Why we said no to AI questions in BIBL's free pack

Three weeks before launch, our QA lead flagged something in our Slack that made me pause. A user had spotted a question in an early build that was technically correct but spiritually off. The verse was real. The answer was defensible. But it missed the point of the passage entirely. That's when I realised we had a choice to make.

The temptation was real

When we started building BIBL, we knew question quality would make or break the game. Bible trivia isn't like general knowledge trivia. People who play this game often know the scriptures better than most. They spot inconsistencies. They catch shortcuts. They'll forgive a typo; they won't forgive a question that twists what a passage actually says.

We could have scaled faster. A team of three could generate hundreds of questions in a week. Run them through verification layers. Ship them. Move on. Plenty of apps do this. We looked at the cost benefit pretty seriously. The Genesis Pack, which every player gets for free, needed 100+ solid questions ready on day one. That's not nothing.

But every time we looked at the shortcut, we kept coming back to the same problem. The technology works well for filling in blanks and generating variations on safe topics. But Bible trivia has teeth. Ask a question about what Jesus said to the Pharisees, and the wording matters. The context matters. A machine can tell you whether it's technically scriptural. It can't tell you whether it captures the actual weight of the moment.

What went wrong (and right) in testing

We tried anyway, to be honest. Early on, we generated a batch of questions and sent them to a small group of people we trusted, mostly church leaders and Bible study folks. The feedback was consistent. Some questions felt academically correct but spiritually hollow. Others had the right idea but the wrong emphasis. One question asked players to identify a verse by its description, but the description was so generic it could have matched three different passages.

What surprised me was the tone of the feedback. People weren't angry. They were disappointed. One person wrote: "This feels like you're treating the Bible as a trivia book rather than scripture." That's when we made the call.

We decided that every question in the free Genesis Pack would be written and verified by humans. Not because we're Luddites or because we're against efficiency. But because the people downloading this app deserve questions they can trust. They're not playing for points. They're playing because they love these stories and they want to test themselves against that knowledge.

The Basic tier is different, and that's intentional

Here's the nuance, though. The Basic tier (Kingdom Pack) does include a limited set of AI generated questions, alongside human verified ones. We made that call deliberately. It's a paid tier. Players have made a choice to engage deeper. They're willing to explore new content, including questions that might be generated and refined rather than hand crafted. That felt like a fair trade.

But the free Genesis Pack? That's your first handshake with us. That's where we set the tone. And the tone we wanted to set was: we care about accuracy. We care about integrity. We're not cutting corners here.

What it actually costs

The honest answer is it costs us time and consistency. Our small team spends cycles writing and verifying questions. We can't generate content as quickly as we might if we gave a machine the task. We have to think about each question individually. Does it work for a competitive game? Is it fair? Is it true to the passage? Does it land?

That's also why the Genesis Pack is smaller than it might be, and why we release new question packs incrementally rather than flooding the game with content immediately. It's a constraint we've chosen.

But it's also why, when someone plays BIBL for the first time, they're not playing against a version of the Bible that's been smoothed and genericised by an algorithm. They're playing against the actual text. They're competing in knowledge that actually matters to them.

The thing nobody asked for

What's strange is that nobody complained about the decision. We didn't have a crowd demanding AI questions. We had the opposite problem, really: people telling us they appreciated that we weren't taking the easy route. One person in our testing group said, "It's nice to know someone actually cares about getting this right."

That mattered to me more than I expected it to. We're not a big studio. We're not chasing viral adoption or funding rounds. We're trying to build something that's worth the time people spend on it. That doesn't sound like a business strategy, and it isn't one. It's more basic than that. It's just how we want to build things.

The Genesis Pack isn't perfect. No question pack ever is. But it's ours. It's thought through. It's real.

If you're building something where trust matters, does taking the long route ever become the competitive advantage instead of the liability?

Ready to try BIBL by MRVL?

One tap to download. No sign-up wall.

Get it on the App Store Get it on Google Play

Want to try Bibl?

Visit Bibl →