The many twisting moralities of AI — what algorithmic AI could be: a Confucian and animist approach to AI ethics

I think therefore I am is a general refrain for conceptualising what a fully fleshed and possibly sapient AI could be. This Descartian view of human consciousness serves us well on a general level, as it applies a notion that to be a thinking being is to have an id, possibly a soul. We root our concepts of AI in a Christian/Hellenic philosophical tradition, built on Plato, Aristotle, and all who came after them both to uphold and dismantle their ideas. Most AI ethics are conceived in the crucible of western thought, as this is how students are educated and trained to think. But what if there were different ways of conceiving AI, and what if they provide better models for conceptualising AIs development from rote algorithm to potential sentience?

I must preface this as a general think piece, as I am not overly familiar with the cultures I am about to discuss. The spirit of this piece it to provoke conversation and engagement with these ideas in the hope that students and academics grounded in these traditions can add their voices and expertise. I am working through this thought process, and this article is hopefully the first of many that will explore this issue deeper.

I am interested in this because I am grappling with the notion of AI within society in a pluralistic sense, not just a singular AI coded by one organisation. From basic algorithms that bear some semblance to artificial intelligence to specific code that is pushing the boundaries, we are used to framing our concepts of AI from a western cultural and philosophical perspective, especially in terms of academia and Hollywood. AI invariably conjures notions of Skynet, cyberpunk, and dire warnings of futures yet to come. However, by framing it purely within these ethical and moral boundaries, are we potentially missing an opportunity to have a deeper cultural conversation and examination about our base assumptions?

I want to explore two ideas in this article, namely Confucian and animist concepts that could lead to a different perspective and conceptualisation of how AI is seen. This takes both a deeper anthropological perspective and examines why culturally we are so hung up on a western perspective. I believe this matters, as while we frame AI within a capitalistic and Christian/Hellenic worldview, essentially we treat AI both as a tool and a potential threat to our existence. This is not to say that all scientists and people working on AI ascribe to a rigid view of this philosophy, but our maps of the world have been so profoundly shaped by these ideas that it potentially could be causing us to misalign what AI is with what we actually wish and want AI to be. Ever since Alan Turing set down his base test nearly 70 years ago, there has been an ongoing dialogue with respect to sapience, intelligence, and machine learning, and possibly now is the time to reframe that discussion by drawing on wider cultural ideas. Neither of the examples I am using are fully fleshed or perfect, but represent different approaches that could be explored.

Confucian thought with respect to AI could primarily centre around two concepts: filial piety and the benevolent ruler. The Analects is a wide ranging text that carries a vast array of ideas and concepts, but these two are the most striking with respect to possible sapient AI.

The notion of a filial child that is respectful to its parent, and then looks after its parent in their old age is rooted within Chinese culture. Indeed, it is one of the cornerstones of most Asian cultures, one which industrialised western societies have been let slip since the neo-liberal revolution of the early 1980s. While it is easy to say that Hellenic and enlightenment philosophers would look at the root value of both filial obligation and the utilitarian value of a child looking after their parent, Confucian ideas root this deeper, seeing it as a core tenet upon which society is build. It is both obligation and expectation, an endless loop that families run through from cradle to grave to shrine. It almost harks back to Asimov’s three laws of robotics, especially the second one which expect all human orders to be followed, and it cannot allow a human to come to harm. There is a deep conversation amongst AI ethicists exploring this idea, so potentially by drawing on a more Confucian moral perspective it could allow for deeper cultural roots to be placed.

Flowing from filial obligation is the notion of the benevolent ruler with the mandate from heaven. This idea addresses the complex we have with AI endangering us all, which at the same time asking what sort of AI we wish placed over us. It is reductive to expect that AI are not already intrinsically affecting our daily lives, something which Cathy O’Neil dissects in Weapons of Math Destruction. She highlights that mortgages, credit cards, insurance, police, justice, schools, governments, and much more already have algorithms that draw on flawed databases intersecting and directing our lives without us realising. That we are rebelling against the iron grasp of these flawed processes, be it A-Levels being overturned or immigration systems being shut off, culturally we are starting to tackle the issue. However, what we have not really addressed is essentially what will AI’s mandate from heaven be, and what benevolence will actually look like. Yes, western ethics can address what the good life should be, and we can draw on Aristotle, Rawls, Dworkin, and Sandel for guidance. But, unless we want to utilise a Hobbsian model, could not a Confucian idea of the benevolent ruler with a mandate from heaven provide a critical framework within which to explore these ideas?

The second cultural idea that could be explored with be an animist approach. Many African, Asian, and South American cultures practice forms of animism, with Japanese Shinto already exploring the boundaries of what AI means within their society. This is a very complex set of philosophies, as each culture has their own approach and ideas about the spirit world. There is no one set pattern, aside from that spirits abide with and around us. It borders on theological and spiritual to reframe AI within an animist world view, but potentially it could help better make sense of the intersections of many overlapping AIs within our lives. Do we have the vocabulary to succinctly understand the intangible, yet functional, within our lives? The Christian/Hellenic lexicon is replete with the notion of singular gods and individual, we are used to thinking of our relationship with us and the spiritual as deeply personal with one deity (I am being highly reductive in this respect as Judeo-Christian faith is vastly more complex than this). Animist philosophy could potentially help us find the words and understanding to better contextualise many AI in our lives, framing them as more than simple machines.

There is the issue of drawing on both these schools of thought in a post-colonial world. If this conversation is to have any meaning it must be had between scholars across all cultures, not just a top-down examination from scholars within the Christian-Hellenic tradition. It must be a partnership, with deep rooted desires to explore and find new ways of conceiving AI ethics and morality. Simply leaving it to a Hobbsian, Locke, Mills, Aristotle et al could potentially alienate or miss out on concepts that could revolutionise the way we think about our machines.

The Christian-Hellenic philosophical tradition is built on rationality, the scientific method, and boundless inquiry. I am not suggesting we jettison this methodology in favour of other approaches; rather, I am suggesting that we work with scholars and practitioners from outside this tradition to broaden our understanding of AI and the impact AI will have on our culture. Along the way this dialogue could equip us with new language and concepts to better help us frame ideas, while also drawing those cultures into AI research in new and meaningful ways.