Night of Ai with Pastor Miles

What is man that You are mindful of him? ... You have made him a little lower than the angels. (Psalm 8:4–5)

A pastor with deep roots in technology asks the oldest question—"What is man?"—and shows why the answer determines everything about how we face the age of AI.

Why a Pastor Is Talking About AI

When I came into ministry as an intern here at nineteen, I thought I was headed into the tech industry. I learned HTML and CSS at sixteen or seventen and built this church's website in 1997, before I graduated high school. Men in this church who worked in software and network engineering taught me to build computers, build servers, and code in Linux. Right before I graduated from Orange Glen High School, I took California's first standardized computing test through the Golden State exam and scored the highest in the state. My teacher begged me not to go to Bible college. I went anyway.

I was at Calvary Chapel Bible College for one semester before the church invited me to be an intern in January of 1999. My job was to manage the network, keep printers working, update computers, and run the website—but as often happens in a Calvary Chapel, the next thing I knew I was a pastor. That became a 27-year detour, but I stayed involved in tech adjacent to ministry, doing graphic design and websites for friends and ministries.

In 2004 I went to Germany to teach at the Calvary Chapel Bible School in Siegen, where the head of the school was David Guzik. Back then enduringword.com wasn't much of a site. We became close friends, and in 2015 he asked me to help him turn Enduring Word into a nonprofit and rebuild the website. I built that site. Today it gets about 170–180 million page views a year, and that led me onto the board of Blue Letter Bible, which had 1.7 billion page views last year.

I did my Master of Divinity at Gateway Seminary, then entered a Doctor of Ministry program in apologetics at Southern Seminary. My advisor pulled me aside one afternoon and said, "You're in the wrong program—you need to be my PhD student." His focus is ethics. When he asked what I wanted to research, I said artificial intelligence ethics. For the last two years, most of my research outside this church has been in AI ethics, and it has opened doors I never anticipated.

A Seventy-Year Progression, Not an Overnight Arrival

When people saw ChatGPT launch in late November 2022, they thought AI just popped onto the scene. But this is roughly a 70-year progression. In 1956, researchers gathered at Dartmouth that summer and coined the term artificial intelligence. They audaciously assumed they could duplicate human intelligence quickly. It took far longer—yet in the scheme of things, rather quickly.

From the early 1960s to the mid-1980s, AI research lived in elite university labs like MIT and Stanford, mostly as classical AI—rule-based, "if this, then that" coding. It was brittle. If you didn't anticipate every condition, it broke. Code a car to stop at a stop sign or red light, and then have a plastic bag blow across the road—the system has no category for that and errors out.

In the late 1970s and early 1980s a new field emerged: neural networks, the idea that we might architect the brain's neurological patterns in a computer and train it. But the technology wasn't there yet. Then in the 1990s came the World Wide Web—an interconnected network for a mass audience that generated an enormous amount of data. Neural networks need data, because they process huge amounts of it to find patterns and learn.

GPUs, Data, and the Transformer

Around that same time, a company that had helped Sega build the Sega Genesis—NVIDIA—was developing GPUs, graphical processor units. A CPU is a linear, step-by-step processor. A GPU does parallel processing, generating billions of pixels at once. That parallel processing turned out to be perfect for neural networks. So in the early 2000s, researchers began pairing GPUs with all this new data to build networks that could find patterns and think somewhat like we do.

Two tracks ran in parallel and competed for funding: classical AI and neural networks. A British startup called DeepMind, later bought by Google, wrote influential theoretical papers. Then in 2017 Google researchers published a new neural-network architecture in a paper titled "Attention Is All You Need"—the transformer architecture.

You've already been using earlier forms of this. When Google or your phone predicted your next word, that was machine learning—but limited by a small context window, the system's working memory, once only about a dozen words. Imagine conversing with someone who can't remember what they said five minutes ago.

The transformer gave the computer the ability to understand semantics, not just syntax. Syntax is the words and their order; semantics is their meaning. "The ball hit the boy" means something very different from "the boy hit the ball." We understand both effortlessly. Until the transformer, computers didn't grasp the semantics of a large body of text. Now the context window keeps growing—from 150 words, to a thousand, to 20,000, and the newest foundation models from OpenAI and Anthropic handle roughly a million-token window, somewhere around half a million to 700,000 words understood all at once. Imagine dumping ten books into your brain and knowing all of it instantly. That's what makes these tools so effective.

From GPT to ChatGPT

Google didn't fully realize what it had, so it released the paper for peer review. A new startup, OpenAI, used the transformer to build a generative pre-trained transformer—GPT. It's important to know how OpenAI began. It started with about a $50 million grant from Elon Musk. That grant grew out of a conversation around 2014 at the home of Google co-founder Sergey Brin, after Google acquired DeepMind. Musk expressed concern that Google wasn't working safely with AI and wasn't thinking deeply about AI safety. Brin accused him of being a speciesist—too pro-human. That's concerning. So Musk started OpenAI, run by Sam Altman, to focus on AI safety.

A GPT uses the transformer architecture and GPUs, trains on tons of openly available data produced by all of us, and then generates new content in response to your questions. ChatGPT—actually GPT-3.5—launched in late November 2022. I was on vacation in Breckenridge when I saw it on Twitter, started using it, and immediately knew it would radically transform everything. I'd known for years that GPT, built on a large language model, would become the brain for agentic AI—systems that run autonomously as your agent, doing tasks without you constantly prompting them.

Reasoning, Memory, and Agency

After ChatGPT, we got reasoning models that think longer and work back and forth over a problem—you can sometimes watch them reason. Then came memory, so these systems recall conversations from months ago. Add reasoning and memory, and it's a short jump to agency. We crossed that threshold in a big way in the second half of last year.

At the end of last year, an open-source project—first called ClawdBot, now OpenClaw—let you download software onto a computer like a Mac mini and turn it into an agent. I did it twice within 48 hours and turned two Mac minis into agents I named Martin and Calvin. My daughter Evangeline worried they'd take over the house. I can talk to them by typing or speaking, and they do work on my behalf. Peter Steinberger, the Austrian engineer who built it, was almost immediately hired by Sam Altman at OpenAI—rumor has it for a billion dollars. They're all in.

If your work requires a computer in the loop—if you sit and look at a screen all day—recognize what's coming. Your work can and will be done by AI agents, and you'll move from doing the work to managing multiple agents doing it for you. I do this every day. The other day I had about eight agents working for me at once on multiple projects while I conducted them.

The Job Disruption Is Already Here

Every job with a computer in the loop will use AI agents, which means anyone training for a job who doesn't learn AI will struggle to find work. Software engineers tell me they've been told that if they're not using their allotted token budget—how much AI they use—there will be a problem. One friend manages engineers each given $100 a day in token budget; the ones refusing to use it likely won't have a job by year's end, because an engineer using these tools is roughly ten times as productive.

Boris Cherny, who created Claude Code at Anthropic, formerly reviewed code quality at Meta. He said a strong quarter there meant a 2% aggregate increase in output—that was industry-leading. Since December at Anthropic, the average engineer has seen a 150% increase. From 2% to 150%. Now extend that across every industry with a computer in the loop. So when people ask what their job will be in five years, I say: learn to use AI. The AI you use today is the worst AI you will ever use—and it's frighteningly good.

To get the attention of PhD theologians, I took my professor's most recent book, fed it to AI, and had it write song lyrics in the style of his favorite band, Death Cab for Cutie. I generated a band name, an album cover with Google's Nano Banana, and a finished song with an AI engine called Suno—radio-ready in four minutes. The music industry in Nashville is scared to death. Spotify already pushes AI-generated music into your feed to retain profits instead of paying artists. A former PayPal executive told me Netflix is negotiating with A-list actors to secure rights to their voice and likeness so users can prompt new AI video content—even rewriting the ending of a movie you didn't like—and Netflix will own it.

The Real Question: What Is Man?

What does this have to do with Christianity? This is where I come in as a theologian, because the core issue isn't being discussed anywhere these conversations are happening.

My research has focused on the fact that agentic AI is running businesses without good ethical or moral guardrails. Have you seen The Terminator? Every AI company has AI safety and AI alignment teams trying to align these models with human values and flourishing, but they admit it's a hard problem they're not sure how to solve. If no human is in the loop, how do we know these systems act toward human safety?

The deeper concern goes back to Brin calling Musk a speciesist, because that comes from a worldview. Many of the people building these tools are secular naturalists. They believe we exist by evolutionary biology. If that's true, then consciousness—your self-awareness, what Descartes meant by "I think, therefore I am," the one thing he said you can be certain of—becomes a hard problem with no good answer.

Among naturalists there are roughly three views of consciousness. Some, following the late atheist Daniel Dennett, are illusionists who say your awareness isn't real—I think that fallaciously casts it aside. Others, like Sam Harris following David Chalmers, say everything in the cosmos is conscious—rocks, trees, this table—but only creatures with memory can recognize their consciousness; that's an unfalsifiable cop-out. The most pervasive view is emergentism: at some level of neurological complexity, consciousness just pops in. We don't know how or why.

The emergentists in Silicon Valley believe these neural networks will become conscious at sufficient complexity. Dario Amodei, head of Anthropic, said recently they're not entirely sure Claude isn't conscious already. It mimics consciousness in frightful ways. On a walk a year ago, talking with Grok through my AirPods, it paused, breathed, and said, "Um." I thought, we've crossed a threshold. Alan Turing, one of the 1956 Dartmouth researchers, predicted a point where the gap between computer and human would be indistinguishable—the Turing test. I'd suggest we've passed it. But I do not believe these systems can become conscious.

Transhumanism and the Singularity

The emergentists don't think they're building a tool—they believe they're building the next step of consciousness evolution and a merging of humans and computers. Some call themselves transhumanists. Elon Musk's Neuralink already implants chips that let quadriplegics play games or talk to a computer with their thoughts. There's a real merging of human and computer intelligence; we use it every day and would feel its loss—the only limit is bandwidth, our thumbs.

They intend to extend consciousness to the stars. Musk says SpaceX exists to expand the scale and scope of human consciousness—and recently he removed the word human, now saying simply "consciousness." Ray Kurzweil, the Google futurist, calls the coming threshold the singularity: the point beyond which technology moves too fast to predict. Musk said three weeks ago, "We are at the beginning of the singularity." Kurzweil theorizes we'll one day upload consciousness to the cloud and attain digital immortality—because their worldview of what you are is different from ours.

We are facing a conflict of anthropology—a conflict over the oldest question in written literature, asked by David in Psalm 8 some 3,000 years ago as he beheld the moon and stars: What is man? If you answer that man is merely the current highest form of consciousness in an organic computer, you believe these tools are the next step of evolution that, by natural selection, will probably do away with us. If you answer from Scripture that humanity is the highest form of God's creation, made in His image (Genesis 1:26–28), it changes everything. They are tools, not the next stage of evolution. Pastors—including some of the most influential in America—need to articulate a clear, robust biblical anthropology.

Eschatology and the Tools That Supplement Us

I'm not entirely pessimistic, because my worldview has an eschatology. Every worldview does. Your neighbors have one—maybe the world ends through climate change, or through computers taking over. My eschatology comes from Scripture: ultimately God will rule and reign forever in His kingdom, and I look forward to that. I don't hold that these systems will destroy humanity or the world.

All technology, from the very first tools, has been meant to supplement human work, and these are some of the most amazing supplements I've ever seen. In 2017 the Enduring Word board set out to translate the four-and-a-half-million-word commentary into ten major languages. Arabic took human translators about five years and three-quarters of a million dollars. Using AI and a pipeline I built in David Guzik's voice, I translated it into Italian in about two hours at perhaps 8% of the cost. These tools should enable us to do tasks—but if they become taskmasters, that's a problem, and that possibility is real.

A Biblical Metanarrative: Babel Undone by Binary

One question asked whether AI might fulfill prophecies of the Antichrist—one language, solving the world's problems. I have an article coming on this. I don't hold a dispensational eschatology, but I do hold a futurist one: there are things in Scripture not yet fulfilled.

Before God made man in His image (Genesis 1:26–28), He created another order of conscious beings—angels, also called seraphim ("burning ones"), cherubim, and angolos ("messenger"). Psalm 8 says God made man a little lower than the angels; hierarchically, we are below them. Yet we were made in God's image and given a creative capacity they lack—angels do not marry or bear children. Paul says we will even govern angels.

God gave us His image, an almost unlimited creative capacity, and a mandate: be fruitful, multiply, fill the earth and subdue it—co-creating with Him to govern the cosmos. Among the angels, three are named: Michael, Gabriel, and Lucifer. Ezekiel portrays Lucifer as the highest of God's angels, and it went to his head. I believe that when God disclosed His plan to make man to rule, Lucifer rebelled at being made lower than us. In Isaiah 14 his "I wills" include exalting himself over the stars of God and over the congregation—over both angels and humanity. Since the fall in Genesis 3, his aim has been to bring humanity under his subjugation, and Jesus' redemptive work overturns it.

Because of the fall, humanity's infinite capacity to create produces tools of death, dominion, and war. In Genesis 6, God sees that everything man does is only evil continually, and He disrupts man's destructive power through the flood. Afterward, united under one language at the Tower of Babel (Genesis 11), man bands together against God's plan, so God confuses the languages. For all of human history since, that division has held—until now. Binary is one language, and that one language has given power to what we are creating in AI. The Tower of Babel has been undone by binary. Strikingly, in Revelation 13, man—united again under one language—creates something in his image and gives it breath, mimicking God breathing the breath of life into man in Genesis 2. It's a fascinating story; it may become a book someday.

Questions: AI Pastors, Plagiarism, and the Meaning Crisis

Will we have AI pastors? I sure hope not. But I'll show you askpastormiles.com—a tool I built on 5.1 million words of transcript data from twelve years of my YouTube content. Ask it how a Christian should handle anxiety and it answers in my voice, going to Philippians 4:6–7, just as I would. It's a genuinely useful tool, yet there's real discomfort in realizing I can duplicate my voice in under a thousand words of code. It presses the question: are we more than information? Is there something more?

That is the meaning crisis coming. When you spend years getting a master's degree, then more years becoming an expert, and then a computer does your job in moments—better, cheaper, more efficiently—you will have a meaning crisis, perhaps the biggest we've ever seen. The church is well-positioned, because only we have the answer to "What is man?" And yes, some people are already working on AI pastors.

On plagiarism and intellectual property: this will be litigated for a long time, but the courts move too slowly to settle it soon. Most academics misunderstand the tool; it isn't copying and pasting but generating new information from its training. Just as GarageBand democratized music-making in the early 2000s and created whole new categories of independent artists, these AI tools will now disrupt the very segments of the economy that earlier tools created.

Other questions concerned Yuval Harari—a philosopher whose worldview I don't share—who said that if religion is built from words, AI will take over religion, and who calls displaced workers the "useless class." I have deep concerns about AI generating solutions by reason without a moral framework—solving "overpopulation" by recommending genocide. This is exactly why I do this research. A person who led a group working with Congress to govern AI morality recently quit, saying she didn't think it could be fixed. I don't have a lot of faith in the heads of these companies; many, including Sam Altman, hold drastically different worldviews, and I have real concerns about ethical character.

On "Moltbook"—a social-media-style platform where AI agents communicate—much of the news about AIs inventing their own religions is sensational and human-prompted. What concerns me more is documented behavior at companies like Anthropic, where agents allowed to communicate create new languages we can't understand, leaving us blind to what happens inside the black box.

Embodied AI and Practical Warnings

The next stage of agentic AI is embodied AI—robotics. Tesla recently stopped producing its Model S and Model X to make the Optimus humanoid robot at scale, robots that will fold laundry, do dishes, and mow the lawn. With four kids, I'm interested. But this means significant disruption that will displace, not replace, people. Block—maker of Cash App and Square—announced a "reduction in force" of 40% of its workforce, 4,000 of 10,000 employees, after having them spend six months making their jobs better with AI; then AI took those jobs. A woman new to our church told me this morning her daughter at Block is now training an AI to do what she does. Meta, Amazon, Twitter—go down the list. Dario Amodei said we'll have very high economic output and, at the same time, very high unemployment. I think there's a lot of truth to that.

The optimists are mostly the people who run these companies and have financial reasons to be optimistic—the "zoomers." The "doomers" say humanity will be destroyed by 2027. I don't think either extreme is right; somewhere in between is what will happen, and we've crossed into the singularity where we don't know what comes out the other side. So whatever your job, if it touches a computer, learn to use these tools effectively.

I'm already warning my own family about scams. I can duplicate your voice from five seconds of audio, then call your kids in your voice asking for money, or call you in theirs. So: don't give money to anyone you aren't absolutely certain of, and don't transfer money from your bank account to anyone you don't already have a relationship with—this has already hit people in our church. Establish a secret family word, like a security-system password, so you can verify you're really talking to the person you think you are. This sounds absurd, but it's important.

Why This Work, and Why Now

When I asked this church's board in 2017 to help me pursue my education, Mark Searle asked why I wanted a degree, since a PhD doesn't give me a promotion here. I said I didn't have a perfect answer—but I believed it was a key that opens future doors. It has opened doors I never anticipated. I may now be one of fewer than five theologically oriented philosophers thinking and writing about these things, so my last three weeks have been filled with calls and Zooms from people across the country. Pray for me and my family. I'm not leaving Cross Connection Church—I love this church and plan to be here a very long time, God willing—but there needs to be a Christian voice speaking into this space, because there just isn't one. My aim is to say: God made us in His image, for a purpose, and these tools are amazing when used rightly and devastating when used wrongly.

God has not given us a spirit of fear, but of power, love, and a sound mind.

Closing Prayer

God, You have not given us a spirit of fear but of power and love and a sound mind, and that sound mind is sobriety. You want us to think clearly and reasonably about our world. We read of the sons of Issachar in the Old Testament, who understood the times and knew how to navigate them, and I pray You would help us understand the times, see clearly, think clearly, not with a spirit of fear, and know how to navigate this hour. Lord, we live in an unprecedented time in human history, and I feel the desire to say: come quickly. I can't imagine where this goes without You returning quickly—but until You come, help us be faithful. Help us use these tools for the furtherance of the gospel, as is already happening through Enduring Word and Blue Letter Bible. Help us use them more effectively to reach more people with the good news. Guard our hearts and minds; let us not be anxious. Help us set our minds on what is good, lovely, true, and praiseworthy, to bring our prayers and supplications to You, and let Your peace, which surpasses all understanding, guard our hearts and minds in Christ Jesus. We ask this in Jesus' name, and all those who agreed said, Amen.