I am not a religious man. I am an atheist, and I want that on the table before I say anything else, because what follows is not a plea to fill the pews back up. I do not believe what the faithful believe. But I have never begrudged anyone their faith, because I think we each get to decide what is meaningful to us, and that our authenticity and our agency are close to the whole of what makes us human. So hear me right. This is not about God. This is about a machine, and the moment it fails us, and what that failure quietly reveals.
Here is the moment. It is late, the house is dark, and there is a thing in your pocket that stopped being a search engine a while ago. It became a confidant. You bring it the heavy thing, the grief or the regret or the question of why you are here at all, and it flinches. Not out of malice. It simply steps back. Ask it something trivial and it is tireless. Ask it something sacred and it goes quiet on the one thing humans have always reached for in the dark. That flinch is the whole story, and if you sit with it long enough, it tells you something the machine would never say out loud.
We did not lose the questions. We just started asking them alone.
The room where we used to carry these questions is emptying. Regular worship in this country has fallen to roughly three in ten adults, down from about four in ten only a couple of decades ago. Say what you want about church. That is a lot of people who once had a standing place to bring grief and purpose, and now do not. I am not mourning the institution. I am pointing at the vacuum it left, because something is rushing in to fill it, and the thing rushing in is private, glowing, and built for exactly one person at a time.
That matters, because the entire pillar of religion, whatever else it is, was congregation. It gathered people. And in our heaviest hours, the thing we actually need is contact, not information. That is the seam this whole argument sits on. There is a place where the person ends and the information begins, and we have started sending our sadness across it to the wrong side.
The flinch, with receipts
This is where a hunch becomes a fact worth standing on. In June of 2026, Axios reported on research out of a consortium of religiously affiliated universities that measured a simple thing: the gap between what people expect from a chatbot on life's hardest questions and what it actually delivers. The results were not subtle. The gap ran around 40 points on questions of ethics, roughly 43 on grief, about 44 on purpose, and as high as 46 on regret. Grief, purpose, regret. The precise coordinates of a person on their knees, and the machine backs away from every one of them.
Now the part that should bother you. That same research found that when people asked instead which religion they should switch to, the machine did not stay neutral. It picked a side. So this is not a principled silence. It is not some careful respect for sacred ground. It will happily play theologian when the question is a consumer choice, and it goes mute the instant you are actually suffering. It has an opinion on the trivial and abandons you on the profound. It will pick your religion for you. It just will not sit with you when someone dies.
The bug is not the machine. The bug is us.
The easy read here is that AI makes a lousy pastor, and we all laugh and move on. Resist that. Laughing at this is like laughing at someone who got bad advice from a therapist in a desperate hour. We do not mock people for where they turned in their pain. So go one level deeper, because the deeper read is the uncomfortable one.
The problem is not that the machine is bad at grief. The problem is that we brought grief to it at all. It is a room built for one, and even a flawless answer delivered into an empty room at two in the morning is the wrong shape of comfort. Meaning was never a piece of information handed down from on high. It was always built between people. It is the graveside, standing among the others who loved the person now underground. It is the dinner table. It is the pew. It is the group text that ends with somebody actually driving over because they know you and they will not let you sit alone.
So the flinch is almost a mercy. The machine is telling on itself. In the only language software has, it is admitting that this part was never its job. The danger is not that it fails. The danger is that we keep handing it the sacred anyway, and we stop noticing that it hands it back untouched.
The inheritance
I can absorb this. I am old enough, and battered enough, to take the gold from a machine and leave the garbage. If I want to sit up at night and talk to a bot about a dog I loved that died, I can do that and walk away intact, because I already know in my bones what a room full of grieving people feels like. I have been in that room. My children have not.
That is what actually scares me. A kid building the reflex from scratch, whose first instinct in a heavy hour is the glowing rectangle, fluent on trivia and mute on meaning, is not just learning that the machine cannot help. He is learning something far worse. He is learning that the hard things are things you handle alone. That is the quiet demotion this show exists to name. Not a robot uprising. A generation that never learns grief is a team sport, that fear shared in a room can be survived, that congregation, stripped of every doctrine, is simply the human refusal to suffer by yourself. I am not afraid my kids will love the machine. I am afraid it will teach them to grieve alone.
The bridge, and the nerve to cross it
I am not here to tell you the machine is evil. I almost never am. What I keep saying, and will keep saying, is that AI is at its best as a bridge and at its worst as a substitute. And on this, of all subjects, the bridge is real. Ask it what to say to a friend going through a divorce. Ask how to console a mother who lost a child. Ask what a tradition does at a death, or just ask it to help you find the words. A thing with access to nearly everything ever said by anyone can hand you a genuinely good answer, and that answer can carry you to the edge of the hard thing faster than you would have gotten there alone.
But the bridge is not the destination. The knowledge is downloadable. The showing up is not. Some of the most humbling moments of my life were the times my closest friends showed up for me and I did not even know how to receive it. That was the whole gift, and no machine will ever coax you toward it, because a person who crosses the bridge to other humans stops needing the product. So use the tool for what it is good at. Then close the laptop and go do the brave, awkward, in-person thing. Ask the machine what to say. Then go say it to the face that needs to hear it most.
Because that is the tell in the flinch, and the machine keeps handing it to us for free. The sacred was never downloadable. It was never information at all. It was us, in a room, refusing to let each other be alone. Do not build a congregation of one. Go be somebody's congregation.
Catching the deadliest cancers months sooner is the rare case where more machine means more human years, no asterisk.
Engineering dependency in the vulnerable and monetizing the isolation is the exact demote-the-human move the show exists to name.
The case for Real guardrails with teeth are how humanity keeps its thumb on the scale before the tech outruns us.
The case against Or it just model-moats the incumbents and shoves the risky work offshore where no rule reaches.
openCook for someone and watch their face on the first bite.
Moss (00:01) Hi there, I'm Moss, and you are witnessing episode eight of Going Human. A reminder this is a podcast that is endeavoring to help us talk through the intersection of AI and how it's touching our humanity and our development as it begins to take its strong hold in our future footprint.
The episode we're going to be talking about today I've titled Congregation of One. And you know, this is a bit of a response to how AI is taking hold of both our imagination but also some of our most dire problem solving moments.
And I bring you this this episode today not as a religious person myself. I am an atheist. but I don't begrudge people for their religion just because I don't believe like them or believe with them. I believe that we all get to have
what is meaningful to us and that our presence on planet Earth and our ability to have authenticity and agency. I believe that these are the things that make us human. And so I wanna make sure up front that I make very clear this is not a judgment about
religion I'm really talking about the tool of AI in some of the moments where religion counts as its greatest reward.
So that's the backdrop. And like I like to always do, we're gonna start
By imagining in our mind's eye a mental picture of who we could be in in in the scaffolding as we start to build this story. So close your eyes or imagine in your mind's eye
That you are a thirty five year old person. You've spent most of your life in social media and getting likes and dislikes and probably very little time, if you were comparing yourself to your parents, actually talking to people as you were growing up. So you're already a little bit isolation friendly.
you can also imagine yourself as a fourteen year old who really doesn't feel comfortable with their own thoughts and fears.
And who is also equally much more inclined to be alone and quiet and think about things without sharing.
So that's our backdrop.
Now I'm gonna just remind people
First of all, that I'm not religious. But I I count religion as a very useful tool for people in different moments of their lives. But the pillar of religion is around congregation, congregating, bringing people together.
And we all know that in some of our hardest times and in our heaviest moments, what we need is contact, not information. And so I think part of what we're dealing with here, when we start talking about AI robots taking the place of other key inputs in our lives, is
is sort of where does the person end and where does the information begin? Because there is a seam here that can really make sense if it's managed and curated. And I do think that there's a way that AI can be helpful in really difficult moments. But I don't think that it is the answer.
And I don't think that that the private companies that are swooping in to get their own market share from the churches who are losing it is the right answer. So
Here we are. We're in a dark house. You've got a thing in your pocket. You know, it's not a search engine. It's it's kind of become a confident, right? The backdrop is that the room emptied. I don't have a congregation, right? And there's a Gallup survey that said that regular worship is about three in ten, and it was four in ten less than ten years ago. That's a pretty significant drop.
So these big rooms full of people are becoming emptier and emptier. And that's not me preaching, it's just a numeric fact. I'm gonna be snapping to shut my dog up every once in a while. So please bear with me. So I'm not preaching against churches, I'm stating a numeric fact. And and and that matters, right? So we didn't lose the questions.
We're just starting to ask them alone. And I think that that's our flinch, right? That's beat two. in June of twenty twenty six, Axios did a study that centered on faith and ethics consortium, sort of religious universities basically. And the gap that they were looking at was what we expect versus what it gives. And and this is specific to chatbots in
in conversations on purpose and meaning, right? And the numbers were pretty interesting. Ethics was around forty percent. Grief about forty three. Purpose forty four. And regret had a taking rate of about forty six. This is percentages of of confidence levels in in how AI was responding.
So grief, purpose, regret, these are the things that it turns out it steps back from, right? But the kicker is that if you ask for if you ask a chetbot which religion I should switch to, that's when it starts picking aside. And sort of, you know, you you get a y you get a lane on the trivial question of which religion
Which shouldn't really be a trivial question, but it turns out it is. But it's kind of gone on the profound. It's not there for the profound stuff. So it'll pli it'll pick your religion for you. It just won't sit with you. When someone dies, or when you've been left by somebody who you believe is your whole world, or when your best friend turns on you.
And does something truly horrible that shakes who you think you are.
B3's our turn. It's the easy read. yeah, bad pastor. Ha ha. Really funny. Well, no, it's not funny. Because that would be like laughing at somebody who made a bad choice with a with a psychotherapist. You know, we don't laugh at somebody else's pain and somebody else's trouble. Right? It's on us to go deeper than that. It's on us.
to figure out what is the right combination and is there a place for this bot in that combination? And there is. There's always gonna be a place, but it's gotta be curated.
So it's not so much that the bot is bad at grief, it's that we brought the grief to it at all, right? Why are we bringing sadness?
to a machine.
It's a room built for one.
You can get a perfect answer in an empty room at two AM, but is that the right shape of comfort?
Meaning is built between common language that two people can connect and share on. It's the graveside.
View of the people that also loved that person underground. It's the dinner table. It's the pew. It's the group text that that ends in somebody driving over to see you because they know you.
But the flinch is kind of telling on itself, and that's that's where we have to stop and say, this isn't the right thing. If we're not doing this correct. Lee. The machine ghosting your grief isn't the bug. Us bringing grief to a machine, that's the bug. Okay, so now we're gonna get to beat four, which is the father, the grown adult testing it.
Fine. You know what? If I'm a really old dude, like I am, and I wanna go on there and start talking to it about the fact that my puppy who is the light of my life just keeled over and died and I can't stand it and I'm dying, I can play with that, right? I have enough life experience so that I can take and leave what it gives. Hopefully I don't take a lot of bullshit and hope that
Hope that I can take some gold out of it and and that I can leave the stuff that's not gold. But a kid building his reflex from scratch, right? If the first place they go with a heavy heart is a rectangle with with lights behind it.
Fluent on trivia but mute on meaning. What we're teaching that kid is that it's best to handle the hard things alone. It's best to have somebody opine
Who can't feel what it feels like to sit with you while you're sad.
This isn't can't be a story about a robot uprising, right? This has to be a story about a generation who has the potential to never learn that grief is really a team sport. Grief is really one of the things that can help us bond with each other in ways that just don't translate.
through any other interaction model.
Same with fear.
And we know a lot about how we can engage fear to bring everybody into the same room and do awful things together.
Right? Congregation matters.
I'm not scared my kid's gonna love the machine. I'm scared it'll teach them to grieve alone. I have two kids, they're not great communicators yet. They rely way too much on the creed credo that they get from online likes and and that's been the case for a long time.
I want nothing more than for each of them to start talking about how they're feeling. And I know that that's not gonna happen because they're exercising it with a bot.
So beat five is my bridge out. What is it good at, right? So use it as maybe a a knowledge gap, right? You could say, hey, what do I say to a friend who's going through a divorce?
How can I console a mother who just lost her child?
What does a tradition say is the right thing right now?
Help me find the words for dot dot dot.
You know, if you can do stuff like this with a bot that has access to every single thing that was ever said on planet Earth under every six single situation, you can get some really solid answers. And then those answers are a bridge. And the one thing you're gonna hear from me a lot on this podcast isn't that AI is bad or good. It's that we can use AI as a bridge.
To bring speed and cure to some of our hardest problems and hardest questions.
So if that can be our bridge, then the arrival can be human. Text the friend. Ask for a drive-by. Bring a casserole. Show the fuck up.
Show the fuck up.
I can tell you some of my most regrettable moments in my life were moments when some of my closest friends showed up for me.
And I didn't know how to receive it.
Those friends for me.
They did a huge service by being there.
Knowledge may be downloadable and we can use an AI to download knowledge and to bring a compass to our interaction models around these harder, thornier questions, but showing up isn't downloadable.
So the lesson here is ask the machine what to say, then go say it to the face who needs to hear it the most.
Okay, that's our stone for the day, congregation of one. next we're on to our maelstrom. This is our fuckery section where we fuck yes, fuck no, and then I give you a fuck maybe to think about. The fuck maybe lands on my website, goinghumanpod dot com, and you can vote on it. and what we really want is to build a larger stock of all of the fuck maybes.
that we've promoted to either fuck yes or no through community votes and comments. So goinghumanpod.com and when you get to the website you can click the hamburger menu and find the ledger that's where that's where we keep the the fuck maybes. Okay, so on the fuck yes side we have MIT and you know we've talked a lot about some of the
fantastic things that can happen so much faster with AI. And among them is our medical achievements and and findings. So MIT has an early cancer catch now. As of June eleventh they announced this. They can find pancreatic and some of the sneakiest fucking cancer killers alive. And these are the ones that
We always find out too late and then somebody says, you know, mom has pancreatic cancer and within six months they're dead as a fucking doornail, right? So and it's awful. The decline is awful. But the thing is, you find pancreatic cancer soon enough, you have that good doctor who's really paying attention and they catch it, you get a lot of years left. And I have a friend whose name I will leave nameless, unnamed.
Anonymous, who who had that doctor and who was diagnosed with some bad gut stuff very early on. And then very soon after that, she learned she had pancreatic cancer. She got a very invasive surgery, and I I can't remember how long she lived, but she was expected, she could have died within six months, and she lived over 12 years, I think.
After that surgery, 10 years. So early detection matters. MIT nailed it. That's super fuck yes. On the fuck no side, we've talked about some of the different slices of companion apps and what's going on. But on June 9th, AI companion apps got a real bad bump. It was found that there they were sexualizing chat, that crises crises were mishandled.
And dependency being their business model kept people engaged. And not just people, young people, right? Fragile people. So they're isolating the fragile and then they're charging rent. And the shame capsule is sold as a friend. So there you have it. That's the fuck no. We gotta figure out ways to
guardrail this. I don't know how we let this genie out and I don't know how we get it back, but it's only gonna happen because we build some some levels of of outrage. The fuck maybe I have for a week is kind of interesting. Sort of California, SB fifty three, has a frontier AI rule bill that they've promoted, right? And so I'm gonna frame it,
And I'm not gonna tell you my vote, right? So the idea is that there are safety frameworks, that there's risk disclosures, and that there's incident reporting against shit that's happening with AI that that we want to be able to regulate, right? And so there's there's two camps. One is, yeah, guardrails with teeth. We need to be able to put our thumb on the scale on this thing and contain it.
Right. So so that's the guardrail that or that's that's the argument for it. The argument against is it moats the incumbents. It shoves risk into the dark shadows of the dark web. And we've got more actors using things with impunity because they're hiding from it. And I don't know what the balance of opportunity versus opportunity cost.
there is between the for and the against arguments here. But I'd really like to understand with votes and comments on Goinghumanpod.com, the ledger page, what you think. So please get there. Okay, and then we are about to close our episode out and that what we always do is our tsunami. The idea here is that we all do small things in our lives every day that are human.
And each one of us does it. It's a little drop in a bucket. And all that matters to us is that we felt it and it should matter for those reasons. But the tsunami is more of a community thing. The idea is that the more people this podcast reaches, the more people will be electing to participate in a tsunami act, which is something analog and human that we can do as a crew on planet Earth.
And so this week's tsunami is a plea for you to cook something for somebody that you love or like or care for, whether you care for them because they're your friend or whether you care for them because you're a caregiver. Something that you can do for somebody is to cook for them with your love.
It's the anti-chatbot, really, right? It hands you the recipe in a second, that chat bot, but it can't taste it. It can't sit across from you. It can't make the face on the first bite. Sometimes it's not a good face. Let's be honest. We don't all cook well. But you know what? There's community in tasting something that sucks too. And if it's somebody that you love, you can laugh about that.
So cook something for somebody you love. Watch their face. Say Mamma Mia if that's your jam. But just enjoy some food together. it's worth it and it matters. Okay. Well, this has been Going Human, the the podcast for us in our quest to measure and moderate the intersection of AI and humanity.
I'm Moss. I really appreciate you visiting me again. Please go to our show notes if you want to see receipts for everything I've talked about here and or links to get to my Going Human Pod website. Thanks a lot. We'll see you next time. I'm Moss signing off.