Objectives and competences
The goal of this course is to provide a comprehensive understanding of the ethics, fairness, safety, and alignment of AI systems, with a focus on their impact across industries, workplaces, and daily life. Students will explore techniques for evaluating AI system effectiveness, ethical considerations, and strategies for addressing bias and fairness in AI. The course will also delve into AI alignment, transparency, and the integration of human oversight in AI-driven decision-making processes, while touching upon the role of generative AI models, such as large language models (LLMs).
Content (Syllabus outline)
• Overview of the current state of generative AI models (such as large language models LLMs) and its impact on various industries, workplace and everyday lives.
• Techniques for evaluating the effectiveness and accuracy of AI systems, including benchmarking and performance metrics.
• Ethics in AI and the potential consequences of AI-driven decision making.
• The role of fairness in AI and the ways in which bias can be introduced into AI systems.
• Current efforts to address the fairness of AI, including algorithmic auditing and bias detection.
• The AI alignment problem and its implications for AI safety and long-term decision making.
• Techniques for building transparent and explainable AI systems, including model interpretability and visualization techniques.
• The impact of data quality and availability on AI decision making, and techniques for ensuring data quality and completeness.
• The role of human oversight and decision making in AI systems, and best practices for integrating AI into decision-making processes.
Learning and teaching methods
• Lectures
• Case studies
• Lab work
• Individual work
Intended learning outcomes - knowledge and understanding
On completion of this course the student will be able to
• Comprehend the current state of generative AI models, their influence on various sectors, and techniques for evaluating AI system effectiveness.
• Recognize ethical considerations, fairness, and bias in AI, and analyze efforts to address these issues, including algorithmic auditing.
• Understand the AI alignment problem, its implications, and strategies for building transparent, explainable AI systems with model interpretability.
• Assess data quality and its impact on AI decision making, and integrate human oversight and best practices into AI-driven processes.
Transferable/Key skills and other attributes:
• Critical thinking and adaptability: Analyzing the impact of generative AI models on industries and evaluating AI systems' effectiveness using benchmarking and performance metrics.
• Ethical reasoning and responsibility: Assessing ethical aspects, potential consequences, and fairness in AI-driven decision making, while identifying and addressing bias in AI systems.
• Technical proficiency and collaboration: Applying techniques for building transparent and explainable AI systems, ensuring data quality, and integrating human oversight in AI decision-making processes.
Readings
• Christian, B., 2020. The alignment problem: Machine learning and human values. WW Norton & Company.
• Tegmark, M., 2018. Life 3.0: Being human in the age of artificial intelligence. Vintage.
• O'neil, C., 2017. Weapons of math destruction: How big data increases inequality and threatens democracy. Crown.
• Karakatič, Sašo, Fister, Iztok. Strojno učenje: s Pythonom do prvega klasifikatorja. 1. izd. Maribor: Univerza v Mariboru, Univerzitetna založba: Fakulteta za elektrotehniko, računalništvo in informatiko, 2022.
