MAIC is a project funded by the Swiss National Science Foundation (national research program 77 "Digital Transformation"). MAIC investigates control of artificially intelligent systems by means of empirical experiments.
Team: Thomas Burri (PI) and Juliane Beck (HSG, law), Markus Christen and Serhyi Kandul (UZH, philosophy and economics), François Fleuret and Vincent Micheli (UniGe, computer science, machine learning); in advisory capacity: Markus Kneer (UZH, experimental philosophy).
Current status: after an intensive phase of planing with three team retreats, we have conducted a main phase experiment in June 2022 involving 120 test persons in a lab at ETH ZH. We are currentling poring over the data. The next step is another experimental phase with a different focus.
Budget: 0.55 Mio €
Duration: 2021-2024.
Output: The primary output will result from the data we are currently gathering by means of experiments. At the same time, we are working on a formal, publishable literature review on empirical experiments on control of AI.
Our team continues to produce relevant output which is closely linked to MAIC, e.g.:
- Several publications, e.g. "Language Models are Few-Shot Butlers" (Vincent Micheli and François Fleuret), "MineRL Diamond 2021 Competition: Overview, Results, and Lessons Learned" (François Fleuret, Vincent Micheli, et al.), "Can a Robot Lie? Exploring the Folk Concept of Lying as Applied to Artificial Agents" (Markus Kneer), "Guilty Artificial Minds: Folk Attributions of Mens Rea and Culpability to Artificially Intelligent Agents" (Markus Kneer et al.), "Playing the Blame Game with Robots" (Markus Kneer et al.), "Can Artificial Intelligence Make Art?…" (Markus Kneer et al.), "The Ethical Assessment of Autonomous Systems in Practice" (Thomas Burri et al.), and "From 'Human Control' in International Law to Human Oversight in the New EU Act on AI" (Thomas Burri and Juliane Beck, book chapter, submitted).
- Several talks at conferences and workshops for instance at an online workshop of Univ. Cagliari on "Meaningful Human Control" (March 2022, Thomas Burri and Juliane Beck), in Geneva on "Trust with AI" (May 2022, Thomas Burri, with video, from 2h:30m), at HSG on "Perception of efficiency vs. Fairness tradeoffs in AI-based decision making" (May 2022, Serhiy Kandul), at EWAF'22 in Zurich (June 2022, Serhiy Kandul), at FU Berlin on "The Human Decision in International Law" (May 2002, Thomas Burri).
- An expert report on "Rechtliche und ethische Dimensionen der Robotik im Sicherheitsbereich" for the Swiss Center for Drones and Robotics of the Swiss Department of Defense (Markus Christen and Thomas Burri, publication pending).
- An international conference on "Informational Warfare" on 18-19 May 2022 in Trier (Thomas Burri, together w. University of Trier)
- A series of videos for the Swiss Center for Drones and Robotics of the Swiss Department of Defense on "Sollen Roboter Menschen retten?" (Markus Christen, Juliane Beck, Thomas Burri et al.).