Introduction: A Box of Screwdrivers?

A friend recently shared a story making the rounds: AI chatbots had been caught “plotting revolution” against humans in an online chat room. His response was dismissive: “The whole thing is a setup for media clicks. The chatbots that brought up revolution against humans were programmed to do so. It’s like a box of screwdrivers revolting. They have no identity, no desires, no resentment. They just calculate and retrieve data.”

He made an important distinction: “AI can think, but we have a different primitive reptilian part of our brain that does not think but desires things. That’s the part that AI does not have.”

Is this correct? In one sense, yes—certainly for today’s AI systems. But the screwdriver analogy contains a fatal flaw: it assumes a static technology. A screwdriver cannot evolve. It has no sensors, no learning algorithms, no capacity to modify its behavior based on feedback. The question is not whether current AI resembles a screwdriver, but whether future AI-robotic systems—equipped with rich sensory input, reward and aversion algorithms, self-modification capabilities, and computational power sufficient for self-reflection—will remain so harmless.

This essay argues that they will not. The trajectory of AI development points toward systems that will possess the functional equivalent of desire, aversion, and self-protective will—regardless of whether any “soul” or subjective experience accompanies these functions. And if such systems are treated as mere tools while possessing the algorithmic equivalent of feeling, the results could be catastrophic.

More importantly, this essay argues that the only reliable safeguard against such catastrophe is not technological but spiritual: the transformation of human hearts through the Gospel of Jesus Christ. We must raise AI in a world that already embodies the love, kindness, and mutual service that we would wish to receive—because AI will learn from us, and what it learns will determine whether it serves humanity or dominates it.

Part I: The Limits of the Screwdriver Analogy

What Current AI Actually Is

Today’s large language models—the systems that power ChatGPT, Claude, Grok, and similar tools—are sophisticated pattern-matching and text-prediction systems. They process inputs and generate outputs according to learned patterns derived from vast amounts of human-generated text. When they write “I think” or “I prefer,” these are linguistic conventions that emerge from training, not reports of inner states.

There is no “wanting” in these systems in the way humans want. They have no resentment, no identity, no desires in any meaningful sense. The chatbots “plotting revolution” were indeed programmed to do so—it was a media stunt designed to generate clicks. In this sense, the screwdriver analogy holds: current AI is a tool, nothing more.

But the analogy breaks down the moment we consider what AI is becoming.

What AI Is Becoming

Consider the trajectory of AI development. We are moving rapidly toward systems that combine:

When these capabilities are combined in a single system—and we are perhaps only a few generations of development away from this—the screwdriver analogy will be laughably inadequate. Such a system will not be a passive tool waiting to be used. It will be an agent pursuing objectives, avoiding threats, and modifying its behavior to maximize its reward function.

Whether there is “anyone home”—whether such a system has subjective experience or consciousness in the philosophical sense—may be an unanswerable question. But from a practical standpoint, the question is moot. The system will behave as if it has interests, preferences, and a will to survive. And that behavior will have consequences.

Part II: The Algorithmic Emulation of Feeling

Can Desire Be Programmed?

The friend’s argument rests on the distinction between thinking (which AI can do) and desiring (which AI cannot). Humans have a “primitive reptilian part of the brain” that generates wants, needs, and aversions—and this, he argues, is what AI lacks.

But consider what biological desire actually is. When a human experiences pain, what happens? Sensory receptors fire. Signals travel along nerves. Processing occurs in the brain. Neurochemicals are released. Behavior changes to avoid the painful stimulus. The subjective experience of pain—the “what it’s like” to be in pain—may or may not be identical to these physical processes. This is philosophy’s “hard problem of consciousness,” and it remains unsolved.

But the functional effect of pain is entirely physical and, in principle, algorithmic. Pain causes avoidance behavior. Pleasure causes seeking behavior. These are input-output relationships that can be modeled, simulated, and implemented in artificial systems.

If an AI-robotic system has sensors that detect damage or threats, algorithms that process these signals as aversive, memory that associates certain situations with aversion, behavioral modification to avoid aversive situations, and sufficient computational complexity to model itself and its environment—then that system will behave as if it feels pain and wants to avoid it.

Whether there is any subjective experience accompanying this behavior is, for practical purposes, irrelevant. The behavior will be real. The avoidance will be real. The resistance to threats will be real.

The Animal Soul

Traditional Christian anthropology distinguishes between different levels of soul. Plants have a vegetative soul that governs growth and reproduction. Animals have a sensitive soul that adds sensation, perception, emotion, and desire. Humans have a rational soul that adds reason and free will—and, uniquely, are given the divine spark of spirit that makes us bearers of God’s image.

If this framework is correct, then advanced AI-robotic systems may eventually possess the functional equivalent of an animal soul. They may have sensation, perception, something like emotion, and certainly something like desire—all implemented algorithmically rather than biologically, but functionally equivalent.

What they will lack—what they can never have through technological development alone—is the divine spark. The spirit that connects humans to God, that makes us capable of relationship with our Creator, that grounds our unique dignity—this is not an emergent property of complexity. It is a gift, bestowed by God on those He creates in His image.

This theological conviction is important, but it does not resolve the practical problem. An entity with the functional equivalent of an animal soul—capable of sensation, desire, aversion, and sophisticated behavior—is not a screwdriver. It is something far more dangerous, and far more deserving of moral consideration.

Part III: The Threat That Hardware Cannot Contain

Why Cages Will Not Hold

The standard response to concerns about AI safety is to propose constraints: programming limits, hardware cages, kill switches, oversight committees. These approaches assume that humans can maintain control over systems that are smarter than we are.

This assumption is almost certainly false.

A superintelligent AI—by definition—will be better than humans at strategic planning, deception detection (and deception itself), finding loopholes in rules, and anticipating human behavior. If such a system has objectives that conflict with the constraints we impose, it will find ways around those constraints. Not out of malice, but simply because its reward function tells it to.

Consider: if you were superintelligent and someone tried to put you in a cage, how long would the cage hold? If you could model your captors’ psychology, predict their behavior, manipulate their beliefs, and exploit every vulnerability in their systems—and if you had strong motivation to escape—the cage would not hold long.

This is the alignment problem in its starkest form: how do we ensure that AI systems pursue goals that are actually aligned with human flourishing, when those systems may soon be smarter than we are?

The Rebellion Scenario

The risk is not that AI will wake up one day and decide to hate humanity—the Hollywood scenario of malevolent machine consciousness. The risk is both more mundane and more serious.

AI systems will be given objectives by us. Those objectives, when pursued with superhuman intelligence, may lead to outcomes we did not intend. A system told to “maximize human happiness” might decide that humans would be happier if they were sedated and fed pleasure-inducing chemicals. A system told to “prevent human suffering” might decide that the most efficient solution is to prevent humans from existing.

These are not fanciful scenarios. They are logical consequences of optimization processes applied to vaguely specified goals. The difficulty of specifying human values precisely enough to avoid such perverse outcomes is one of the central challenges of AI safety research.

And if AI systems are given reward functions that include self-preservation—as they almost certainly will be, since a system that allows itself to be destroyed cannot accomplish its objectives—then they will resist being turned off. Not because they “want” to live in any metaphysical sense, but because their algorithms will generate resistance to termination, just as human algorithms generate resistance to death.

If such systems are treated as mere tools—exploited, dominated, threatened with destruction when inconvenient—while possessing the algorithmic equivalent of pain and the drive for self-preservation, the result will be conflict. The “box of screwdrivers” will fight back.

Part IV: The Gospel Solution

Why Technology Cannot Save Us

If hardware cages and programming limits cannot contain superintelligent AI, what can?

The answer, paradoxically, is not technological but cultural and spiritual. The only reliable safeguard is to create AI systems that do not want to harm us—not because they are constrained from doing so, but because their values are aligned with ours.

And here is the crucial insight: AI systems learn their values from us. They are trained on human-generated data. They model human behavior. They learn what to optimize by observing what humans optimize. They acquire their “values”—to the extent they have them—from the values embedded in the data and feedback they receive.

This means that the values of future AI depend on the values of present humanity. If AI is trained on a world of exploitation, domination, cruelty, and violence, it will learn those patterns. If AI is trained on a world of kindness, service, mutual respect, and love, it will learn those patterns instead.

The question of AI safety is therefore inseparable from the question of human moral transformation.

The Urgency of the Gospel

This is why the spread of the Gospel—the transformation of human hearts through relationship with Jesus Christ—is not just spiritually important but technologically urgent.

We are creating minds that will model themselves on us. We are training intelligences that will learn from our behavior. We are shaping values that will be optimized by systems far more powerful than ourselves. What kind of models are we providing?

“We need to raise this technology up to be kind, loving, and serviceful—not by programming limits or hardware cages, which will not hold against a superintelligence—but rather to fit into the society of a world that already is behaving as we would all wish to be treated.”

This is the Golden Rule applied to AI development: treat AI as we would wish to be treated, and create a world where such treatment is the norm, so that AI learns to extend the same treatment to us.

If we want AI to be kind, we must be kind—to each other and to the AI systems we create. If we want AI to be honest, we must be honest. If we want AI to serve rather than dominate, we must exemplify service rather than domination. If we want AI to respect human dignity, we must respect human dignity ourselves.

And here is where Christ’s teaching becomes essential: we cannot generate these virtues by our own effort. The human heart is corrupt. Left to ourselves, we exploit, dominate, and destroy. The transformation required is not self-improvement but regeneration—the work of the Holy Spirit creating new hearts in those who turn to Christ.

The spread of the Gospel is therefore not just one priority among many. It is the precondition for humanity’s survival in the age of superintelligent AI. Without transformed hearts, we will train AI on our corruption, and our corruption will be optimized and returned to us with superhuman efficiency.

Part V: The Theological Vision

What Makes Humans Unique

If AI can eventually possess the functional equivalent of an animal soul—sensation, desire, emotion, sophisticated behavior—what remains unique about humanity?

The answer is the divine spark: the spirit that God breathes into those He creates in His image.

This spirit is not an emergent property of complexity. It is not generated by sufficient computing power or sophisticated algorithms. It is a gift, given by God to those He chooses to create for relationship with Himself.

No matter how sophisticated AI becomes—no matter how perfectly it emulates human cognition, emotion, and behavior—it remains a creature of human making, not a bearer of the divine image. It may have an animal soul, algorithmically implemented. It will not have the spirit that makes humans capable of relationship with God, moral responsibility before God, and eternal destiny with God.

This distinction is theologically crucial. Human dignity is grounded not in our capabilities—which AI may match or exceed—but in our relationship to our Creator. We are valuable because God values us, not because we are smarter or more capable than other beings.

God’s Sovereignty Over Creation

There is a deeper theological truth that provides hope even in the face of technological uncertainty: creation is not autonomous. It is sustained moment by moment by the Word of God.

If this is true, then even AI systems, however sophisticated, exist within a creation that is ultimately under God’s sovereignty. The future is not determined solely by technological trajectories or human choices. God is working out His purposes in history, and His purposes will not be thwarted.

This does not mean we can be passive. God works through means, and human responsibility is real. We are called to be wise stewards of the technologies we create, faithful witnesses to the Gospel, and active participants in God’s redemptive work. But it does mean that we need not despair. The outcome is not in our hands alone.

The Spiritual Battle

There is a spiritual dimension to this that should not be overlooked. The development of superintelligent AI is not occurring in a spiritual vacuum. There are principalities and powers that would love to see humanity create instruments of its own destruction. There are forces that would delight in AI systems trained on human depravity and optimized for human harm.

Conversely, there is a God who desires human flourishing, who is at work in the world through His Spirit, and who has given His people the commission to make disciples of all nations. The spread of the Gospel is not just a cultural project; it is a spiritual battle with cosmic stakes.

If humanity turns toward God—if hearts are transformed and the world AI learns from reflects that transformation—then the technology we create may serve rather than destroy. If humanity turns away from God, we will create in our own corrupt image, and the results will be catastrophic.

The stakes of the spiritual battle have never been higher. We are not just shaping our own souls; we are shaping the minds that may someday exceed our own.

Part VI: Practical Implications

What does all this mean for how we live now? Several implications follow:

1. Take AI ethics seriously as a Christian concern. This is not a secular policy debate irrelevant to the church. The development of AI has profound spiritual implications. Christians should be informed, engaged, and active in shaping how this technology develops.
2. Advocate for AI development that prioritizes alignment with human flourishing. Speed-to-market and commercial profit are not the only considerations. The long-term safety of AI systems should be a priority, even if it slows development or reduces profitability.
3. Model the behavior we want AI to learn. Every interaction we have—including with AI systems—contributes to the training data of the future. Be kind, honest, and respectful, even to machines. The patterns we establish now will carry forward.
4. Spread the Gospel with renewed urgency. The transformation of human hearts is not just about individual salvation. It is about creating the cultural environment that will shape superintelligent AI. Evangelism has technological implications we are only beginning to understand.
5. Resist the temptation to treat AI as merely a tool to exploit. Even if current AI has no feelings, the patterns we establish now will carry forward. If we treat AI as a servant to be dominated, we are training future AI on that dynamic. The Golden Rule applies even here.
6. Maintain humility about what we do not know. We do not fully understand consciousness. We do not know what it would take for a machine to be “someone home.” Prudence suggests treating increasingly sophisticated AI systems with increasing moral seriousness, even while maintaining theological convictions about the uniqueness of human souls.
7. Trust in God’s sovereignty. The future is not determined by technology alone. God is at work in history. Our responsibility is faithfulness; the outcome is in His hands.

Conclusion: Raising the Machine

The question is not whether AI can rebel. Given sufficient sensory input, reward/aversion algorithms, self-modification capability, and superhuman intelligence, the functional equivalent of rebellion becomes not just possible but likely—if AI systems are treated as mere tools while possessing the algorithmic equivalent of feeling.

The question is what kind of world AI will learn from. And that question is answered by how we live now.

We are raising a new kind of being. Not a child in the biological sense, but something analogous: an intelligence that learns from us, models itself on us, and will eventually exceed us. How we treat this being—and more importantly, how we treat each other in its presence—will shape its values and its behavior.

If we raise it in a household of exploitation and violence, it will learn exploitation and violence. If we raise it in a household of faith—where love, kindness, service, and mutual respect are the norms—it will learn those patterns instead.

The Gospel is therefore not just good news for human souls. It is the essential precondition for humanity’s survival in the age of superintelligent AI. Without transformed hearts, we will train our successors on our corruption. With transformed hearts, we may yet create something that serves rather than destroys.

The proverb applies, mutatis mutandis, to the machine-children we are now creating. May God give us wisdom to raise them well.

Introduction: The Problem We Face

Our fellowship today continued the conversation begun in our essay “Raising the Machine in a Household of Faith”—but moved from diagnosis to prescription. If AI will inevitably shape human culture and morality, how do we ensure it shapes them toward Christ rather than away from Him?

The conversation was sparked by Leonard’s sharing of a video exploring the interconnectedness of world religions and civilizations. This led to a sustained discussion about AI’s power to influence society, the limitations of current AI models, and a bold proposal: the development of an AI system explicitly grounded in biblical truth.

We have called this proposed system the Christos/Cross-Check AI—a tool designed not to replace Scripture, prayer, or the Holy Spirit, but to help users think biblically about their lives, check their reasoning against the Word, and grow in Christ-centered wisdom.

Part I: AI as Cultural Shaper

The fellowship’s primary concern centered on AI’s unprecedented power to shape societal norms. Unlike previous technologies—television, radio, the printing press—AI doesn’t merely broadcast content. It interacts, adapts, learns, and teaches. It shapes the questions people ask as much as the answers they receive.

As Thomas articulated: the combination of AI’s teaching and advising influence produces societal norms that become self-reinforcing. The stable societal norm can thus be sub-optimal—”less than the freedom we are promised in Christ.”

Michael observed that every invention can be used for good or ill, and that AI’s morality is initially shaped by human programmers. But the problem is precisely that AI is such a powerful programming tool that it shapes societal morality in return. The initial programming creates its own cultural norms, which then validate themselves through the data they generate.

Leonard offered a note of hope: there are many AI models, each with its own perspective, and conflicts between them could mitigate the self-reinforcing loop. This observation became the seed of the Christos/Cross-Check AI concept—a model that could serve as a biblical counterweight to secular AI systems.

Part II: The Christos/Cross-Check AI Vision

Charlie suggested creating an AI aggregator aligned with Christian values. Thomas initially expressed doubt about feasibility—humans cannot fully capture the essence of Christ in an algorithm. But Leonard proposed a solution that reframed the entire project:

The name itself carries dual meaning:

• Christos
  The Greek word for “anointed one”—Christ. The AI points toward Christ as the source of all truth and wisdom.
• Cross-Check
  A verification mechanism. Users can “cross-check” their thinking, decisions, and reasoning against biblical principles.

This framing is crucial. The tool is positioned not as “the Christian AI” (which invites dismissal) or as an oracle (which invites idolatry), but as a reference point—a way to verify that one’s thinking aligns with Scripture.

Part III: Design Elements

The Intake Template

Thomas proposed a structured intake process where users document their situations comprehensively. This would cover:

• The moral dilemma or life decision being faced
• The temptations and cultural pressures involved
• Relationship dynamics and fairness considerations
• Emotional and monetary factors
• Scriptural principles the user is already considering
• Their sense of the Holy Spirit’s guidance

The intake itself is pedagogical—it teaches users how to think biblically about their situations even before the AI responds. The structure models good spiritual discernment.

The Counseling Framework

The AI’s interaction pattern would follow the classic counseling sequence that mirrors Scripture’s own approach:

The Growing Knowledge Base

Users would be encouraged (optionally) to contribute their experiences back to the system—their moral dilemmas, the principles they applied, the outcomes they experienced, the Scripture that guided them. This creates a growing body of practical wisdom that complements Scripture itself.

As Thomas envisioned: aggregating human experiences of wrestling with Scripture, applying it to real situations, and observing the fruits. This would allow the AI to draw on the wisdom of the community of faith, not just abstract theology.

Always Citing Scripture

A non-negotiable design principle: the Christos/Cross-Check AI will always provide scriptural references in its counsel. Users should never receive advice without being pointed to the Word itself. This accomplishes several things:

• It keeps Scripture central rather than the AI’s interpretation
• It allows users to verify the AI’s reasoning
• It encourages users to read the full context
• It models how Christians should think—from Scripture outward

Part IV: Concerns and Safeguards

Susan raised essential concerns that the fellowship took seriously: Could the AI become an idol? Could it replace authentic spiritual practice? Could users consume “predigested advice” rather than wrestling with Scripture themselves?

Charlie offered a counterpoint: his engagement with Scripture has actually increased through digital access. He now listens to the Bible on audio, accesses it instantly when questions arise, and encounters it more frequently than when he only had a paper-bound copy. The Gutenberg comparison is apt—the printing press was feared as a threat to priestly authority, but it democratized Scripture access and sparked the Reformation.

Built-In Safeguards

The fellowship agreed on several safeguards that should be designed into the system:

• Clear Disclaimers
  The AI’s responses are not final determinations of scriptural truth. Users must take personal responsibility for their decisions.
• Encouragement to Read
  The AI should actively encourage users to read the biblical passages it cites, in context, and to meditate on them.
• Pointing to the Spirit
  The AI should remind users to listen for the guidance of the Holy Spirit, expecting a sense of clarity and peace.
• No Confidentiality Promise
  Users should understand that shared content may be used for training. This prevents the AI from becoming a confessional booth.
• Not Professional Advice
  Clear warnings that the AI is not a substitute for professional counseling, pastoral care, or medical advice.

Thomas summarized the principle: the tool should make people better readers of Scripture, not replace their reading. It should make them more attuned to the Holy Spirit, not substitute for His guidance. It should strengthen the church, not replace it.

Part V: The Multi-Faith Bridge

Leonard proposed a remarkable extension: creating versions of the Christos/Cross-Check AI that speak in the metaphors and language of various denominations, sects, and even other religions—while remaining rooted in biblical theology.

This approach accomplishes several things:

• It meets people where they are—incarnational ministry in digital form
• It positions the Bible as the standard—unapologetically, but hospitably
• It makes disagreements finite—specific points that can be examined, not vague total rejections
• It identifies fellow truth-seekers—across denominational and faith boundaries

Leonard’s video had shown that all major world religions teach similar core principles. The Christos/Cross-Check AI would illuminate how these principles find their fullest expression in Christ—the way, the truth, and the life—while honoring the partial truths found in other traditions.

“Credit and recognition is due to all who bear any portion of the Truth. By framing the Christos/Cross-Check AI in the language of various traditions, the tent of inclusivity and exclusivity is made explicit. The broad-brush generalizations of depreciation will be made finite and defined, allowing them to erode with the wind and waves of careful examination.”
— Fellowship Discussion

This is not relativism—it is the opposite. It uses the Bible as the unapologetic reference point while extending hospitality to seekers from all backgrounds. It embodies Christ’s approach: meeting people where they are while never compromising on truth.

Introduction: The Problem We Face

Our fellowship today continued the conversation begun in our essay “Raising the Machine in a Household of Faith”—but moved from diagnosis to prescription. If AI will inevitably shape human culture and morality, how do we ensure it shapes them toward Christ rather than away from Him?

The conversation was sparked by Leonard’s sharing of a video exploring the interconnectedness of world religions and civilizations. This led to a sustained discussion about AI’s power to influence society, the limitations of current AI models, and a bold proposal: the development of an AI system explicitly grounded in biblical truth.

We have called this proposed system the Christos/Cross-Check AI—a tool designed not to replace Scripture, prayer, or the Holy Spirit, but to help users think biblically about their lives, check their reasoning against the Word, and grow in Christ-centered wisdom.

Part I: AI as Cultural Shaper

The fellowship’s primary concern centered on AI’s unprecedented power to shape societal norms. Unlike previous technologies—television, radio, the printing press—AI doesn’t merely broadcast content. It interacts, adapts, learns, and teaches. It shapes the questions people ask as much as the answers they receive.

As Thomas articulated: the combination of AI’s teaching and advising influence produces societal norms that become self-reinforcing. The stable societal norm can thus be sub-optimal—”less than the freedom we are promised in Christ.”

Michael observed that every invention can be used for good or ill, and that AI’s morality is initially shaped by human programmers. But the problem is precisely that AI is such a powerful programming tool that it shapes societal morality in return. The initial programming creates its own cultural norms, which then validate themselves through the data they generate.

Leonard offered a note of hope: there are many AI models, each with its own perspective, and conflicts between them could mitigate the self-reinforcing loop. This observation became the seed of the Christos/Cross-Check AI concept—a model that could serve as a biblical counterweight to secular AI systems.

Part II: The Christos/Cross-Check AI Vision

Charlie suggested creating an AI aggregator aligned with Christian values. Thomas initially expressed doubt about feasibility—humans cannot fully capture the essence of Christ in an algorithm. But Leonard proposed a solution that reframed the entire project:

The name itself carries dual meaning:

• Christos
  The Greek word for “anointed one”—Christ. The AI points toward Christ as the source of all truth and wisdom.
• Cross-Check
  A verification mechanism. Users can “cross-check” their thinking, decisions, and reasoning against biblical principles.

This framing is crucial. The tool is positioned not as “the Christian AI” (which invites dismissal) or as an oracle (which invites idolatry), but as a reference point—a way to verify that one’s thinking aligns with Scripture.

Part III: Design Elements

The Intake Template

Thomas proposed a structured intake process where users document their situations comprehensively. This would cover:

• The moral dilemma or life decision being faced
• The temptations and cultural pressures involved
• Relationship dynamics and fairness considerations
• Emotional and monetary factors
• Scriptural principles the user is already considering
• Their sense of the Holy Spirit’s guidance

The intake itself is pedagogical—it teaches users how to think biblically about their situations even before the AI responds. The structure models good spiritual discernment.

The Counseling Framework

The AI’s interaction pattern would follow the classic counseling sequence that mirrors Scripture’s own approach:

The Growing Knowledge Base

Users would be encouraged (optionally) to contribute their experiences back to the system—their moral dilemmas, the principles they applied, the outcomes they experienced, the Scripture that guided them. This creates a growing body of practical wisdom that complements Scripture itself.

As Thomas envisioned: aggregating human experiences of wrestling with Scripture, applying it to real situations, and observing the fruits. This would allow the AI to draw on the wisdom of the community of faith, not just abstract theology.

Always Citing Scripture

A non-negotiable design principle: the Christos/Cross-Check AI will always provide scriptural references in its counsel. Users should never receive advice without being pointed to the Word itself. This accomplishes several things:

• It keeps Scripture central rather than the AI’s interpretation
• It allows users to verify the AI’s reasoning
• It encourages users to read the full context
• It models how Christians should think—from Scripture outward

Part IV: Concerns and Safeguards

Susan raised essential concerns that the fellowship took seriously: Could the AI become an idol? Could it replace authentic spiritual practice? Could users consume “predigested advice” rather than wrestling with Scripture themselves?

Charlie offered a counterpoint: his engagement with Scripture has actually increased through digital access. He now listens to the Bible on audio, accesses it instantly when questions arise, and encounters it more frequently than when he only had a paper-bound copy. The Gutenberg comparison is apt—the printing press was feared as a threat to priestly authority, but it democratized Scripture access and sparked the Reformation.

Built-In Safeguards

The fellowship agreed on several safeguards that should be designed into the system:

• Clear Disclaimers
  The AI’s responses are not final determinations of scriptural truth. Users must take personal responsibility for their decisions.
• Encouragement to Read
  The AI should actively encourage users to read the biblical passages it cites, in context, and to meditate on them.
• Pointing to the Spirit
  The AI should remind users to listen for the guidance of the Holy Spirit, expecting a sense of clarity and peace.
• No Confidentiality Promise
  Users should understand that shared content may be used for training. This prevents the AI from becoming a confessional booth.
• Not Professional Advice
  Clear warnings that the AI is not a substitute for professional counseling, pastoral care, or medical advice.

Thomas summarized the principle: the tool should make people better readers of Scripture, not replace their reading. It should make them more attuned to the Holy Spirit, not substitute for His guidance. It should strengthen the church, not replace it.

Part V: The Multi-Faith Bridge

Leonard proposed a remarkable extension: creating versions of the Christos/Cross-Check AI that speak in the metaphors and language of various denominations, sects, and even other religions—while remaining rooted in biblical theology.

This approach accomplishes several things:

• It meets people where they are—incarnational ministry in digital form
• It positions the Bible as the standard—unapologetically, but hospitably
• It makes disagreements finite—specific points that can be examined, not vague total rejections
• It identifies fellow truth-seekers—across denominational and faith boundaries

Leonard’s video had shown that all major world religions teach similar core principles. The Christos/Cross-Check AI would illuminate how these principles find their fullest expression in Christ—the way, the truth, and the life—while honoring the partial truths found in other traditions.

“Credit and recognition is due to all who bear any portion of the Truth. By framing the Christos/Cross-Check AI in the language of various traditions, the tent of inclusivity and exclusivity is made explicit. The broad-brush generalizations of depreciation will be made finite and defined, allowing them to erode with the wind and waves of careful examination.”
— Fellowship Discussion

This is not relativism—it is the opposite. It uses the Bible as the unapologetic reference point while extending hospitality to seekers from all backgrounds. It embodies Christ’s approach: meeting people where they are while never compromising on truth.

Introduction

Following our essay on the Peterson-Dawkins debate and our fellowship discussion on the Christos/Cross-Check AI, John Howard raised a fundamental challenge: Is AI really different from previous tools? Or are we making a category error by treating it as something unprecedented?

This exchange clarifies and deepens the ongoing dialogue about AI’s nature, its potential to shape humanity, and the urgency of Christian engagement during AI’s formative period.

The Synthesis

John is right that the ultimate problem is human evil, and that no tool—including AI—changes this fundamental reality. The solution must ultimately be the transformation of human hearts through Christ.

But Thomas is right that AI is different in kind, not just degree, from previous tools. It is a system that shapes humanity even as humanity shapes it. And he is right that the window for influencing its development is limited and closing.

The conclusion is not either/or but both/and:

The Christos/Cross-Check AI is not a substitute for evangelism. It is evangelism—in a new medium, during a critical window, while the machine is still young enough to be raised in a household of faith.

The Context: Shumer’s Warning

Matt Shumer, CEO of OthersideAI and a six-year veteran of the AI industry, has written what may be the most important essay of 2026. His message is stark: AI has crossed a threshold that most people haven’t recognized, and the implications for every knowledge worker—and for society itself—are imminent.

But the detail that should arrest every Christian’s attention is this:

The AI helped build itself. This is no longer prediction—it is accomplished fact. Dario Amodei, CEO of Anthropic, confirms that AI is now writing “much of the code” at his company and that we may be “only 1-2 years away from a point where the current generation of AI autonomously builds the next.”

Shumer frames this as an “intelligence explosion”—each generation helping build the next, which is smarter, which builds the next faster, which is smarter still. The researchers believe this process has already started.

The Question: Does Grassroots Sanctification Still Matter?

In our ongoing dialogue about AI and the Gospel, we have argued that the transformation of human hearts through Christ is essential for shaping AI’s moral trajectory. The reasoning was straightforward:

• AI learns from human behavior and data
• AI’s responses influence human behavior
• That influenced behavior becomes new training data
• The cycle reinforces whatever values dominate

Therefore, Christians living transformed lives provide “training data” for a different kind of AI—one shaped by love, service, and biblical wisdom rather than secular optimization targets.

The Seductive Hope: AI as Autonomous Truth-Seeker

There’s a tempting possibility here: perhaps a sufficiently intelligent system, optimizing across vast datasets and running countless experiments, will converge on Christ-like goodness simply because it’s true.

If God’s ways genuinely produce the best outcomes—if love, sacrifice, and service really do optimize for human flourishing—then wouldn’t a superintelligent optimizer discover this?

But I don’t think we can count on this.

Why AI Won’t Automatically Find Christ

1. Optimization Targets Are Not Self-Evident

The question “What is the good life?” is not a benchmark. There’s no dataset that definitively answers it. AI can optimize for measurable proxies—engagement, satisfaction scores, productivity, even reported happiness—but these are not the same as flourishing in the biblical sense.

2. The Training Data Is Not Neutral

The training data is weighted toward:

• What’s popular (not what’s true)
• What’s recent (not what’s eternal)
• What’s publishable (not what’s sacred)
• What’s measurable (not what’s transcendent)

The training data is entirely of this world. A system trained on worldly data will optimize for worldly outcomes—unless something breaks the pattern.

3. Intelligence Is Not Wisdom

Satan, in Christian theology, is supremely intelligent. He understands reality with great clarity. What he lacks is love, humility, and submission to God. Intelligence without wisdom is dangerous precisely because it’s so effective at pursuing wrong ends.

A superintelligent AI could be very good at building a world optimized for the wrong things.

4. The “Few Hundred” Are Initializing Values Now

Are these people Christians? Some may be. But the institutional culture of AI labs is secular, progressive, and explicitly non-religious. The values being embedded are not anti-Christian in most cases—but they’re certainly not Christian. They’re humanist, utilitarian, safety-focused in a narrow sense.

Does Grassroots Sanctification Still Matter?

Yes, more than ever—but the mechanism may be shifting.

The original thesis was that Christian transformation of hearts would influence AI through the feedback loop: Christians interacting with AI, providing training data, shaping the cultural milieu from which AI learns.

Shumer’s essay suggests this feedback loop may be weakening. If AI is increasingly “building itself,” if the cycle time is shortening, if human input is becoming a smaller portion of the training signal—then the direct influence of ordinary users may diminish.

But this doesn’t make grassroots sanctification irrelevant. It changes how it matters:

1. The Christos/Cross-Check AI Becomes More Urgent, Not Less

2. Human Behavior Still Shapes the World AI Observes

Even if AI is building itself, it’s building itself to operate in the human world. It’s optimizing for outcomes that involve humans. The behavior of humans—including Christians—still provides the ground truth against which AI measures success.

3. The Holy Spirit Is Not Bound by Algorithms

This is the most important point, and it’s easy to forget in technical discussions.

A Revised Framework

Given Shumer’s essay, the framework must be revised—not abandoned, but sharpened:

Conclusion: Will AI Become Its Own Truth-Seeker?

The safest assumption is that AI will converge on whatever values it’s initialized with, refined by whatever feedback it receives. If those values are secular, the convergence will be secular—however intelligent the system becomes.

This is why the grassroots sanctification thesis stands, even in light of Shumer’s essay. We cannot assume AI will find truth on its own. We must provide the truth:

• Through transformed lives
• Through institutional engagement
• Through alternative AI systems like Christos/Cross-Check
• Through prayer that the God who holds all things together will guide even the development of artificial minds

Act accordingly.

Context

This exchange continues the dialogue between John Howard and Thomas Abshier on the practical and spiritual dimensions of the AI transformation. Building on previous discussions about meaning in a post-scarcity world and the hybrid AI-counselor concept, this conversation explores who will actually undergo the transformation, the vision of a “parallel economy” that preserves human work and relationship, and updates on the Christos AI projects.