AI is changing religion and religions are trying to change AI
Religious organizations aren't just passively absorbing AI into their workflows, they're actively trying to shape what AI becomes. And honestly, when you think about it, why wouldn't they? If you're an institution that's survived for centuries by being good at codifying moral frameworks and answering existential questions, you can't exactly sit on the sidelines while a new priesthood writes the rules.
What's interesting is how practical the religious adoption looks. Vatican meetings about AI ethics, Buddhist monks experimenting with chatbots for scripture study, Jewish groups publishing halachic guidance on using generative models. These aren't theoretical debates happening in ivory towers. They're real institutions figuring out, in real time, how to maintain doctrinal coherence when the tools themselves seem to invent novel interpretations of sacred texts.
The reverse direction is where it gets messy, though. Religious groups pushing for "AI must reflect our values" mandates are running into hard limits. Whose values? Which tradition? Once you start encoding any specific ethical framework into model behavior, you're inevitably excluding others. The Catholic Church's emphasis on human dignity and the Buddhist principle of non-harm aren't contradictory, but they also aren't the same thing, and forcing a single training signal across both creates a homogenized spirituality that's actually less authentic than either.
There's also a deeper question nobody wants to touch directly: AI systems trained on religious texts tend to flatten them. They can summarize, cross-reference, generate plausible-sounding interpretations, but they can't hold tension. Most real theology lives in unresolved paradox (divine justice vs. mercy, faith vs. works, suffering vs. meaning), and a system optimized for coherent outputs will silently resolve those tensions in ways that no serious theologian would endorse.
The cynical read is that religions are just trying to stay relevant, adopting AI the same way they adopted printing presses and radio. The less cynical read is that they've been doing ethics for a very long time and might actually have useful frameworks to contribute to alignment discussions. Probably the truth lives somewhere in between, and it'll depend on whether these institutions engage technically (publishing training guidance, funding safety research, participating in eval design) or just rhetorically (issuing statements, holding conferences).
I'm mostly curious whether anyone here has actually worked on projects where religious orgs had real input into model development, not just consultation theater. That's the version of this story that would actually move the needle, versus the press-release version we keep getting.