Abstract
In contemporary Christian ethics, Noreen Herzfeld is a pioneer when it comes to a theologically rich appraisal of artificial intelligence. After more than twenty years from her first and seminal work on AI, In His Image: Artificial Intelligence and the Human Spirit (Fortress, 2000), her latest offering, The Artifice of Intelligence, confirms Herzfeld as a direction-setting voice in a field that is shifting at breakneck speed, surrounded by hype, and outpacing ethical deliberation.
Her most intriguing suggestion comes early in the book and sets the stage for her entire inquiry: What if the endgame of the AI project, the deepest drive behind the frenetic research and gargantuan investment is not control and domination but the creation of an Other to whom we might relate? Framed this way, AI is from the start morally and mythologically charged in interesting ways.
Herzfeld's analysis pivots on two axes: human to human relationships, as mediated by AI systems, and human to AI relationships, respectively. Regarding the latter, Herzfeld asks a simple but clarifying question that organises her material: should we treat AI as a tool, as a partner, or as our ‘mind children’? Spoiler alert: she ultimately comes down on tool and vigorously waves the red flag around the prospect of AI as a surrogate human or as a ‘new digital species’, as Microsoft's Mustafa Suleyman would put it (see ‘What is an AI anyway?’, https://www.ted.com/talks/mustafa_suleyman_what_is_an_ai_anyway/transcript).
The starting point for Herzfeld’s appraisal of AI is Karl Barth's relational conception of the imago Dei and his criteria for authentic human relationships. These provide the scaffolding for the appraisals of various AI systems and use cases that are referenced throughout the book.
First, authentic relationships, she writes, requires looking the other in the eye—the acknowledgement of an Other capable of truthful self-disclosure; sooner or later, this calls for physical presence. Second, an authentic encounter is one where the partners ‘speak to and hear one another’ (p. 18), which points to the importance of verbal communication predicated on, and giving rise to mutual understanding. The third criteria Barth puts forward is that partners ‘render mutual assistance’ (p. 19). This entails both agency and interdependence. Finally, an authentic relationship presupposes that such mutual assistance is given gladly and therefore freely.
Having established her framework for authentic relationships, Herzfeld turns her attention, in chapter 2, to a fundamental question about human existence in the digital age: ‘Do we need bodies?’ Herzfeld mounts a bracing critique of transhumanist ambitions of mind uploading and ‘virtual immortality’, drawing on the typology of ghosts (souls without a body) and zombies (bodies without a soul) from popular mythology and folklore. She shows the scientific and philosophical incoherence of this vision while making a persuasive case for the body as essential to personhood. But ghosts and zombies are not just an exotic device for critiquing futuristic projects, but apocalyptic categories for existing ways of living and patterns of relating in the age of smartphones and social media. Aren't we all, in some sense, ghosts in the social media machine or zombies when we live absorbed by our devices and indifferent to our bodies?
While Siri, ‘smart’ devices, and the latest generation of Large Language Models (LLMs) like ChatGPT allow us to communicate fluently with AI, such communication does not an authentic relationship make. This is the subject of chapter 3—‘Do you hear me, Alexa?’ Speaking and listening in the context of reciprocal relationships involves more than a transfer of information. They express and enact commitments and conscious acts of self-disclosure. No AI system is capable of such morally complex actions. But this does not rule out the possibility that we can be fooled into thinking there is indeed a conscious Other on the receiving end of our emotionally laden queries and disclosures. Indeed, as the ability of AI systems to simulate human features and capabilities increases, so does our propensity to deceive ourselves (about their nature) and our susceptibility to deception by commercial or political actors.
Drawing on this framework, the chapter is a quilt of short discussions of brain computer interfaces, self-driving cars, smart devices, the privacy-infringing flow of data in the digital economy (a.k.a. surveillance capitalism), as well as various negative impacts of an increasingly technological mediation of relationships and decision making. Herzfeld's main worry is not that machines are becoming increasingly human but that humans are becoming increasingly machine-like.
Continuing with Barth's criteria for authentic relationships, chapter 4 indexes the various ways that AI systems come to our aid: in warehouses, medical laboratories, court houses, and various white-collar work spaces and activities. They improve accuracy, efficiency, productivity and other similar metrics—except, of course, when they don't, and instead entrench harmful stereotypes, biases and discriminatory actions.
Furthermore, to be more than a tool that augments human abilities, AIs require a degree of autonomy and self-awareness, something they fundamentally lack. Indeed, AIs are not autonomous moral agents, nor should they be, argues Herzfeld. This issue takes on a special urgency in the context of Lethal Autonomous Weapons developed and deployed by militaries across the world, which take autonomy to new and worrisome levels. Chapter 4 ends by laying out two different directions of living with AI: either AI is allowed to automate and replace humans or AI is used to augment human capacities. Human dignity requires the latter, even though the incentive structure of late capitalism clearly skews towards the former. ‘Human dignity depends on our working with our tools rather than letting them supplant us, and this is at its most important in matters that involve questions of life and death’ (p. 96), Herzfeld concludes.
Chapter 5 touches on Barth's final criteria for authentic relationships: that mutual aid be given gladly and therefore freely, with authentic emotion, not merely the empty performance of it. Herzfeld rejects the possibility that AIs will be capable of love in the absence of a biological body. The rest of the chapter is made up of summary discussions of various emotion-detection and emotion-simulation systems that are embedded in social media and social robotics, respectively. She is clear-sighted about the dangers of emotional manipulation, deception and self-deception, which are arguably growing with the advent of human–AI relationships in the form of AI assistants, agents, and humanoid robots.
Chapter 6, on ‘The Dreams of Reason’, deals with the varieties of transhumanism: the life-extensionist end of the spectrum, of AI, nanotech and biotech which are plausibly poised to improve health-span and lifespans alike, as well as the post-human end, which envisions the superseding of organic life altogether, through uploading or the creation of a new, putatively superior cyborg entity that would live indefinitely. Herzfeld is open to biotech therapeutics in the service of the Dominical call to heal the sick, but is conspicuously silent on enhancement. She draws clarifying contrasts between transhuman and orthodox Christian eschatologies, pointing to the qualitative difference between digital, human-made ‘immortality’ and the eternal life that Jesus’ resurrection epitomises and enables.
Does AI pose an existential threat to humanity? Unlikely, according to Herzfeld. But just because the scenario of a super intelligent AGI (or ASI) that goes rogue is implausible does not make current AI systems less existentially salient. Indeed, AI is disrupting the conditions and experience of life for a growing number of people, through job displacement, robots used in the care professions, algorithmically-driven social media, the military, and much more. Chapter 6 ends with a clarion call to resist personification of AI and eschew the ‘death-defying religions’ (p. 153) of post- and trans-humanism, as Jaron Lanier calls them.
In her dénouement, ‘An Embodied Faith’, Herzfeld anchors her argument in the premise that Western culture's alienation from, and quiet contempt for the body—often cloaked in the rhetoric of care—is fertile ground for neo-gnostic visions that give pride of place to the rational mind. But ‘a Christian theology centred in our relationships—with God, neighbour, and self—is necessarily an embodied theology’ (p. 160), Herzfeld reminds readers. Barth's criteria for authentic relationships are only met through embodiment. She marshals linguistic and ecological arguments to defend the necessity of the body. Her targets are, again, transhumanist visions of superseding organic life, which fail to appreciate that the mind and body are intimately intertwined. But theologically, it is the Incarnation that is the strongest, even definitive affirmation of the value of materiality and embodied life. Through the Incarnation, ‘God enters and sanctifies the dynamic web of interrelatedness of body, mind, and environment’ (p. 171), says Herzfeld in a beautiful turn of phrase.
If the beginning of the book is somewhat ambivalent and tentative in its judgements, to allow for the complexity surrounding AI to come into full view, towards the end of the book, Herzfeld is unequivocal in her pronouncements: AIs are ‘machines, not living things’; ‘tools and nothing more’ (p. 174). As such, people and their need for loving care must always come before machines. Even as they become increasingly sophisticated, and may even become conscious (Herzfeld tentatively leaves open that possibility), machines should always be treated as an Other—neither human, let alone a god—precisely because humans and AI will never share the same organic body. Lack of shared embodiment is a chasm we ignore at our own peril. The bitter irony is that we habitually treat AI as a pagan god when we trust AI systems unquestioningly and give them unwarranted authority. AI priests might be a fringe phenomenon (although this may change) but AI worship or technolatry is arguably the more widespread, if subtler danger in our technologically advanced late modern world.
Herzfeld's journey through the ethical landscape of AI ultimately leads her back to her initial question: What is the true purpose of AI? Herzfeld lands on an unequivocal note: ‘AI can be a good tool when used with care. It is an incomplete partner and a terrible surrogate for other humans’ (p. 179). Future relationships with AIs may be ‘safe and made to measure’ (p. 178) but they will not satisfy the depth of our relational longings. If anything, they may have a deformative effect, eroding the skills and virtues necessary to sustain our authentic relationships that ‘stretch our boundaries, … test our preconceived notions, … draw us out of our petty preoccupations, and make us grow in the image and likeness of God’ (p. 178).
All in all, Herzfeld's examples and cultural references will inevitably become obsolete as the technological landscape continues to shift, but the moral clarity, robust theological framework, and her overall example of nuanced theological critique will remain an evergreen resource for anyone committed to thinking straight about AI in the years to come.