The March 2023 value connection workshop

In a ground-breaking interdisciplinary workshop, philosophers, social scientists, computer scientists, and designers gathered to discuss the topic of value alignment. Value alignment refers to the challenge to align the behavior of automated systems, such as deep learning AI models, with legitimate values.

About the Value alignment workshop

The participants discussed various aspects of value alignment, including ethical and philosophical considerations, technical challenges, and design principles. They exchanged insights on how to design AI systems that are capable of understanding and respecting human values, as well as ways to ensure that these values are reflected in the decisions made by these systems.

The workshop was structured around a series of presentations, panel discussions, and group activities. Participants shared their research and practical experiences, and engaged in lively debates about the theoretical and practical dimensions of value alignment. They also explored the potential implications of value misalignment, such as unintended consequences, discrimination, and loss of trust.

The interdisciplinary nature of the workshop proved to be a fruitful approach, as the different perspectives and expertise of the participants enabled them to identify new avenues for research and collaboration. Many of the participants expressed their enthusiasm for continuing the dialogue and pursuing joint projects that address the challenges of value alignment in AI. The newly established network – the Value Connection – will gather again in September to follow up on current collaborative projects and to plan next steps.

Overall, the workshop was a significant step towards bridging the gap between different disciplines and stakeholders in the field of AI, and towards ensuring that AI systems are aligned with human values. The insights and recommendations generated by the workshop will be disseminated widely among academics, policymakers, and industry professionals, with the aim of promoting an approach to AI design and development that embodies the principles of the design for values approach championed at Delft.

The workshop was held at the Trippenhuis in Amsterdam and funded by the Royal Dutch Academy of Arts and Sciences (KNAW) and the Delft Design for Values Institute. It was organized by our MT-member Michael Klenk, in the context of his Early Career Partnership awarded by the KNAW.

https://delftdesignforvalues.nl/wp-content/uploads/2023/04/2217705_640x640-1-500×500.webp