairness, accountability, transparency, inclusivity) 2. Ethical Frameworks and Principles Key ethical theories and their application to AI Existing frameworks for ethical AI Case studies on ethical AI failures and successes 3. Understanding and Mitigating Bias in Generative AI Types and sources of bias in AI models Methods for detecting and mitigating bias Practical exercises in bias identification and mitigation 4. Transparency and Accountability in AI Importance of transparency in AI systems Techniques for improving transparency Ensuring accountability in AI deployment 5. Auditing and Red Teaming Approaches Introduction to auditing AI systems Red teaming methodologies for stress-testing AI models Hands-on projects in auditing and red teaming generative AI 6. Aligning Large Language Models with Societal Values Challenges specific to large language models (e.g., GPT, BERT) Techniques for aligning language models with ethical standards Practical projects in fine-tuning and aligning language models 7. Balancing Innovation and Ethical Responsibility Case studies on innovation vs. ethical dilemmas in AI Strategies for responsible AI development Policy implications and regulatory considerations.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86122) AI-Powered Social Systems: Computational Social Science and Agentic Intelligence - Seminar 

seminar

This module consists of a seminar and an exercise. This module introduces students to computational social science with a focus on AI agents and their impact on social systems. It covers topics such as social network analysis, natural language processing, big data ethics, and agent-based modelling. Students will engage in hands-on projects to analyse digital social dynamics and simulate real-world scenarios across sectors such as governance, NGOs, and the private sector.

Held by Dalia Ali, M.Sc.

(SOT86122) AI-Powered Social Systems: Computational Social Science and Agentic Intelligence - Exercise

exercise 

This module consists of a seminar and an exercise. This module introduces students to computational social science with a focus on AI agents and their impact on social systems. It covers topics such as social network analysis, natural language processing, big data ethics, and agent-based modelling. Students will engage in hands-on projects to analyse digital social dynamics and simulate real-world scenarios across sectors such as governance, NGOs, and the private sector.

Held by Dalia Ali, M.Sc.

PhD Kolloquium - Societal Computing

colloquium (free subject)

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86065, SOT86066) Machine Learning and Society 

lecture

This lecture sheds light into the profound impact of machine learning (ML) on society, emphasizing the ethical, societal, and technical challenges posed by advanced algorithms. It aims to equip students with a critical understanding of the implications of ML technologies and the responsibility of practitioners in this field. 1. Algorithmic Bias Explore the concept of bias in machine learning algorithms. Understand how bias originates from data sources and algorithmic design. Examine case studies where algorithmic bias has impacted decision-making in areas like recruitment, criminal justice, and lending. 2. Fairness: Formal Definitions Introduce formal definitions of fairness in the context of machine learning. Analyze different models of fairness and their applicability in various scenarios. Discuss the trade-offs and challenges in implementing these models in real-world applications. 3. Alignment of Large Language Models (LLMs) Investigate the alignment challenges in large language models. Discuss strategies for aligning LLMs with human values and ethical norms. Evaluate the effectiveness and limitations of these alignment strategies through practical examples. 4. Red Teaming Learn about the practice of red teaming in the context of machine learning. Understand how adversarial approaches can identify vulnerabilities in ML systems. Study real-life instances where red teaming has been used to improve system robustness and safety. 5. Auditing Examine the role of auditing in ensuring the ethical use of machine learning. Discuss methodologies for conducting effective audits of ML systems. Assess the impact of regulatory and compliance requirements on ML auditing processes. 6. Limits of Prediction Investigate the theoretical and practical limits of predictive modeling in machine learning. Understand the implications of these limits on the reliability and utility of ML predictions. Explore scenarios where over-reliance on predictive models can lead to adverse outcomes. 7. Algorithmic Harms Define and categorize different types of algorithmic harms. Analyze the societal impact of these harms, with a focus on vulnerable and marginalized groups. Discuss strategies to mitigate and prevent algorithmic harms, including policy interventions and technical solutions. 8. ML in Society Understand how ML systems are influenced by and, in turn, influence social structures, norms, and values. Discuss case studies that illustrate the deep interconnection between ML applications and social dynamics. 9. ML as Society Understand the collaborative and often complex interactions between these stakeholders in the development and deployment of ML systems. Examine the roles and responsibilities of each stakeholder group, emphasizing the ethical considerations at each stage of ML development. Explore the concept of ML systems as socio-technical systems, where technical aspects are deeply intertwined with social, ethical, and organizational dimensions.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86065) Machine Learning and Society (part of 6 ECTS module) 

seminar

Further information (ECTS, description of assessment methods, etc.) can be found in the module description in TUMonline.

Held by Prof. Dr. Orestis Papakyriakopoulos

PhD Kolloquium - Societal Computing

colloquium (free subject)

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86074) Aligning Generative AI to Social Values (part of 6 ECTS module)

seminar

This module explores the intersection of generative AI and social values, focusing on ethical considerations, bias mitigation, and the development of AI systems that reflect diverse societal values. Students will engage with theoretical frameworks, practical tools, and case studies to understand and address the ethical challenges in generative AI.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86097, SOT86074) Aligning Generative AI to Social Values

lecture

This module explores the intersection of generative AI and social values, focusing on ethical considerations, bias mitigation, and the development of AI systems that reflect diverse societal values. Students will engage with theoretical frameworks, practical tools, and case studies to understand and address the ethical challenges in generative AI.

Held by Prof. Dr. Orestis Papakyriakopoulos

PhD Colloquium - Societal Computing

colloquium (free subject)

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86065) Machine Learning and Society (6 ECTS)

lecture

This lecture sheds light into the profound impact of machine learning (ML) on society, emphasizing the ethical, societal, and technical challenges posed by advanced algorithms. It aims to equip students with a critical understanding of the implications of ML technologies and the responsibility of practitioners in this field.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86065) Machine Learning and Society (part of 6 ECTS module)

seminar

This lecture sheds light into the profound impact of machine learning (ML) on society, emphasizing the ethical, societal, and technical challenges posed by advanced algorithms. It aims to equip students with a critical understanding of the implications of ML technologies and the responsibility of practitioners in this field.

Held by Prof. Dr. Orestis Papakyriakopoulos

(SOT86066) Machine Learning and Society (lecture for 3 ECTS)

This lecture sheds light into the profound impact of machine learning (ML) on society, emphasizing the ethical, societal, and technical challenges posed by advanced algorithms. It aims to equip students with a critical understanding of the implications of ML technologies and the responsibility of practitioners in this field.

Held by Prof. Dr. Orestis Papakyriakopoulos