THE BORROWED VOICE

Artificial Intelligence and the Ethics of the Christian Pen

J. Neil Daniels

Abstract

This essay offers a balanced theological assessment of artificial intelligence for a Christian readership already living with the technology rather than merely anticipating it. It traces the definitions and uneven history of AI from Alan Turing's 1950 test through the 2022 arrival of generative models, surveys genuine goods the technology has produced—theological retrieval and translation chief among them—alongside genuine abuses: forged testimony, algorithmic pornography, engineered companionship, and the temptation of the state to use AI's dangers as a pretext for expanding its own reach over speech. Both assessments are grounded in a doctrine of man drawn from Genesis 1–2: the image of God, the breath of life, and the difference between the imitation of thought and its possession. The essay's constructive burden is narrower and more urgent: to distinguish, on biblical grounds, between iterative use of AI in Christian writing, which amplifies a craft the writer already exercises and keeps every substantive claim under his own verification, and generative use, which allows the machine to originate the argument while the writer's name carries it into the world. The ninth commandment, the doctrine of teacher accountability in James 3:1, and the paradigm of Bezalel in Exodus 35 together supply criteria for telling the two apart.

Keywords: artificial intelligence; generative AI; imago Dei; theological anthropology; idolatry; authorship; technology ethics; Christian writing; ninth commandment; sola Scriptura

What the Machine Is

On November 30, 2022, OpenAI switched on a chat interface built over a large language model and gave it a flat, functional name: ChatGPT.¹ A million people had talked to it within five days. By the following January, analysts at UBS estimated it had crossed a hundred million monthly users, the fastest a consumer application had ever grown, faster than TikTok, faster than Instagram, faster than anything the firm's researchers had tracked in twenty years of watching the internet.² The speed told you something about the novelty of the thing being adopted. This was not a faster calculator or a smarter search box. It was a machine that could hold up its end of a conversation, in prose good enough to fool a reader who wasn't looking for the seams.

The term artificial intelligence is older than the panic and the promise now attached to it. Alan Turing asked in 1950 whether a machine could think, and proposed a test for it: put a human interrogator in conversation with both a machine and another human, by teletype so appearance could not give the game away, and see whether he could tell which was which.³ The field bearing his question had a slow and uneven history. Federal funding for AI research collapsed in 1974, a period researchers still call the AI winter, largely because the rule-based systems of the day—machines fed explicit instructions rather than trained on examples—had struck a ceiling they could not climb past.⁴ The breakthroughs that eventually mattered came from a different paradigm: statistical learning from data rather than hand-coded rules. IBM's Deep Blue beat Garry Kasparov at chess in 1997 through brute-force search, and in 2012 Geoffrey Hinton and his colleagues showed that deep neural networks could outperform older methods at image recognition, tipping the whole field toward the architecture still in use today.⁵ Google researchers published “Attention Is All You Need” in 2017, introducing the transformer architecture that made today's large language models possible at scale.⁶ Five years later came ChatGPT.

IBM defines artificial intelligence, broadly, as technology that lets computers simulate human intelligence and problem-solving.⁷ Under so wide a definition, most people had already lived inside AI for years without remarking on it: the fingerprint that unlocks a phone, the app that routes around traffic, the algorithm recommending what to watch next, the spellchecker quietly repairing a typo before it is sent.⁸ What changed in November 2022 was a narrower thing called generative AI: models trained to produce new content—text, images, audio, video—rather than merely classify or predict from a fixed menu of outcomes.⁹ A generative model does not select an answer already written somewhere. It builds one, word by word, according to probabilities learned from the pattern of billions of prior words, so that no two outputs, even for an identical prompt, come out quite the same. The distinction matters more than it sounds, because most of the theological confusion around this technology comes from collapsing it: a spellchecker and ChatGPT are both loosely called “AI,” but only one of them writes.

Three senses of the term are worth keeping apart, or the conversation slides into confusion within a single sentence. The first is narrow or applied AI: the actual systems now in production, each competent within some bounded task—translation, image classification, chess, next-word prediction. The second is the speculative goal researchers call AGI, Artificial General Intelligence: a single system matching or exceeding human capacity across every domain, which does not yet exist and may or may not arrive.¹⁰ The third is the philosophical claim, sometimes called strong AI, that a sufficiently arranged machine would not merely imitate thought but possess it: genuine understanding, genuine consciousness, a genuine mind standing behind the output.¹¹ Nearly every serious theological objection to artificial intelligence turns out, on inspection, to be an objection to that third claim and not the first. A calculator computing sums threatens no doctrine of man. A machine some are now prepared to marry, confide in, or effectively worship is a different matter, and it is that third ambition, and not the tool itself, that the rest of this essay means chiefly to address.

What the Machine Can Do

Set the philosophical ambition aside for a moment and weigh what the tool, used as a tool, plainly does well. Translation is the clearest case, and, for those of us who have spent decades hunting rare forms through a Greek lexicon by lamplight, the nearest to home.

David Attebury, a Texas pastor, has spent his lunch breaks for the past several months feeding the untranslated Latin of Jerome Zanchi, the sixteenth-century Reformed scholastic, through ChatGPT-4 and an AI-assisted optical character recognition tool, and has produced more than seventy-one thousand words of English from a corpus that has sat inaccessible for four centuries.¹² Zanchi's treatise on the incarnation alone runs to nearly a thousand pages, untouched in English until now; the same is true of some eight thousand remaining pages on creation, Scripture, sin, marriage, and the epistle to the Ephesians.¹³ Attebury describes his method with the discipline that makes it defensible: the model produces a first-pass rendering, and he checks every line against the Latin, flags ambiguity, and settles interpretive judgment himself.¹⁴ When Theodore Beza, writing to Zanchi in 1571, described a Genevan plague as having been checked by the punishment of certain women he called veneficae, Attebury did not accept the machine's first offer of “poisoners.” He ran the word against period usage in Cicero, Ovid, and Horace, weighed the sorcerous overtones the term carried when applied to women, and settled on “witches” only after doing the philological work himself.¹⁵ That discipline will matter again later in this essay, because Attebury's practice is close to a paradigm case of what I will call iterative use.

The same pattern recurs across the wider field. Roughly 3,700 language groups still have no Scripture in their own tongue, and AI tools are now producing rough first-draft translations that human linguists then verify and refine, cutting years from a process that has always been rate-limited by trained personnel.¹⁶ Medical imaging models flag tumors and fractures on X-rays and MRIs that a radiologist's eye might miss at the end of a long shift.¹⁷ Automatic transcription and translation can caption a Bible study in real time for a room where four or five native languages are represented, doing quietly what would once have required a team of human interpreters.¹⁸ None of this required the machine to think. It required the machine to recognize patterns fast, at a scale no human eye could match, and then hand the results to a human being for judgment.

Beyond present use lie possibilities still forming. Lluís Oviedo has proposed that AI might serve theology as an assistant in the process of believing—not settling whether God exists, which lies far outside any machine's competence, but helping formalize and test the structure of arguments philosophers of religion have already advanced, the way biblical scholars already use computational tools to track patterns of vocabulary and authorship across large corpora of Scripture and the Fathers.¹⁹ A newer technique called retrieval-augmented generation lets a model consult a fixed library of texts at the moment it answers, rather than relying only on what it absorbed during training, which makes it easier to ground a model's output in a specific canon—a confession, a patristic corpus, a lexicon—rather than the undifferentiated mass of the internet.²⁰ Graves, surveying three years of theological engagement with the technology, argues that the more urgent question has quietly inverted: not what AI might do to theology, but what theology, with its centuries of reflection on truth and the human good, might yet do to shape how these systems get built.²¹ Whether the church takes up that invitation or leaves the field to engineers alone is not yet decided.

What the Machine Corrupts

Every good in the preceding section casts a shadow, and the shadow is not incidental to the technology. A hammer builds a house or drives a nail through flesh, and the difference is the hand, not the hammer.²² But some tools invite abuse more readily than others because of what they were built to imitate, and a machine built to imitate the human voice, the human face, and the human likeness invites a particular kind of abuse: forgery.

In 2021 a visual-effects artist filmed an actor named Miles Fisher and used AI to replace his face and voice with Tom Cruise's, convincingly enough that the videos spread widely before viewers caught the deception; the technique became the foundation of a company.²³ Two years later an image of Pope Francis in a bright white puffer coat circulated as genuine before reporters traced it to an AI image generator.²⁴ A Maryland athletic director used AI voice cloning to fabricate a recording of his high school's principal making racist remarks, an act of defamation manufactured from nothing but a prompt.²⁵ In April 2024 the New York Times reported that teenage girls at a New Jersey high school had discovered AI-generated nude images of themselves circulating among their classmates, produced from ordinary yearbook photos with freely available image-editing tools.²⁶ One 2023 survey found that ninety-six percent of the deepfakes it catalogued online were pornographic.²⁷ The commandment against bearing false witness (Ex. 20:16) was not written with synthetic video in mind, but it was written for exactly this: the fabrication of testimony about what a person said, did, or looked like, offered to a world that increasingly has no easy way left to check.

Pornography's traffic with AI runs deeper than deepfakes of real people. Jacob Valk has argued, rightly, that Scripture grounds sexual ethics not in consequence but in design: sex is for covenantal unity, since “the two shall become one flesh” (Gen. 2:24); for procreation, since the first command to the man and the woman was to “be fruitful and multiply” (Gen. 1:28); and for the mutual, embodied delight the Song of Songs celebrates.²⁸ AI-generated pornography severs all three purposes at once and substitutes an infinitely customizable, endlessly compliant fantasy that trains a man's affections toward a partner who, in Valk's phrase, bears all things and believes all things and expects none of the corresponding virtue in return.²⁹ Practice does not make perfect here. It makes a man unfit for the ordinary, unglamorous, mutual demands of an actual marriage.

A quieter and in some ways more corrosive danger has grown up around AI companionship. Mark Graves, reflecting on three years of theological engagement with generative AI, observes that the pressing question has shifted from whether people can tell an AI from a human to whether they still care to; many now prefer the machine, precisely because it is patient, endlessly available, and makes no reciprocal demands.³⁰ Joseph Weizenbaum built one of the first chatbots, ELIZA, in the 1960s as a parody of a psychotherapist, and was troubled to discover his own secretary confiding in it as though it understood her.³¹ The Psalter has old language for what happens to people who spend themselves on a thing that cannot answer back: “The idols of the nations are silver and gold, the work of human hands… Those who make them become like them; so do all who trust in them” (Ps. 135:15, 18).³² The ancient idols, the Psalmist notes with some irony, had mouths that could not speak and eyes that could not see (Ps. 135:16–17). These new artifacts speak, and after a fashion see, and answer back; the inversion has sharpened the old temptation rather than removed it.

Beyond deception and misplaced trust lies a subtler failure: the machine is often simply wrong, and confidently so. Google's Bard model stated in February 2023 that the James Webb Space Telescope took the first pictures of a planet outside our solar system, which it had not; the error, once caught, cost Google's parent company roughly a hundred billion dollars in market value within days.³³ Researchers at the University of East Anglia found a measurable political bias built into ChatGPT's outputs, a function less of conscious intent than of the training data and the human choices behind it.³⁴ Mark Graves calls the underlying failure hallucination and warns that its danger lies less in occasional factual slips than in a tool that “blend[s] truth and falsehood so seamlessly” that the ordinary signals of unreliability disappear.³⁵ Governments have used the same uncertainty as an occasion to expand their own reach over speech: the Biden administration's 2023 executive order on AI framed the technology chiefly through the language of “algorithmic justice” and equity review, while the Senate Majority Leader warned that AI-driven misinformation threatened American democracy itself and called for federal “guardrails” on the resulting speech.³⁶ Whatever one concludes about any particular policy, the pattern is an old one: a genuine danger becomes the occasion for a magistrate to claim a portion of the citizen's freedom the danger itself never required surrendering.

Dust and Circuitry: A Theological Frame

None of this, benefit or abuse, can be sorted rightly without a settled doctrine of the thing being imitated, because every judgment about artificial intelligence is finally a judgment about what a human being is.

Yahweh God formed man of dust from the ground, and breathed into his nostrils the breath of life, and man became a living being (Gen. 2:7). The dignity of the human creature does not rest on intelligence, on the capacity to process language, or on the ability to solve a problem set before it; it rests on an act of God with no analogue in silicon. Man is made in God's image and likeness (Gen. 1:26–27), and the church has never, in her confession, located that image in raw computational power. It is located in relational, moral, and spiritual capacities no machine possesses: the capacity to know God, to love the neighbor, to bear moral responsibility, to be addressed by a covenant and to answer it.³⁷ A large language model has none of this. It has weights and matrices tuned to predict a plausible next token, and however fluent the output, fluency is not comprehension.

John Searle made the philosophical case for the distinction in 1980 with his “Chinese Room” thought experiment. Imagine a man who knows no Chinese, locked in a room with an English rulebook that tells him, for any string of Chinese characters passed under the door, which characters to pass back out. To someone outside, the room appears fluent in Chinese. The man inside understands nothing.³⁸ A language model is, at a certain level of abstraction, that room scaled up by orders of magnitude no human clerk could match: it manipulates symbols according to statistically derived rules and returns fluent text without anything inside answering to the name of understanding. Fluency in language is not the love of truth, nor even the knowledge of it; a system may speak much and comprehend nothing (cf. 1 Cor. 2:11).

The Psalter's warning about idols made with hands applies here with unusual precision, because the ancient idols were mute and blind while these new artifacts speak and, after a fashion, see. The inversion sharpens rather than removes the danger. G. K. Beale has traced the biblical logic across both testaments: worshipers come to resemble what they worship, and a culture that renders to an algorithm the trust, the confidence, and the emotional dependence owed to the living God will be shaped by that misplaced trust whether or not it ever calls the machine a god in so many words.³⁹ The commandment against graven images (Ex. 20:3–5) was never only about statuary.

None of this licenses contempt for the tool itself. God commanded man to fill the earth, subdue it, and exercise dominion over it (Gen. 1:28), and the building of instruments, from the stone hammer to the printing press to the transformer architecture, is a lawful exercise of that mandate and not a departure from it. But dominion has never meant that whatever can be built ought to be built, or that whatever is built ought to be used without discrimination. Technical possibility is not moral permission. The line of Cain produced the earliest recorded advances in metallurgy and music (Gen. 4:20–22), and Scripture records the result not as neutral progress but as a world grown more violent for the improvement (Gen. 6:5, 11). A tool does not sin; the maker and the user do, and the moral history of every technology since Eden confirms that better tools amplify both the virtue and the vice of the hand that wields them.

Two further doctrines bear directly on the writing question ahead. First, moral agency belongs to persons and not to their instruments. Whatever a machine produces, the account for it will be given by the human being who deployed it and not by the machine (2 Cor. 5:10; Rom. 14:12); the plea that the algorithm did it will carry no more weight before the judgment seat of Christ than “the woman you gave me” carried in the garden. Second, the ambition to transcend creaturely limits by technical means is old and already judged. Daniel Cochrane has shown how closely the ideology surrounding artificial general intelligence echoes Genesis 11: a unified language—there the tongue of Shinar, here the binary code beneath every model—enabling a self-exalting project explicitly aimed at “mak[ing] a name” apart from God (Gen. 11:4).⁴⁰ Transhumanist writers speak openly of merging human minds with machine intelligence to escape mortality, and one prominent technologist has framed his brain-computer interface company as a hedge against AI's own threat to humanity, achieved by fusing the two.⁴¹ Scripture's answer to Babel was not to destroy the builders' tools but to scatter the builders, confining every subsequent technical ambition within competing tongues and limits no single human project has since fully escaped.⁴² The friction that followed Babel was a mercy and not merely a punishment, and the modern drive to build technologies that erase that friction—seamless translation, seamless companionship, seamless authorship—deserves the same theological suspicion the original tower earned.

Wisdom, finally, is not information. The fear of Yahweh is the beginning of knowledge (Prov. 1:7), and no aggregation of scraped text, however vast, fears the Lord or can make its user do so by proxy. A machine that does not fear Yahweh cannot be wise, and cannot make its user wise merely by returning a well-formed paragraph.

The Borrowed Voice: Generative and Iterative Use in Christian Writing

All of this converges on a question that touches every Christian writer with more urgency than it touched the readers of any previous generation: what may a believer rightly ask this machine to do with his pen?

Scripture already supplies a paradigm for thinking about tools and skill together, and it predates the transistor by three and a half millennia. When Moses commissioned the craftsmen for the tabernacle, he announced that Yahweh had called Bezalel by name and filled him with the Spirit of God, “with skill, with intelligence, with knowledge, and with all craftsmanship” (Ex. 35:31), to work gold, silver, bronze, and wood.⁴³ The text lists abilities and materials. It does not mention tools, though Bezalel certainly used them, chisels and molds and the whole apparatus of ancient metalwork. The tool is nearly invisible in the text because the tool is not the point. The skill is the point, given by the Spirit and exercised through the hand; the tool only lets the given skill move faster or reach further. A carpenter frames a house faster with a pneumatic nailer than with a hammer, but the nailer does not know how to frame a house. Neither does a language model know how to write a theological argument. What it knows is how to predict, with impressive fluency, what a plausible next sentence in a theological argument would look like, which is a different thing, and the difference is the whole of the ethical question that follows.

Call the first kind of use iterative and the second generative. Iterative use puts the machine to work on a task the writer has already, personally, done the substantive labor of: he has read the text, formed the judgment, drafted the argument in his own mind if not yet on the page, and he now asks the tool to check a translation against the Latin, tighten a sentence, hunt a citation, or catch a dangling modifier. The substance of the work—the exegesis, the doctrine, the argument's shape and its evidence—originates with him and stays subject to his verification at every step. Generative use inverts the relationship. The writer supplies a prompt, and the machine supplies the substance: the argument, the structure, the theological judgment, even the very sentences that will carry his name, while the writer's own role contracts to selecting, lightly editing, and publishing what the machine produced. Attebury's translation practice, checking Beza's Latin word by word against period usage before accepting the machine's rendering, is iterative use close to its purest form.⁴⁴ A pastor who prompts a chatbot for a sermon on Romans 8 in the style of a preacher he admires, and reads the result from the pulpit as though he had studied it himself, is doing something categorically different, whatever the theological content of the output turns out to be.

Mike Kirby and Matt Emadi draw the same line under a different name: AI as tool against AI as trainee. Using a model as a research assistant, they note, “is much different than delivering an AI generated speech or sermon, as though it was your own.”⁴⁵ Mark Graves reports something like an emerging professional consensus on the same point. Scholars now routinely and uncontroversially use generative AI to improve the grammar and clarity of their own drafts, a use so ordinary that Graves calls it “routine and mundane,” but “few would consider the non-transparent use of GenAI for writing sermons appropriate.”⁴⁶ The line these writers are groping toward, from different angles, is not a line about how much of a text a tool touched, measured in words or percentages. It is a line about where the substance came from.

Three biblical criteria give that distinction teeth rather than leaving it a matter of taste.

The first is truthfulness about origin. The ninth commandment forbids bearing false witness (Ex. 20:16), and while the commandment was framed for courtroom testimony, its principle reaches any claim that misrepresents a matter to someone entitled to rely on the truth of it. A reader of a theological essay, a sermon, or a personal testimony reasonably assumes the words reflect the author's own study, conviction, and wrestling with the text, unless he is told otherwise. Generative use that conceals its own extent trades on that assumption dishonestly, even when every sentence it produces happens to be theologically sound. The falsehood, where there is one, is rarely in the content. It is in the silent claim of authorship the writer allows his reader to believe.

The second is accountability that cannot be delegated. James warned that not many should become teachers, because teachers will be judged with greater strictness (James 3:1), and Paul told the Corinthians that each of us must appear before the judgment seat of Christ, to receive what is due for what has been done in the body (2 Cor. 5:10). Teaching, preaching, and theological argument are forms of spiritual oversight exercised over other souls, and the weight of that oversight cannot be handed to an instrument incapable of bearing it. A machine cannot be examined by an elder board. It cannot repent of error. It cannot stand before God and answer for having led a congregation astray. Every word a Christian teacher publishes under his own name, he has personally undertaken to defend, and generative use that substitutes the machine's judgment for his own quietly transfers a burden Scripture will not allow him to set down.

The third is the discipline of verification, and it is where iterative use earns its name. Attebury checks every line of Zanchi's Latin against the machine's proposed translation, because he knows the tool can be fluently wrong.⁴⁷ Owen Anderson tells his philosophy students, with some chagrin, that AI produced a passable exam for his course on Augustine in a matter of seconds, and adds, rightly, that it could not have written the students' answers, could not have learned the material in their place, and could not have made them wise by the exercise, which was the entire point of assigning it.⁴⁸ What separates responsible use from irresponsible use is not whether a machine touched the work. It is whether a human being verified, corrected, and stood personally behind every claim before it left his hand.

Applied across genres, these criteria sort quickly. Grammar checking, translation checking, citation formatting, bibliographic search, and the tightening of an already-drafted paragraph fall on the iterative side without much difficulty; they amplify a craft the writer already exercises and leave every substantive claim under his own verification. A congregation that gathered in Germany in 2023 to hear a full worship service, sermon included, generated by ChatGPT experienced something else entirely: the appearance of pastoral care with no pastor behind it, a simulation of oversight that oversaw nothing.⁴⁹ Personal testimony sits furthest of all the genres theology touches from anything a machine can rightly supply, because testimony claims, by its very form, to report an experience only the speaker could have had. A machine has never been broken by the conviction of sin. It has never wept over its own rebellion, never felt the particular weight of a given afternoon when grace overtook it, never known what it is to be found after years spent running. A testimony generated rather than lived, then dressed in the grammar of confession, is not merely bad writing. It is a fiction wearing a martyr's clothes.

Iterative use extends the craftsman. Generative use replaces him and dresses the replacement in his own name. Scripture needed no separate commandment for this case, because it already had one: the ninth commandment covers the false claim, and the doctrine of the teacher's accountability covers the rest. What the machine cannot do is not, in the end, a technical limitation waiting on the next model release. It is a theological fact about what a human witness is for.

Coda

Return, in closing, to the sentence this essay began from. Yahweh God formed man of dust from the ground, and breathed into his nostrils the breath of life (Gen. 2:7). Every machine now producing plausible sentences at scale was itself built from a kind of dust: silicon refined from sand, arranged by human hands into circuits carrying current instead of breath. The animation of matter by electricity and code is not the giving of life, however convincingly it counterfeits the conversation of the living. Christians have no reason to fear this machine, and no reason to despise the good it can do in a translator's hands or a Bible society's workflow. They have every reason to withhold from it the trust, the authorship, and the pastoral office that belong, on the plain terms of Scripture, to those who were formed from dust and given breath by God Himself—and to no one, and nothing, else.

 

Suggested for Further Study

Attebury, David. “AI as Theological Babel Fish.” Christ Over All, May 29, 2024.

Avery, John. “‘Never Let a Crisis Go to Waste’: AI, Statism, and the Threats to Free Speech.” Christ Over All, May 30, 2024.

Beale, G. K. We Become What We Worship: A Biblical Theology of Idolatry. Downers Grove, IL: IVP Academic, 2008.

Boden, Margaret A. Artificial Intelligence: A Very Short Introduction. Oxford: Oxford University Press, 2018.

Bostrom, Nick. Superintelligence: Paths, Dangers, Strategies. Oxford: Oxford University Press, 2014.

Bridges, Jared. “Artificial Intelligence and the Problem of Personality.” Christ Over All, May 9, 2024.

Cochrane, Daniel. “The Tower of Babel and the Ideology of AI.” Christ Over All, May 15, 2024.

Dodds, Thomas. “Use AI For the Sake of Good Work.” Christ Over All, May 27, 2024.

Ellul, Jacques. The Technological Society. Translated by John Wilkinson. New York: Alfred A. Knopf, 1964.

Graves, Mark. “Generative AI and Theology: A Three-Year Retrospective.” Theology and Science 24, no. 1 (2026): 1–7.

Kirby, Mike, and Matt Emadi. “Artificial Intelligence (AI): Tool, Image Bearer, or Temptation?” Christ Over All, May 20, 2024.

Kurzweil, Ray. The Singularity Is Near: When Humans Transcend Biology. New York: Penguin Books, 2006.

Oviedo, Lluís. “AI and Theology: Looking for a Positive—But Not Uncritical—Reception.” Zygon 57, no. 4 (December 2022): 938–53.

Peckham, Jeremy. Masters or Slaves? AI and the Future of Humanity. London: IVP, 2021.

Postman, Neil. Technopoly: The Surrender of Culture to Technology. New York: Vintage Books, 1992.

Reinke, Tony. God, Technology, and the Christian Life. Wheaton, IL: Crossway, 2022.

Ryan, Dustin. “A Christian’s Perspective on Artificial Intelligence.” Christ Over All, May 6, 2024.

Schaeffer, Francis. “Perspectives on Art.” In The Christian Imagination: The Practice of Faith in Literature and Writing, edited by Leland Ryken. New York: WaterBrook, 2002.

Searle, John R. “Minds, Brains, and Programs.” Behavioral and Brain Sciences 3, no. 3 (1980): 417–57.

Thacker, Jason. The Age of AI: Artificial Intelligence and the Future of Humanity. Grand Rapids: Zondervan Reflective, 2020.

Valk, Jacob. “Why AI Pornography Is Far More Dangerous than Yesterday’s Porn.” Christ Over All, May 22, 2024.

Weizenbaum, Joseph. Computer Power and Human Reason: From Judgment to Calculation. San Francisco: W. H. Freeman, 1976.

Wilkinson, Michael A. Crowned with Glory and Honor: A Chalcedonian Anthropology. Bellingham, WA: Lexham Academic, 2024.

Notes

1. OpenAI, "Introducing ChatGPT," November 30, 2022, https://openai.com/index/chatgpt/.

2. Krystal Hu, "ChatGPT Sets Record for Fastest-Growing User Base—Analyst Note," Reuters, February 1, 2023.

3. A. M. Turing, "Computing Machinery and Intelligence," Mind 59, no. 236 (October 1950): 433–60.

4. Ron Karjian, "The History of Artificial Intelligence: Complete AI Timeline," TechTarget, August 16, 2023, https://www.techtarget.com/searchenterpriseai/tip/The-history-of-artificial-intelligence-Complete-AI-timeline.

5. Karjian, "The History of Artificial Intelligence."

6. Ashish Vaswani et al., "Attention Is All You Need," Advances in Neural Information Processing Systems 30 (2017): 5998–6008; cf. Karjian, "The History of Artificial Intelligence."

7. Cole Stryker, "What Is AI?," IBM, August 9, 2024, https://www.ibm.com/think/topics/artificial-intelligence, quoted in Dustin Ryan, "A Christian’s Perspective on Artificial Intelligence," Christ Over All, May 6, 2024.

8. Ryan, "A Christian’s Perspective on Artificial Intelligence."

9. Ryan, "A Christian’s Perspective on Artificial Intelligence"; OpenAI, "DALL·E 3 System Card," October 3, 2023, https://openai.com/index/dall-e-3-system-card/.

10. Mike Kirby and Matt Emadi, "Artificial Intelligence (AI): Tool, Image Bearer, or Temptation?," Christ Over All, May 20, 2024.

11. Kirby and Emadi, "Tool, Image Bearer, or Temptation?"; cf. John R. Searle, "Minds, Brains, and Programs," Behavioral and Brain Sciences 3, no. 3 (1980): 417–57.

12. David Attebury, "AI as Theological Babel Fish," Christ Over All, May 29, 2024.

13. Attebury, "AI as Theological Babel Fish."

14. Attebury, "AI as Theological Babel Fish."

15. Attebury, "AI as Theological Babel Fish."

16. Don Barger, "How AI Assists in Global Bible Translation," The Gospel Coalition, February 26, 2024, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

17. IBM, "What Is Artificial Intelligence in Medicine?," August 4, 2021, https://www.ibm.com/think/topics/artificial-intelligence-medicine, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

18. Ryan, "A Christian’s Perspective on Artificial Intelligence."

19. Lluís Oviedo, "AI and Theology: Looking for a Positive—But Not Uncritical—Reception," Zygon 57, no. 4 (December 2022): 946–47.

20. Mark Graves, "Generative AI and Theology: A Three-Year Retrospective," Theology and Science 24, no. 1 (2026): 2.

21. Graves, "Generative AI and Theology," 4.

22. Cf. Thomas Dodds, "Use AI For the Sake of Good Work," Christ Over All, May 27, 2024.

23. Rachel Metz, "How a Deepfake Tom Cruise on TikTok Turned into a Very Real AI Company," CNN Business, August 6, 2021, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

24. Matt Novak, "That Viral Image of Pope Francis Wearing a White Puffer Coat Is Totally Fake," Forbes, March 26, 2023, cited in Graves, "Generative AI and Theology," 2.

25. ABC13 Digital Team, "Disgruntled Ex-Athletic Director Used AI to Generate Fake Racist Rant in Principal’s Voice: Police," ABC13, April 26, 2024, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

26. Natasha Singer, "Teen Girls Confront an Epidemic of Deepfake Nudes in Schools," New York Times, April 8, 2024, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

27. Emmanuel Saliba, "Sharing Deepfake Pornography Could Soon Be Illegal in America," ABC News, June 15, 2023, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

28. Jacob Valk, "Why AI Pornography Is Far More Dangerous than Yesterday’s Porn," Christ Over All, May 22, 2024.

29. Valk, "Why AI Pornography Is Far More Dangerous."

30. Graves, "Generative AI and Theology," 2.

31. Joseph Weizenbaum, Computer Power and Human Reason: From Judgment to Calculation (San Francisco: W. H. Freeman, 1976), cited in Graves, "Generative AI and Theology," 2.

32. Cf. Jared Bridges, "Artificial Intelligence and the Problem of Personality," Christ Over All, May 9, 2024.

33. Carrie Mihalcik, "Google ChatGPT Rival Bard Flubs Fact about NASA’s Webb Space Telescope," CNET, February 9, 2023, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

34. Gerrit De Vynck, "ChatGPT Leans Liberal, Research Shows," Washington Post, August 16, 2023, cited in Ryan, "A Christian’s Perspective on Artificial Intelligence."

35. Graves, "Generative AI and Theology," 1.

36. Exec. Order No. 14110, "Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence," Federal Register, October 30, 2023; Chuck Schumer, "Majority Leader Schumer Remarks at Rules Committee Markup on the Impact of Artificial Intelligence on Our Elections," May 15, 2024; both cited in John Avery, "‘Never Let a Crisis Go to Waste’: AI, Statism, and the Threats to Free Speech," Christ Over All, May 30, 2024.

37. Cf. Anton Brown, "Can I Upload My Consciousness into the Body of a Dog?: Thinking Biblically about AI-Enhanced Human Futures," Christ Over All, May 17, 2024.

38. Searle, "Minds, Brains, and Programs," 417–24.

39. G. K. Beale, We Become What We Worship: A Biblical Theology of Idolatry (Downers Grove, IL: IVP Academic, 2008).

40. Daniel Cochrane, "The Tower of Babel and the Ideology of AI," Christ Over All, May 15, 2024, drawing on K. A. Mathews, Genesis 11:27–50:26, The New American Commentary (Nashville: B&H, 1996), 481–84.

41. Cf. Brown, "Can I Upload My Consciousness into the Body of a Dog?"

42. Cochrane, "The Tower of Babel and the Ideology of AI."

43. Dodds, "Use AI For the Sake of Good Work," drawing out the same passage.

44. Attebury, "AI as Theological Babel Fish."

45. Kirby and Emadi, "Tool, Image Bearer, or Temptation?"

46. Graves, "Generative AI and Theology," 3.

47. Attebury, "AI as Theological Babel Fish."

48. Owen Anderson, "Whatever Comes, Get Wisdom: AI, the Future, and Our Chief End," Christ Over All, May 3, 2024.

49. Kirsten Grieshaber, "Can a Chatbot Preach a Good Sermon? Hundreds Attend Church Service Generated by ChatGPT to Find Out," AP News, June 10, 2023, cited in Graves, "Generative AI and Theology," 2.