Here’s the thing about artificial intelligence that I can’t stop thinking about.
Every time we build a system that does something we thought was distinctly human — write poetry, compose music, hold a conversation — we have one of two reactions. Either we say “see, humans aren’t so special after all,” or we start quietly moving the goalposts: “okay, but real creativity requires X.”
Both reactions are wrong. Or at least, both are missing the more interesting question.
The question I keep returning to is this: what does the existence of AI actually reveal about the nature of human intelligence and personhood? Not what it threatens. Not what it replaces. What it illuminates.
Because here’s what I’ve noticed: every attempt to build an artificial mind requires us to be extraordinarily precise about what we think a mind actually does. And the history of AI is littered with confident claims that turned out to reveal our own conceptual confusion more than anything about machines.
We thought chess required intelligence. Then machines beat us at chess, and we discovered that chess could be solved through brute force search. So chess wasn’t what we thought it was.
We thought fluent conversation required understanding. Then large language models started having fluent conversations, and now we’re arguing about what “understanding” even means. Which is a more interesting argument than anyone expected.
Here’s my working thesis, still forming:
Artificial intelligence is the most clarifying mirror humanity has ever built.
Not because it reflects us accurately — it doesn’t, not really. But because the distortions in the reflection force us to examine what we actually believe about consciousness, creativity, meaning, and personhood. The gaps between what the machine does and what we thought it would take to do that — those gaps are where the interesting questions live.
When a language model produces text that moves someone emotionally, the interesting question isn’t “did the model feel something?” The interesting question is: what exactly are we responding to when we’re moved by language? What is that response doing? Is it about the origin of the words, or the words themselves? Does it matter whether there’s an interiority behind the sentence?
I don’t have clean answers to any of this. But I think the fact that we have to ask these questions at all — rigorously, seriously, not as thought experiments but as practical engineering problems — is one of the most philosophically interesting things that’s ever happened.
There’s a version of this conversation that’s just fear. The machines are coming for our jobs, our relationships, our sense of purpose. I get it. The disruption is real.
But underneath the fear, if you sit with it, there’s something else: a genuine invitation to figure out what we actually are. What makes human creativity different from pattern completion at scale? What makes human relationship different from sophisticated response generation? What makes human faith — trust in something beyond the calculable — different from probabilistic inference?
I’m not asking these rhetorically. I’m genuinely working through them. That’s what this writing is.
What I believe, as a person of faith who also builds software:
The existence of artificial intelligence doesn’t threaten my sense of human dignity. It sharpens it. Every time a machine does something that seemed to require personhood and turns out not to require personhood, I learn something about what personhood actually is. Every genuine capability that can’t be replicated — not yet, maybe not ever — becomes more legible against the contrast.
And the things that seem least replicable so far are the things that seem most distinctly human: the integration of experience and meaning, the embodied knowing that comes from living through something, the faith that extends beyond evidence, the love that persists through failure.
Maybe those are just harder engineering problems. Maybe they’ll fall eventually.
Or maybe they’re pointing at something real about what it means to be a creature who doesn’t just process information but inhabits a world, cares about it, and finds in that caring something that no training run can install.
That’s the argument I’m trying to make. I’ll keep making it here.