James Hwang: Dr. Zhu, thank you for your time. You hold a PhD in physics from Cornell. How did that journey—from understanding the fundamental laws of the universe—lead to founding an AI company focused on serving the Christian Great Commission?
Dr. Zhu: It’s a fascinating journey that God led me through. I became a Christian in my junior year of college. In high school, I was all about STEM and math. In college, I majored in physics as a way to learn all the math I wanted without departmental restrictions. A number of my friends there did the same—we saw physics as the gateway to unlimited math.
My faith journey is another long story. I was a typical atheist, opposed to all religion, viewing it as a crutch for the mind. But everything changed when I encountered the Bible. I took it seriously—evaluating whether these things actually happened or whether it was all a hoax. In a sense, I used my scientific mind to assess what is true. That’s how I came to the gateway of faith.
Once I decided it was more likely that everything happened as the Bible said, I was willing to accept that possibility. Then I opened my heart to God, and that became a spiritual experience—finding him to be real. During my entire time in grad school, I was on fire for God. Studying physics became a way of understanding what my heavenly father had created.
When I was graduating, it was around the time the Cold War ended. Many physicists left defense and flooded the job market, and academic positions were scarce. But I wasn’t attached to being a physicist—I had been pursuing truth. So I didn’t mind changing direction. God led me to IBM Research, where I worked on speech recognition, specificallymachine learning. My wife and I spent many years there, working on the beginnings of machine translation and information retrieval—things that companies like Google and Amazon are known for today.
After seven years, I moved on to Wall Street. It wasn’t just about money. I wanted to understand how people work—how economics shapes behavior—rather than only studying the physical world. Then, after about ten years, our family felt led to move to California. That’s when I started to embrace my entrepreneurial side.
The timing was providential. Friends from my faith circles who also had tech backgrounds reached out. They had a vision for using big data to serve faith-based organizations. So we teamed up, as an experiment, to create AI and other cutting-edge tools tailored for ministry. Our model isn’t typical consulting. It’s proactive. Most faith leaders are trained in theology—not in tech—so we build things and say, “Look what’s possible. Would this help you?” That’s how we began. And with the recent AI wave, we’ve seen growing openness among Christian organizations.
James Hwang: Your company has developed an intelligent Bible search engine that aligns modern translations with the original languages. How does this meaning-based approach help pastors or Bible students, particularly those who may not know Greek or Hebrew?
Dr. Zhu: When it comes to Scripture, accuracy is paramount. Our AI work usually happens behind the scenes—we build tools that other organizations use. And when we build those tools, we aim for 100% accuracy. That’s possible because the Bible is a finite text. It’s not like the news, where something new appears every day. So, we can pre-search all the possible relationships between verses.
We let the machine do the laborious task of comparing everything, and then our team—highly qualified Christians with deep biblical knowledge—cleans up the results. It’s like making a concordance in advance. That human-in-the-loop element ensures theological integrity.
For instance, when we align a modern version with the original Greek or Hebrew, we might use older AI models, not the latest neural nets. Then our team reviews and refines the alignment. This makes the original languages more accessible—even if someone doesn’t know Greek, they can still grasp 80% of the meaning through our system.
In one prototype, for every Greek concept in a verse, you can click a word and instantly explore related contexts across the Bible—even if the exact word isn’t repeated. Our goal is to find patterns in grammar and thought—maybe Paul uses a rare word, but the same mental structure appears somewhere else in Scripture. That’s the kind of connection our algorithms are designed to surface.
James Hwang: Fascinating. Many ministries are exploring chatbot technology and vector embeddings. But often the knowledge base is too small, and hallucinations happen. How can Christian organizations prepare their data for more accurate AI applications?
Dr. Zhu: First, we need to understand the limitations. If you train a vector embedding on just 75 Christian books, that’s not enough. Vector embeddings aim to capture the totality of human knowledge in mathematical form. Without massive data—think World Wide Web scale—you’ll get sparse, unreliable results.
Some pastors ask us to build a chatbot based on their personal sermons. That’s okay for personal use. If it’s just for him, and no one else is likely to use it, he can try tools like Perplexity, which allow you to upload documents and perform semantic search. That’s fine. But building a public-facing chatbot is another matter. You’d need scale and accuracy—and that costs money. Often, we advise churches not to build unless they know it will truly be used.
If you really want to serve the global church—say, through all Chinese Christian literature—then you should integrate with existing AI platforms. Don’t reinvent the wheel. Pay for a platform that already does 90% of what you need.
James Hwang: And if they do want to build, what’s the next frontier? How do ministries make their resources agent-friendly?
Dr. Zhu: We’re entering the era of AI agents—tools that can act on your behalf. You give it a task, and it executes it—summarizes an article, fetches data, even buys something online.
For these agents to work, they need a shared protocol. That’s where Model Context Protocol (MCP) comes in. MCP is like HTTP for the AI age. Just as websites talk to each other via HTTP, AI agents use MCP to talk to each other and to data sources.
So if your organization wants to be visible to these agents—whether they’re helping seekers or researchers—you need to make your content MCP-accessible. It’s like having a website in the 1990s. You don’t need to know the technical details yourself, but you need someone on your team who can get you on that new web.
And here’s the key: let the technologists lead. Pastors should clarify the goal—spreading God’s word—but let the tech folks figure out how to get there. Too often, we’ve seen well-meaning leaders dictate the tech stack without knowing the constraints. That’s risky and delays progress.
James Hwang: Hallucinations are still a concern. What practical advice do you have for avoiding them?
Dr. Zhu: For starters, if you’re using a large language model (LLM), you can prompt it by saying, “Only tell me what you’re 100% sure about and show me the source.” That alone can dramatically reduce hallucinations.
But also, we need to recognize that the fear of hallucination is often exaggerated. Many pastors haven’t explored AI deeply. They hear fears secondhand. Most LLMs today don’t try to be the final authority. If you ask a theological question, they’ll summarize the major perspectives: “Catholics say this, Protestants say that.” It’s like having a balanced research assistant.
That’s where retrieval-augmented generation (RAG) comes in. You ground the AI in your own content—sermons, books, or curriculum—and the model retrieves relevant passages before generating a response. It’s far more reliable than vanilla prompting.
Still, my personal advice is this: don’t just use AI for search. Use it to think. Treat it like a team of PhD assistants. Don’t waste their brainpower fetching things you already know—ask them to help you see what you can’t yet see.
James Hwang: That’s powerful. So we should use AI not just to retrieve answers but to stretch our thinking.
Dr. Zhu: Exactly. Use it to explore outside your usual categories.
James Hwang: Looking ahead, what future application of AI could most advance the mission of the global church?
Dr. Zhu: If I were advising a group like the Lausanne Movement, I’d tell them: Make sure every ministry worker has access to a paid AI account. Don’t rely on free tools with limited features. We have amazing volunteers and coworkers who are already committed to God’s kingdom—give them the tools to multiply their creativity.
The steam engine equalized physical strength. The internet equalized access to knowledge. AI is the next equalizer. Someone who wasn’t great at memorizing verses might have been overlooked by seminary culture. But maybe that person has insight and vision—something the Spirit wants to release now. With AI, they can start a new ministry or reach a new audience.
So empower them. Let them try things. Maybe it’s not conventional. But remember—AI doesn’t have the Holy Spirit. People do. Our job is to equip them and get out of the way.
James Hwang: A couple of years ago, Harvard Business School’s Karim Lakhani said, “AI won’t replace humans. But humans using AI will replace humans not using AI.” What do you think?
Dr. Zhu: I agree. You can live a peaceful life without technology. But if you want to make an impact, you need to be where people are. AI isn’t just about speed—it’s about reach and stewardship. In ministry, we’re serving people. If AI helps you serve ten people instead of one, that’s worth it.
So, my word to Christian organizations is: don’t be afraid. Embrace technology. Trust your people. And keep asking: What new thing is God doing in our time?
