Lecture 6-2: Artificial Intelligence and Technology Ethics
Contents 31
- Lecture 1-1: What Makes Ethics Distinctly Christian
- Lecture 1-2: Scripture, Natural Revelation, and Conscience — Frame's Triperspectival Method
- Lecture 1-3: Secular Ethical Theories — Utilitarianism, Deontology, and Virtue Ethics
- Lecture 1-4: Natural Law and General Revelation in Ethics
- Lecture 1-5: The Structure of Christian Moral Reasoning in Practice
- Lecture 2-1: The Structure of Old Testament Law
- Lecture 2-2: The Sermon on the Mount as Ethical Vision
- Lecture 2-3: Paul and the Indicative-Imperative Pattern
- Lecture 2-4: The Unity of the Whole Canon's Ethics
- Lecture 2-5: Reading the Bible's Most Difficult Ethical Texts
- Lecture 3-1: Christian Freedom and the Law
- Lecture 3-2: Vocation and Work
- Lecture 3-3: Marriage, Sexuality, and the Body
- Lecture 3-4: Truth-Telling, Speech, and Digital Communication
- Lecture 3-5: Money, Possessions, and Contentment
- Lecture 4-1: Patristic and Medieval Ethics
- Lecture 4-2: Reformation Ethics — Law, Gospel, and the Third Use
- Lecture 4-3: Modern Challenges — Weber's Protestant Ethic Thesis
- Lecture 4-4: Twentieth-Century Christian Ethics — Bonhoeffer and the Limits of Otherwise-Clear Commands
- Lecture 4-5: Niebuhr's Christ and Culture and the Church-World Relationship
- Lecture 5-1: Political Engagement and Civic Responsibility
- Lecture 5-2: Economic Justice and Poverty
- Lecture 5-3: War, Peace, and Just War Theory
- Lecture 5-4: Race, Immigration, and Reconciliation
- Lecture 5-5: Environmental Ethics and Creation Care
This course's companion sequence in eschatology (ST8) already asked, from a different angle, whether artificial intelligence's growing cognitive capability changes the doctrine that humanity alone bears the image of God, and answered no — because Scripture grounds that doctrine in humanity's relationship to God (being addressed, commissioned, morally accountable, capable of redemption), not in comparative cognitive performance at any particular task. This lecture does not repeat that argument; it assumes the conclusion is settled and asks a different, ethical question that only becomes pressing once the doctrinal question is out of the way: given that an AI system does not and cannot bear the image of God in the sense this course has argued for, what do Christians owe one another, and what should they be wary of, in the actual, everyday use of increasingly capable automated systems?
That question breaks into two more specific ones this lecture takes up in turn. The first is a question of moral responsibility: when an automated system — a hiring algorithm, a content-recommendation engine, a medical-diagnosis tool, a self-driving vehicle — produces a harmful outcome, who is morally answerable for it, given that the system itself is not a moral agent capable of bearing responsibility? The second is a question of posture: how should Christians hold together a fair-minded appreciation of what artificial intelligence can genuinely do for human flourishing with an equally fair-minded wariness about what it cannot do and should not be asked to do? John Lennox's 2084: Artificial Intelligence and the Future of Humanity argues that a Christian anthropology supplies real resources for holding those two postures together, resisting both an uncritical technological utopianism that expects AI to solve fundamentally human problems and a reflexive technophobia that treats every advance as inherently threatening. This lecture engages Lennox's argument, and closes by returning briefly to 1 Corinthians 15:53-54's picture of embodied resurrection hope as the standing check against any technological aspiration — AI-driven or otherwise — that treats the body as an obstacle to be engineered around rather than the very thing God intends to redeem.
What these 19 minutes give you
- State why this lecture treats "does AI threaten the doctrine of the image of God" as a settled question inherited from ST8's treatment, and explain what different, ethical question this lecture takes up instead.
- Explain the problem of moral responsibility for automated systems, and state at least two candidate loci of responsibility (designer, deployer, user) along with a reason the "black box" character of some machine-learning systems complicates assigning it.
- State John Lennox's argument in *2084* that Christian anthropology resists both technological utopianism and technophobia, and explain what each of those two errors looks like in practice.
- Explain how 1 Corinthians 15:53-54's picture of embodied resurrection functions as a check on disembodied technological aspiration, and connect this briefly to 3-4's treatment of digital communication as an older sin category in a newer medium.
For this perishable body must become imperishable, and this mortal must put on immortality. But when this perishable body will have become imperishable, and this mortal will have put on immortality, then what is written will happen: 'Death is swallowed up in victory.'" — 1 Corinthians 15:53-54
World English Bible
1. A settled doctrine, a new ethical question
Genesis 1:27's claim that humanity bears the , argued in Lecture 6-1 to ground human dignity independently of capacity, has a further implication this lecture depends on without re-arguing: since the image of God is grounded in relationship to God rather than in comparative cognitive performance, no degree of sophistication an artificial system reaches in generating language, producing images, or optimizing a decision changes the doctrine at all. An AI system that writes more fluently than most humans, diagnoses disease more accurately than most physicians, or plays a strategy game better than any human champion has not thereby entered relationship with God, moral accountability before God, or the possibility of being redeemed — the actual grounds of the doctrine, on this course's own account. This lecture treats that conclusion as inherited rather than needing to be re-established, and turns instead to what actually changes ethically as these systems become more capable and more widely deployed: not what an AI system is, but what human beings owe each other in building, deploying, and relying on it.
2. Who is responsible when a system causes harm
An automated system is not a moral agent; it has no intentions in the sense moral theology requires for praise or blame, and it cannot repent, be held accountable in any personal sense, or bear guilt. Yet automated systems now make or heavily influence decisions with real moral weight — approving or denying a loan application, flagging a defendant as high risk for a judge's sentencing decision, recommending content to a user for hours at a stretch, controlling a vehicle in traffic — and when those decisions cause harm, the harm is real even though the system causing it is not a moral agent. Responsibility does not evaporate simply because a machine, rather than a person, executed the decision; it has to be located in the human beings whose choices made the system's operation possible.
Moral theology's traditional analysis of an act by intention, causation, and foreseeability, already used in Lecture 6-1's treatment of double effect, gives a framework for locating that responsibility, even though it does not resolve every case easily. A designer who builds a system to optimize a stated goal (engagement, approval rate, predictive accuracy) bears responsibility for foreseeable harms that goal produces, whether or not each specific harmful outcome was individually intended — an engagement-maximizing recommendation algorithm that reliably promotes outrage-inducing content because outrage reliably holds attention is not absolved of responsibility for that pattern simply because no single video was personally selected by a human being to provoke anger. A deployer — the company or institution that puts a system into use for a specific purpose, such as a bank using a lending algorithm or a court system using a risk-assessment tool — bears responsibility for choosing to rely on the system's outputs, for the adequacy of its testing and oversight, and for building in genuine human review rather than treating an algorithmic output as beyond question. A user who operates a system within its intended bounds bears comparatively little responsibility for a design flaw they had no way to detect; a user who deliberately misuses a system, or ignores clear warning signs a reasonable person would heed, bears more.
The theologian Noreen Herzfeld's In Our Image: Artificial Intelligence and the Human Spirit names a genuine complication this analysis has to reckon with honestly: many contemporary machine-learning systems, particularly large neural networks, are "black boxes" in a specific sense — their designers can describe the training process and the objective function optimized, but often cannot fully explain, in terms a human being would recognize as a reason, why the system produced a particular output in a particular case. This opacity does not dissolve responsibility, but it does complicate it: a designer who cannot fully predict or explain a system's behavior in every case has correspondingly greater responsibility to test rigorously, to build in safeguards for the failure modes that can be anticipated even if individual outputs cannot, and to resist deploying a system in a high-stakes context — sentencing, medical diagnosis, hiring — before that testing gives real confidence in its behavior, rather than treating impressive average performance as sufficient on its own.
3. Lennox: resisting both utopianism and technophobia
John Lennox's 2084: Artificial Intelligence and the Future of Humanity (2020) argues that Christian anthropology gives believers real intellectual resources for a posture toward AI that is neither credulous nor fearful, and this lecture engages that argument as a needed corrective to two opposite errors it sees among Christians and non-Christians alike.
The first error, technological utopianism, treats sufficiently advanced artificial intelligence as capable of resolving problems that are not, at bottom, technical problems at all — meaning, purpose, moral formation, reconciliation between estranged people, the human confrontation with mortality. Stated at its most ambitious, this utopianism expects AI-driven abundance or AI-augmented cognition to deliver something like the fullness Scripture locates in redemption and the new creation; stated more modestly, it simply overestimates how much of human flourishing a purely technical improvement in capability can actually supply. Lennox's argument, consistent with this course's own doctrine of sin (argued at length in Chapter 1), is that a great deal of what actually damages human flourishing — pride, cruelty, greed, the will to dominate others — is a problem of the human heart that more capable tools do not fix and can in fact amplify, since a more capable tool in a sinful hand produces more capable sin, not less.
The second error, technophobia, treats AI's growing capability as an inherent threat to human significance simply because it approaches or exceeds human performance at particular cognitive tasks. This error, Lennox argues, typically smuggles in the very materialist assumption Christian anthropology rejects: that human worth is a function of comparative cognitive performance in the first place, so that a machine outperforming a human at reasoning or language production is treated as diminishing something essential about being human. But if 6-1's argument and ST8's parallel argument are right that the image of God was never grounded in comparative cognitive performance, then a machine's growing capability at those tasks simply is not the kind of thing that could threaten human significance — it can threaten particular jobs, particular economic arrangements, and particular forms of social status built around cognitive performance, which are real and serious concerns, but not the doctrine of human dignity itself, which never rested on that foundation. Lennox's stated contribution is to give Christians a reasoned basis for engaging AI's genuine benefits (in medicine, scientific research, and practical assistance) without needing either posture — expecting salvation from the technology, or fearing annihilation by it.
4. Embodiment as the standing check
The chapter returns, briefly, to a text this course's companion eschatology course engaged at length: 1 Corinthians 15:53-54's picture of "this perishable body" itself becoming imperishable, the same body transformed by God's own act rather than a pattern of information preserved independent of it. That argument was made there chiefly against transhumanism's promise of digital immortality; this lecture draws a narrower, more immediate application specific to present-day technology ethics rather than a speculative future one. An AI companion application designed to simulate warmth, attentiveness, and personal interest; a curated social feed engineered to substitute algorithmically selected content for a person's own embodied relationships and unmediated attention — both trade, in smaller and more everyday ways, on the same appeal transhumanism trades on at its most ambitious: the promise of relational or experiential goods without the cost, risk, and mutual vulnerability that a real, embodied relationship with another image-bearer actually requires. A Christian anthropology committed, on the ground of 1 Corinthians 15:53-54, to the goodness and permanence of embodiment has principled reason to be wary of technologies that substitute a simulated version of embodied goods for the real thing, not because such technologies have no legitimate use, but because their appeal often depends on exactly the disembodied shortcut this course's doctrine of the resurrection refuses to grant any priority over the real, costly, embodied relationships God actually made human beings for.
This connects directly to Lecture 3-4's treatment of truth-telling and digital communication: that lecture argued social media's distinctive dangers (outrage amplification, decontextualized quotation, anonymity's effect on ordinary restraint) are old sins — gossip, slander, false witness — operating in a faster medium, not a wholly new ethical category. The same pattern holds here: an AI companion or an engagement-optimized feed is not introducing a wholly new sin either; it is offering an old temptation — preferring an easier, less costly substitute to the harder, more costly good God actually intends — dressed in unusually capable and unusually persuasive new clothing.
5. Going to the Sources
Read the Rome Call for AI Ethics (2020), a joint declaration originally signed by the Pontifical Academy for Life together with Microsoft, IBM, the U.N. Food and Agriculture Organization, and the Italian government, proposing what it calls "algorethics" — six principles (transparency, inclusion, responsibility, impartiality, reliability, security and privacy) meant to govern AI's design and deployment. It matters here because it is a genuinely multi-stakeholder document naming the same responsibility question this lecture's second section addressed — who answers for an automated system's effects — and locating that responsibility explicitly in human design, deployment, and oversight decisions rather than in the system itself, from an ecclesial and secular partnership rather than a single tradition's confessional statement.
In short
- The lecture takes as settled, from ST8, that AI does not threaten the image of God, since that doctrine rests on relationship to God and not on cognitive performance. It asks instead what people owe each other when using AI.
- A system is not a moral agent, so responsibility lies with people: the designer, the deployer, and the user. Each carries a different share.
- "Black box" systems, whose outputs even designers cannot fully explain (Noreen Herzfeld), raise the duty to test carefully before use in high-stakes settings.
- John Lennox warns against two errors. Utopianism expects AI to solve problems of the heart. Technophobia assumes human worth depends on comparative performance.
- 1 Corinthians 15:53-54 supports caution toward technologies, such as AI companions, that offer simulated embodied relationships without their cost.
Topic: This lecture asks who is morally responsible when automated systems cause harm, and how Christians should hold a balanced attitude toward AI.
Remember: Responsibility for a machine's harm never disappears. It rests with the people who design, deploy, and use it.
Caution: The lecture is not against AI. It names real benefits, and warns only against expecting salvation from it or fearing it as a threat to human dignity.
One line to carry into ministry: Use AI's real benefits with human oversight, and do not let a simulated relationship replace the costly, embodied ones God made us for.
Check yourself
Not "did you read it" but "can you explain it". Ticks are kept on this device only.
Reflection and discussion
- This lecture distinguishes designer, deployer, and user responsibility for an automated system's harm. Choose one real or plausible example (a hiring algorithm, a content-recommendation system, a medical-diagnosis tool) and explain how you would distribute responsibility among these three roles if the system caused a specific, identifiable harm.
- Lennox argues technophobia often smuggles in the assumption that human worth is a function of comparative cognitive performance. Do you find this diagnosis persuasive as an account of why some people fear AI, or does it miss a form of legitimate concern about AI that is not really about comparative performance at all?
- This lecture connects AI companionship applications to 1 Corinthians 15:53-54's embodied resurrection hope by way of an argument about substituting simulated relational goods for costly embodied ones. Can you think of an application of AI that offers real relational or social benefit without raising this same concern? What distinguishes it from the cases this lecture is wary of?