SEED-SET: Scalable Evolving Experimental Design for System-level Ethical Testing
This addresses the challenge of evaluating ethical alignment in autonomous systems like drones, which is critical for safety and bias prevention in high-stakes domains.
The paper tackles the problem of automated ethical benchmarking for autonomous systems by proposing SEED-SET, a Bayesian experimental design framework that incorporates objective evaluations and subjective stakeholder judgments, resulting in up to 2× optimal test candidates and 1.25× improvement in coverage compared to baselines.
As autonomous systems such as drones, become increasingly deployed in high-stakes, human-centric domains, it is critical to evaluate the ethical alignment since failure to do so imposes imminent danger to human lives, and long term bias in decision-making. Automated ethical benchmarking of these systems is understudied due to the lack of ubiquitous, well-defined metrics for evaluation, and stakeholder-specific subjectivity, which cannot be modeled analytically. To address these challenges, we propose SEED-SET, a Bayesian experimental design framework that incorporates domain-specific objective evaluations, and subjective value judgments from stakeholders. SEED-SET models both evaluation types separately with hierarchical Gaussian Processes, and uses a novel acquisition strategy to propose interesting test candidates based on learnt qualitative preferences and objectives that align with the stakeholder preferences. We validate our approach for ethical benchmarking of autonomous agents on two applications and find our method to perform the best. Our method provides an interpretable and efficient trade-off between exploration and exploitation, by generating up to $2\times$ optimal test candidates compared to baselines, with $1.25\times$ improvement in coverage of high dimensional search spaces.