Advances in artificial intelligence (AI) have been met by rigorous analyses concerning their social and environmental impacts. However, AI ethics literature has curiously expended far less energy theorizing justice and interrogating its ontological boundaries. Meanwhile, environmental philosophy has taken justice seriously, although it has tended to focus exclusively on animals and nature. This chapter seeks to overcome the anthropocentric bias of AI ethics and the biocentric bias of environmental philosophy by developing an Anthropocene-informed theory of justice capable of accommodating technological beings. To accomplish this, I compare several theories of justice hospitable to non-humans—multispecies justice, planetary justice, and socio-ecological justice. After identifying optimal features drawn from these approaches, I devise a list of design principles and explain how they could be employed in the case of a personal delivery device. I conclude by calling for AI ethics to embrace complexity, interdisciplinarity, and epistemic humility in its pursuit of justice.
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
Artificial Intelligence, Design, and More-than-Human Justice
1. AI, Design, & More-than-Human Justice
Josh Gellers, PhD, LEED Green Associate
University of North Florida
Image: MidJourney
2. 1. Argue that AI ethics needs a theory of justice that
includes the more-than-human world
2. Demonstrate how AI can be designed to foster
more-than-human justice
3. “[W]e’re missing a moral framework of
justice in artificial intelligence.”
Le Bui and Noble (2020, p. 163)
4. Only 0.4% of human-robot interaction
research mentions justice.
Ostrowski et al. (2022)
5. Justice in AI Ethics
Individual case or class of injustices related to
the use of AI
Examples: Bender et al. (2021); Birhane (2021);
Sloan & Warner (2020)
Attempts to comprehensively assess AI impacts
across theories of justice or explore the relevance
of justice theory to AI
Examples: Coeckelbergh (2022); Gabriel (2022)
6. A Framework for Justice
1) WHO are the subjects?
2) WHAT is considered a “just” outcome?
3) HOW will justice be achieved?
(Biermann and Kalfagianni 2020; Fraser 2008; Pope et al 2021)
Image: MidJourney
7. Determining the “Who” of Justice
● “principle of humanism”
(Hubbard, 2011)
● “all-affected principle”
(Pogge, 1999)
● “all-subjected principle”
(Fraser, 2008)
AI Ethics
Image: MidJ
8. AI ethics has all but
ignored nonhumans.
Exceptions: Coghlan & Parker (2023); Daley
(2021); Owe & Baum (2021); Singer & Tse (2022)
Image: MidJourney
9. Premises Conclusions
Robots are increasingly found in social spaces and the
environment (Donhauser 2019).
Human/nature and nature/culture binaries are no longer tenable
in the Anthropocene (Biermann 2020; Hoły-Łuczaj and Blok
2022).
Technology is part of and constitutes the environment (Hui 2020).
Human-centered design is ethically and environmentally
problematic (Chan 2018; Forlano 2017).
We need a theory of justice that
accommodates technological entities and
the ecological systems in which they are
embedded.
Robots should be designed to contribute to
the pursuit of more-than-human justice.
10. Comparing Approaches to More-than-Human Justice
Who What How
Multispecies Justice Individual entities and whole
species or ecosystems and
relationships between them
Inclusive, inter-acting,
functioning, and flourishing
environments
Communicative modalities, humans as
diplomats, (re)designing institutions
Planetary Justice Humans OR humans, non-human
animals, and nature
Securing the integrity of the
planetary system for humans
(esp. poor) OR humans, future
generations, and more-than-
human world
Pro-poor policies, Indigenous
innovations, strengthening people’s basic
capabilities OR political rights for the
more-than-human world, incorporating
future generations into decision-making
Socio-ecological
Justice
Humans and non-humans (all-
subjected principle)
Preserving and promoting basic
capabilities for full functioning
Bidirectional communicative process,
fair procedures, a representative
structure, and empowerment
11. Q: Can AI be a subject of more-than-human justice?
A: Yes, if it plays a role in
ecosystems, makes up part of the
Earth system, and/or is subject to
governance structures.
Image: MidJourney
13. Design Principles for More-than-Human Justice
Avoid harm to the integrity, stability, and dignity of entities and systems
Consider context (social, environmental, technical, cultural) where deployed
Design to augment human and non-human capabilities to achieve flourishing
Evaluate how affordances/disaffordances affect different parties
Use participatory design to include diverse users and stakeholders
Utilize intersectional benchmarks during testing