Responsible Innovation in the Digital World: Designing AI with Human Rights and Values

Short Summary

The Royal Institution talk explores responsible innovation in AI and digital technologies, arguing that design must be guided by human rights and core values. It highlights the need to bridge engineering and ethics through a design-by-values approach, showcases Dutch and European examples, and discusses practical methods such as privacy by design, explainable AI, and value-driven requirements engineering to ensure technology serves humanity while addressing risks and geopolitical challenges.

Introduction and Core Message

The presentation, delivered at The Royal Institution, centers on responsible innovation in the digital world as a process that welds ethics with new technologies. The speaker asserts that this concept originated in The Hague, Netherlands, and has since spread globally, influencing how policymakers, engineers, and ethicists think about technology. A central theme is the design perspective: unlike merely describing or predicting the world, designing the world involves imagining how it could be and bridging the gap between utopian values and practical engineering. This bridging is the space where engineers and ethicists converge to articulate specifications that realize a possible world where technology benefits humanity while aligning with fundamental rights and values.

AI as a World-Defining Technology

The talk highlights a geopolitical dimension of AI, noting statements from major tech leaders about the transformative potential and risks of AI, including existential threats and the risk of bias in deployment. The speaker emphasizes that AI is a master technology evolving rapidly, with billions of users and red queen dynamics driving relentless advancement. While some commentators warn about existential threats, others stress the practical harms of biased, opaque, and privacy-invasive systems. The speech references prominent voices and works on the dangers of generative AI and urges a balanced focus on immediate societal changes as well as long-term risks.

From Fear to Responsible Design

Despite concerns about existential risk, the core argument is pragmatic: we must pursue responsible innovation that solves urgent problems in health care, energy transition, circular economy, and mobility, while upholding privacy, autonomy, sustainability, and rights. The speaker advocates a design-by-values approach where ethical principles are translated into concrete design requirements that engineers can implement. This makes ethics actionable rather than abstract, moving from high level values to mid-level norms such as data minimization, accountability, data quality, and security.

Values by Design and Conceptual Engineering

The talk introduces a three-part relationship between design and values: valuing is design consequential, designs are not value neutral, and tensions between values can be overcome by design. The methodology, termed conceptual engineering, involves breaking down abstract values into actionable requirements and evaluating trade-offs. It frames moral decision making as a technical exercise: when given several competing obligations, engineers should decompose the problem and specify how to meet each obligation through design choices. This creates an empirical design cycle where implemented solutions generate feedback and refinements for future iterations.

Concrete Examples: Dutch Innovations

The speaker showcases several Dutch designs to illustrate how a single artifact can satisfy multiple ethical desiderata, such as the FairPhone, which features modular components, replaceable batteries, and sourcing from non-authoritarian regions, thereby aligning with privacy, sustainability, and labor considerations. Other examples include a bus stop in Utrecht that captures rainwater, supports biodiversity, and mitigates heat stress, and a foldable shipping container approach that reduces empty loads and greenhouse gas emissions. These examples demonstrate how design can embed ethics into everyday infrastructure and consumer products.

Privacy, Security, and Explainability as Design Goals

The talk discusses practical privacy-preserving techniques such as coarse graining, differential privacy, and homomorphic encryption for privacy-preserving machine learning. It also outlines the goal of explainable AI, arguing that transparent, controllable systems are essential for accountability. The WHO's guidance on AI in healthcare is cited, encouraging design that conforms to human rights and is demonstrably aligned with core values. In addition, the talk stresses the need for design to address issues of coercive control, as seen in digital devices that could facilitate abuse, and emphasizes designing to prevent such misuse.

Second-Order Obligations and the Moral Axion

A key principle introduced is the moral axiom of responsible innovation: if new technology can satisfy more of our obligations today than it could yesterday, there is a moral imperative to innovate. This principle supports a strategy of building tools that enable both privacy protection and functionality, such as privacy-preserving techniques that do not sacrifice essential services. The speaker uses the concept of second-order obligations to argue that designers should ensure their systems enable multiple values to be realized concurrently rather than trading off one value for another.

European Policy and Global Implications

The discussion reflects on European Union regulations and the GDPR, while noting ongoing debates about AI governance, transparency, and human rights. The speaker emphasizes the importance of a culture of technology by design for Western liberal democracies, backed by the UN Sustainable Development Goals and the technology facilitation mechanisms that connect science with policy. He argues that responsible AI is not an obstacle to progress but a necessary condition for moral progress and societal trust.

Methodologies: From Abstract Concepts to Concrete Designs

Examples and Principles in Practice

The speaker presents a drone designed for blood sample transport and other examples illustrating how ethical considerations can be integrated at the design stage. He emphasizes modularity, replaceable components, and sustainable material choices as practical embodiments of ethics in engineering. The overarching message is that values like privacy and sustainability should not be afterthoughts but integrated into the core architecture of systems and devices, ensuring trustworthy, human-centered technology.

Conclusion: Toward a Moral, Yet Innovative Future

In closing, the talk advocates a balanced, value-driven approach to AI and digital technologies, urging the adoption of a culture of responsible innovation that harmonizes ethics with progress. The trolley problem is used to illustrate design opportunities that remove moral dilemmas through thoughtful architecture, reminding us that the way we build systems shapes the ethical outcomes they produce. The speaker invites continued collaboration across public and private sectors to realize a future where technology serves human rights, dignity, and the public good.