Artificial Intelligence for a Better Future. An Ecosystem Perspective on the Ethics of AI and Emerging Digital Technologies is the latest book from Bernd Carsten Stahl, and part of the Springer Briefs in Research and Governance series (see also this review of Katharina Jarmai’s book in the same collection). Regular readers will know the author’s work from posts on the Foundation website dating from as far back as 2013, Professor Stahl having played an important role in both the theoretical development and implementation of Responsible Innovation approaches.
As the title suggests, this brief book offers the novel categorization of AI as a metaphoric ecosystem, the application of which helps in describing and understanding the complexity of the technological context and debate around ethics that AI developments have created. The metaphor also provides the structure through which the author offers insight into practical interventions for approaches to managing AI ecosystems.
Human Flourishing
Within this ecosystem analysis framework, Stahl proposes an ethical approach based on the concept of human flourishing, a concept that ‘does not commit us to a particular way of life or require the adoption of a particular ethical position’ but reflects the values and approaches of Responsible Innovation.
This concept is drawn from a historical approach to flourishing ethics with these key tenets:
1. Human flourishing is central to ethics.
2. Humans as social animals can only flourish in society.
3. Flourishing requires humans to do what we are especially equipped to do.
4. We need to acquire genuine knowledge via theoretical reasoning and then act autonomously and justly via practical reasoning in order to flourish.
5. The key to excellent practical reasoning and hence to being ethical is the ability to deliberate about one’s goals and choose a wise course of action.
Artificial Intelligence
The book flows effortless through a series of clear chapters: A discussion of perspectives on AI leads into those of the purpose of AI development. 3 major categories are distilled from a spectrum of uses and applications: AI for efficiency, AI for social control and AI for human flourishing, the author demonstrating how these categories are not exclusive but linked with several interesting and thought-provoking examples. Social control has a role to play in human flourishing, as does efficiency and the economic benefits it gleans.
As the argument unfolds, ethical issues and benefits brought on by the development of AI are discussed via empirical accounts, before the author moves on to reviewing some of the proposals that have been put forward to address some of these ethical issues. Stahl then proposes his own vision of approaching AI from an ecosystem perspective, allowing him to highlight the challenges that arise for the ethical governance of such systems before offering some recommendations for interventions that are likely to be able to shape AI ecosystems in ways that are conducive to human flourishing.
AI Ecosystems
After presenting a number of requirements for shaping the AI ecosystem, the author highlights ‘some key aspects of governance, mitigation and interventions that have a high likelihood of making a positive contribution to the aim of shaping AI ecosystems in desired ways’. Within these aspects we find the use of specific terminology rather than AI as a blanket term, recognizing the relationship between excellence and flourishing, understanding expected impacts and the importance of communication, knowledge and capacity building.
As we might imagine from the earlier key tenets and their familial relationship with RI, stakeholder engagement and the problem of understanding societal preferences play a pivotal role, alongside those of regulation and enforcement.
In the Author’s words
The following summary of the book’s analysis of how an ecosystem approach to ethics might be structured is taken from the conclusion, and really encapsulates the book’s message:
The analysis pointed to three groups of requirements that interventions into AI ecosystems need to fulfil, in order to increase their chances of successfully promoting human flourishing:
• Interventions need to clearly delineate the boundaries of the ecosystem: Systems boundaries are not necessarily clear and obvious. In order to support AI ecosystems, the boundaries of the ecosystem in question need to be clearly located. This refers not only to geographical and jurisdictional boundaries, but also to conceptual ones, i.e. the question of which concept of AI is the target of intervention and which ethical and normative concepts are at the centre of attention.
• Interventions need to develop, support, maintain and disseminate knowledge: The members of AI ecosystems require knowledge, if they are to work together to identify ethically desirable future states and find ways of working towards those. AI as a set of advanced technologies requires extensive subject expertise in the technologies, their capacities and uses. In addition, AI ecosystems for human flourishing require knowledge about concepts and processes that support and underpin ethical reflections. And, finally, AI ecosystems need mechanisms that allow for these various bodies of knowledge to be updated and made available to members of those ecosystems who need them in a particular situation.
• Interventions need to be adaptive, flexible and able to learn: The fast-moving nature of AI-related innovation and technology development, but also of social structures and preferences as well as adjacent innovation ecosystems, means that any intervention into the AI ecosystem needs to incorporate the possibility and, indeed, likelihood of change.
Governance structures therefore need to be flexible and adaptable. They need to be open to learning and revisions. They need to be cognisant of existing responsibilities and must build and shape these to develop the ecosystem in the direction of human flourishing.
Summary
This Brief flows beautifully, is very easy to read and suitable across the board (it does not require previous knowledge of ethics). Stahl is able to summarize entire histories and approaches very clearly and precisely, leading the reader through the chapters and the construction of his argument. Several diagrams appear that help in demonstrating the scheme as well as the interconnected and complex nature of the debate around ethics and AI, and while not being prescriptive the book proposes a series of framework and approaches within which this novel analysis sits.
Artificial Intelligence for a Better Future. An Ecosystem Perspective on the Ethics of AI and Emerging Digital Technologies is available Open Access through a Creative Commons License, and free to download here. I would certainly recommend it to all of our readers.
——————