The Runtime Dimension of Ethics in Self-Adaptive Systems

Self-adaptive systems increasingly operate in close interaction with humans, often sharing the same physical or virtual environments and making decisions with ethical implications at runtime. Current approaches typically encode ethics as fixed, rule-based constraints or as a single chosen ethical theory embedded at design time. This overlooks a fundamental property of human-system interaction settings: ethical preferences vary across individuals and groups, evolve with context, and may conflict, while still needing to remain within a legally and regulatorily defined hard-ethics envelope (e.g., safety and compliance constraints). This paper advocates a shift from static ethical rules to runtime ethical reasoning for self-adaptive systems, where ethical preferences are treated as runtime requirements that must be elicited, represented, and continuously revised as stakeholders and situations change. We argue that satisfying such requirements demands explicit ethics-based negotiation to manage ethical trade-offs among multiple humans who interact with, are represented by, or are affected by a system. We identify key challenges, ethical uncertainty, conflicts among ethical values (including human, societal, and environmental drivers), and multi-dimensional/multi-party/multi-driver negotiation, and outline research directions and questions toward ethically self-adaptive systems.

💡 Research Summary

The paper “The Runtime Dimension of Ethics in Self‑Adaptive Systems” argues that the prevailing practice of embedding static, rule‑based ethical constraints or a single pre‑selected ethical theory into self‑adaptive software is fundamentally insufficient for today’s AI‑enabled systems that interact closely with humans. The authors introduce the concept of “runtime ethics,” distinguishing between hard ethics (non‑negotiable legal, regulatory, and safety constraints) and soft ethics (the under‑determined space of value‑driven choices that must remain within the hard‑ethics envelope). They contend that soft‑ethical preferences—such as privacy, fairness, environmental impact, or collective welfare—should be treated as dynamic runtime requirements that are continuously elicited, represented, negotiated, and revised as contexts and stakeholder groups evolve.

Using an autonomous environmental‑monitoring drone as a running example, the paper illustrates how the system must adapt flight paths, sensing modalities, and data‑collection strategies while respecting hard constraints (airspace safety) and balancing competing soft values (data quality vs. privacy vs. wildlife disturbance). This example demonstrates that ethical preferences cannot be reduced to traditional functional requirements because they are subjective, context‑dependent, and often lack crisp satisfaction criteria, a condition the authors label “ethical uncertainty.”

The core technical challenges identified are: (1) modeling and reasoning under ethical uncertainty; (2) handling multi‑party, multi‑dimensional conflicts among human, societal, and environmental values; (3) designing negotiation mechanisms that can operate over non‑commensurable ethical values while guaranteeing compliance with hard ethics; and (4) providing explainable, transparent outcomes to affected stakeholders. Existing automated negotiation research focuses on economic utilities and assumes well‑defined, comparable preferences, which does not translate to ethical values that are often fuzzy, probabilistic, or qualitative.

To address these challenges, the authors propose extending the classic MAPE‑K self‑adaptive loop with dedicated stages for ethical requirement monitoring, analysis, and negotiation. They outline a research agenda comprising five pillars: (i) formalization and runtime extraction of ethical requirements (e.g., via ontologies, surveys, sensor data); (ii) representation of ethical uncertainty (e.g., fuzzy sets, Bayesian networks, multi‑criteria decision models); (iii) multi‑dimensional, multi‑party negotiation protocols (e.g., Pareto‑optimal, minimum‑loss, evolutionary negotiation); (iv) runtime verification that negotiation outcomes stay within hard‑ethics bounds (formal methods, safety contracts); and (v) explainability and transparency mechanisms (natural‑language explanations, visualizations, justification traces).

Each pillar is accompanied by concrete research questions such as: How can stakeholder preferences be quantified in real time and integrated with uncertain value models? What algorithms can efficiently compute Pareto‑optimal trade‑offs under hard‑ethics constraints? How can negotiation outcomes be audited and justified to non‑technical users? The paper emphasizes that addressing these questions requires interdisciplinary collaboration across ethics, value‑sensitive design, runtime requirements engineering, automated negotiation, and formal verification.

In conclusion, the authors argue that moving from static ethical rules to runtime ethical reasoning is essential for the safe, trustworthy deployment of self‑adaptive systems in real‑world socio‑technical environments. By treating ethics as a dynamic, negotiable requirement, future systems can better accommodate pluralistic values, adapt to evolving regulations, and provide accountable, explainable decisions that respect both hard legal mandates and the nuanced soft‑ethical expectations of diverse human stakeholders.