RETRAI Workshop

Human values, such as fairness, security, autonomy, are increasingly relevant for software: Software is becoming an integral part of our lives. However, it is not obvious how human values and requirements are related, and how human values can be considered in a software development process. It would be advantageous to have a champion for important human values. Could LLMs play this role? In this talk, I will discuss this question and touch on a number of related issues: What do we mean when we talk about “human-LLM collaboration”? Should it follow the patterns of a human-human collaboration?

Kurt Schneider studied computer science in Erlangen and received a PhD in software engineering from Stuttgart University. He held a PostDoc position at the Center for LifeLong Learning and Design at the University of Colorado at Boulder, USA. In 1996, he joined the Daimler Research Center in Ulm as a researcher and manager. Since 2003, Kurt Schneider is a full professor of software engineering at Leibniz Universität Hannover, Germany. He specializes in the intersection of humans with technology, such as requirements engineering, software quality – and human values.

The next generation of AI systems will reason, plan, and act autonomously in increasingly complex environments. Building trust in these systems requires more than trustworthy models, it requires engineering trustworthy AI systems. This keynote presents lessons learned from autonomous cyber operations and discusses how they inform the engineering of human-AI teaming, bounded autonomy, verification, and accountability. The talk concludes with a vision for how these principles can shape the future of trustworthy AI and requirements engineering.

Mehdi Mirakhorli is a Professor of Computer Science at the University of Hawaiʻi at Mānoa and the Founding Director of the Mānoa Indo-Pacific Cybersecurity Institute. His research focuses on trustworthy autonomous AI, offensive cybersecurity, software and systems engineering, and human-AI teaming in high-consequence environments. He has led projects of national importance for the U.S. Department of Homeland Security, the Cybersecurity and Infrastructure Security Agency, the Defense Advanced Research Projects Agency, the U.S. Air Force, the Office of Naval Research, the U.S. Department of Transportation, and the National Institutes of Health.

Over the last few years, various public, private, and NGO entities have adopted a staggering number of non-binding ethical codes to guide the development of artificial intelligence. However, this seemingly failed to drive better ethical practices within AI organizations. In light of this observation, this paper aims to reevaluate the roles the ethics of AI can play to have a meaningful impact on the development and implementation of AI systems. In doing so, we challenge the notion that AI ethics should focus primarily on instilling ethical principles in practitioners within AI organizations, as well as the claim that AI ethics can only lead to ethics washing. We propose a two-pronged institutionalist approach to AI ethics, focusing on shaping organizational decision-making processes and emphasizing the necessity of binding legal regulations. First, we argue that AI ethics should give priority to institutional design over the internalization of ethical principles by individual practitioners. We then contend that legally binding rules are needed to this end, both as a motivation for organizations and to contribute to the semantic determination of high-level ethical principles. We then show that promising proposals to operationalize ethical principles require the backing of binding legal norms to be effective. We conclude by highlighting the potential of AI ethics to contribute meaningfully to legislative innovation in AI governance.

Jocelyn Maclure is full professor of philosophy and Jarislowsky Chair in Human Nature and Technology at McGill University. His current work spans across speculative and practical questions paused by progress in AI research and development. His recent publications appeared in journals such as Minds & Machines, AI & Ethics, AI & Society and Digital Society. As a public philosopher, he chaired the Ethics in Science and Technology Commission of the Quebec Government from 2017 to 2024, and contributed to initiatives such as the Montreal Declaration for the Responsible Development of AI and the UNESCO AI Ethics Recommendation. In 2023, he was Mercator Visiting Professor for AI in the Human Context at the University of Bonn.