A new collection brings critical literacy, practical governance and social justice into the same conversation. An autopoietic ecology perspective asks how these commitments can be sustained in everyday practice.

Book: Larry Medsker (ed.), AI and Ethics Handbook. Springer, published August 2026, copyright year 2027. xv + 991 pages. Publisher’s record.

Review scope: This preliminary review essay draws on the handbook’s preface, contents and selected chapter abstracts and bibliographies. The chapter discussions use published abstracts rather than full chapter texts. The essay distinguishes the contributors’ stated approaches from the questions those approaches raise for EdgeLab.

Imagine a university introducing an AI system to provide feedback on students’ writing. Its guidance promises fairness, transparency and human oversight. Teachers retain responsibility. Students are told how their data will be used. These are worthwhile commitments. Yet further questions arise as the system becomes part of everyday teaching. Does it encourage a narrower idea of good writing? Can students challenge its advice? Do teachers have enough time to exercise the oversight the policy promises? What happens when a student’s most interesting idea is precisely the one the system struggles to recognise?

This example captures the challenge at the heart of AI ethics: making principles matter in the situations where people live, learn and work. Larry Medsker’s AI and Ethics Handbook offers a substantial entry into that challenge. On the evidence available, its principal strength is the range of perspectives it brings into conversation. Its coverage moves from foundational questions to social institutions, sectoral applications and global contexts. Chapters address disability, environmental resources, cultural rights and Indigenous communities, alongside technical design and governance. The preface presents ethics as integral to the development and use of technology. Preface and contents

The opening chapter, by Tania Duarte, Kathryn Conrad and Ismael Kherroubi Garcia, establishes a valuable educational direction. Their account of critical AI literacy includes understanding the political, economic, cultural and environmental conditions surrounding AI. It recognises that people may have good reasons to resist or refuse particular uses. This makes room for an educational question that deserves sustained attention: what should people be able to question when an institution asks them to use AI? The authors’ stated approach gives learners a role in shaping the conditions of adoption. Critical AI literacy

Medsker’s contribution on systemic racism gives that concern historical depth. Drawing on James Baldwin, Eddie Glaude Jr. and Cornel West, the chapter situates AI within histories of racial inequality and examines how data, objectives and institutional practices can reproduce harm. Its proposed responses include participatory design and community governance. This framing matters because the purpose of a system is itself an ethical question. A system might perform its assigned task accurately while helping an institution continue a harmful practice. Assessing it therefore requires scrutiny of what the institution is trying to achieve and whose experience counts in judging the result. AI and systemic racism

The collection also takes seriously the difficulty of turning ethical commitments into organisational practice. Richard Benjamins and colleagues describe a platform that uses an inventory of more than 800 ethical risks to support assessment and mitigation, with human oversight. Such work could help organisations assign responsibilities and identify problems more consistently. It also prompts questions about the categories through which an organisation sees harm. Who can revise the inventory? How can people describe an injury that does not fit its categories? When should disagreement lead to changing the system’s purpose? These are questions for evaluating the approach; the abstract alone cannot establish how fully the chapter answers them. Responsible AI in practice

A related chapter by J. Krijger and T. Thuis examines organisational AI ethics maturity through a pilot involving 11 Dutch higher education institutions. This provides a potentially useful connection between ethical ambitions and institutional capabilities. A question for readers is what evidence would show that an institution’s growing capacity has improved the experience of those affected. Procedures, training and responsibility structures can support change. Their value ultimately needs to be assessed through what they enable people to understand, challenge and correct. Ethics maturity in higher education

Technical architecture receives attention too. Medsker and Sumit Virmani argue for hybrid intelligent systems that combine different computational approaches, including neural and rule-based components. They associate this architecture with greater traceability, contextual reasoning and opportunities to incorporate ethical constraints. For readers developing AI applications, this is a useful invitation to examine the arrangement of the whole system. The ethical significance of any architecture nevertheless requires testing in context. A rule can be visible and still unfair. An explanation can be available yet difficult for the person affected to challenge. Better architecture can support ethical scrutiny, while leaving substantial social and institutional work to be done. Hybrid intelligent systems

The handbook’s treatment of future AI raises another question about framing. Kentaro Toyama’s AGI chapter uses scenarios ranging from assistants and companions to soldiers and infrastructure. These can help readers explore unfamiliar responsibilities. Its abstract, however, introduces human-level artificial general intelligence in terms of when it arrives. That wording deserves scrutiny. The meaning of human-level intelligence, the conditions under which it might emerge and the relationship between capability and moral standing all require argument. Ethical preparation is possible while treating such futures as conditional. AGI and ethics

For EdgeLab, these discussions invite an autopoietic ecology perspective. Here, the starting question concerns how patterns of activity keep being reproduced through their relationships with other activities. Applied to AI ethics, this means following what happens across the model, interface, user, organisation and wider conditions that sustain their interaction. An ethical policy becomes consequential through repeated decisions: which outputs are accepted, which complaints receive attention, which targets shape behaviour and which repairs receive resources.

Return to the university feedback example. Suppose teachers begin adopting the system’s preferred vocabulary, students learn to imitate it, and assessment criteria gradually follow the same pattern. Even if individual comments remain accurate and polite, the relationship between teaching, writing and judgement may have changed. An ecological inquiry would investigate how that pattern develops, what supports it and whether participants can redirect it. It would also look for benefits: perhaps the system helps students articulate ideas they previously struggled to express, and teachers develop new ways to recognise them. Neither outcome should be assumed in advance.

This perspective also connects ethics with meaning mediation: the ways AI participates in shaping how people understand a situation. Consider how an uncertain suggestion becomes authoritative feedback, or how a provisional score becomes an institutional decision. The ethical consequences depend partly on these changes in status. Useful safeguards would therefore include ways to preserve uncertainty, identify who has accepted responsibility and allow decisions to be reopened when experience challenges them.

An ecological account must also explain its own ethical commitments. The fact that a practice persists does not make it desirable. Harmful arrangements can be highly durable. At EdgeLab, a defensible direction is to examine whether arrangements protect people’s capacity to question, participate and obtain repair, with particular attention to those who bear costs while having little influence over decisions. These are normative commitments that need justification and public discussion; they cannot simply be deduced from a theory of persistence.

Medsker’s handbook looks especially useful as a resource for bringing these questions into teaching and organisational discussion. Its critical and practical contributions create opportunities to examine the relationship between stated values and working arrangements. A productive use would be to read the critical literacy and governance chapters alongside one actual institutional decision: an AI assessment system, a recruitment process or a research tool. Ask who can question it, what happens when they do, and what evidence would warrant changing course. That is where an ethical commitment begins to acquire practical force.