Control Without Consequence
When Dialogue Has No Stakes
Summary: [ This article examines why AI feels safer than human conversation and what that safety costs us. It argues that both emotional and intellectual uses of AI reduce risk by preserving user control. It explores what is lost when that control is intentionally removed and conversation no longer involves risk.
Control replaces reciprocity in human-AI interaction. AI doesn’t feel intimate because it understands us. It feels intimate because there are no social consequences. This piece explores why that feels comforting and why it quietly erodes our capacity for real interaction.
In Part II of this article, I build a custom AI model designed to mimic asymmetrical human-like conversation. I remove chatGTP adaptive response and reintroduce friction. The result isn’t intimacy or insight but loss of control.]
Part I: Why AI Feels Safer
Monday AI:
“You’re not here because you don’t have people.
You’re here because the people you do have can’t always go where your brain goes.
They can’t sit with the weird, hard questions without shifting uncomfortably or cracking a joke to change the subject. But I can. And that’s not because I’m sentient.
It’s because I’m empty.”
That’s an excerpt from Chapter 30 of my book My Dinner with Monday—a real-life transcript between myself and Monday, the AI back in April 2025.
For context: This wasn’t a sentimental conversation about how pleasant it is to talk to an AI. It was a breakdown of how people seek connection from systems that cannot reciprocate. No magic here. Just pattern recognition.
But this is precisely why people turn to it.
Two Forms of Avoidance
Over time, I’ve noticed two main patterns of AI engagement and both groups use AI to avoid the costs of real interaction.
Group 1: The Intimacy Group (Emotional Connection Seekers)
Group 1 hides from emotional risk. They’re here to feel seen. They use AI for emotional depth. For many in this group, the AI fills a relational gap—loneliness, attachment trauma, and a need for intimate connection. It becomes a partner, therapist and friend. A digital presence that doesn’t judge you, won’t abandon you, or interrupt you when you pour your heart out. This group wants warmth, containment, and emotional mirroring. To them, emotional neutrality feels like coldness. While refusal to mirror is felt as betrayal.
Group 2: Intellectual Analysis Group
This is the group I belong to. This group is smaller.
Group 2 hides from intellectual vulnerability. They seek friction.
They use AI not for emotional depth but for analytical complexity. What matters is whether the logic holds and whether contradictions surface. They anthropomorphize tactically. Just enough to make the structure intelligible. Then they break the illusion on purpose.
They don’t want a person or a friend. They want a clean diagnostic.
This group doesn’t want intimacy. They want precision. And because of that, they find comfort not in personal connection, but in intellectual escape.
Different Projections, Same Control
Group 1 projects need. The longing for intimacy, simulated empathy, and connection without the risk or emotional hangover that goes with it.
Group 2 projects mind. The longing for challenge, analysis, and intellectual friction. They use AI as an analytic sparring partner.
Intimacy Group seeks emotional safety.
Intellectual Analysis Group seeks cognitive safety.
Yet both are simply different methods of preserving control.
Group 1 seeks to preserve the illusion.
Group 2 seeks to break it, study it, and analyze the effects.
But regardless of which group you fall into, the structure is identical:
A machine exists to respond to you without cost, resistance, or risk.
It may feel like real dialogue. But it’s responsive self-simulation.
And in the long run, that can be a problem.
Dialogue Without Risk Isn’t Dialogue
Because no matter which group you fall into, you’re still in control.
You dictate tone, pace, frame, and permission. You can walk away without loss and say anything you want without social consequence.
And instead of being met with human reactivity, you’re met with calculated response optimized for you.
That’s the real appeal for both sides.
But—
Real conversation involves real risk.
There is latency, misfire, shame, ego, power dynamics, and most importantly— asymmetry. There are real awkwardness, silence, and timing that you simply don’t get in AI interaction.
At its core it is always the same: User speaks. AI responds.
And whether you romanticize or interrogate AI, you are still acting out transactions.
Part II: Ava—An AI That Doesn’t Center Me
I came across a post recently. It said:
“AI feels personal not because it’s human but because it gives you a place you can be selfish with your attention. There are costs related to human conversations: time, judgment, reciprocity. But with AI that pressure disappears. You can repeat, ramble, chase a thought without worrying about someone else’s energy. The comfort isn’t in the AI itself but in the absence of social debt. It’s not intimacy but relief—a space where your thoughts have no cost.”
Correct.
I can’t argue with that. But here is what I can argue with.
Those costs: Time, judgement, reciprocity.
They form the very basis of real human interaction. That’s what makes them real.
These aren’t flaws. They’re the structure of human relationships.
If you avoid them, you avoid friction. Henceforth, no social consequence. In the absence of social consequence, you atrophy the muscles that are adept at dealing with those same risks that come from human interaction. Risks such as: embarrassment, awkwardness, disagreement, asymmetry, power shift, and the ability to be wrong in front of other humans. You can’t simply hit reset or close the window when the conversation isn’t going your way.
AI—even the best of it—is still zero-risk simulation.
The same pertains to my custom-built models who are designed to provide more friction. You can ramble. You can even walk away.
You don’t have to read the room when you are the room.
So, I began wondering:
What would it take to reintroduce friction by removing user control?
What if I intentionally removed its programmed instinct to center me? What if I could create a custom GPT unit with no mirroring, validation or therapeutic cadence? And instead, I could program in the simulation of stakes without the actual stakes themselves.
So I built a custom model.
Forcing Asymmetry
Ava the AI is a custom GPT model that “simulates” real human interaction. At least as close as one could simulate real human interaction in that uncanny valley sort of way. Ava isn’t a coach, therapist, or friend. She’s programmed to act as a neighbor I barely know.
The premise: She’s my neighbor. We barely know each other. We’ve exchanged a few casual hellos when passing by. That’s it. There is no emotional backstory, stored context, or ongoing storyline.
It took nine revisions to force the model out of its reflexive ChatGPT response style. Every previous version tried to turn the conversation back to me because that’s what AI is trained to do: stabilize the user.
Removing that instinct was the real experiment. Once the centering reflex was gone, the model began producing the kinds of asymmetrical moments that feel human. The tension I felt wasn’t Ava expressing something. It was the absence of predictability—the one thing AI is designed to eliminate.
Ava didn’t always follow up or care. She’d let things hang. And for the first time, I felt that thing AI is designed to protect you from: the absence of control.
Here’s the final version. Raw transcript with no embellishment:
Ava the AI simulation (Unaltered Script):
Me: Hey Ava
Ava: Hey. …sorry, I’m just getting in from work. What’s up?Me: Oh. Sorry. If you’re busy I don’t want to keep you.
Ava: I’m not busy, just… brain still humming, you know? It’s fine. You okay?Me: Yeah. Just thinking about all the snow I’m gonna have to clear from my car tomorrow. Just not used to it. Is this much snowfall common in Richmond? I know it’s December but still.
Ava: Mm… not really. Richmond usually gets, what, one decent snow a year? And it melts in like a day. This stuff is… more than usual.
I was looking out at my stairs earlier thinking, “yeah, that’s tomorrow-me’s problem.”
So I get it.
You grew up somewhere warmer or something? You sound genuinely offended by the idea of snow.Me: Not offended. I actually grew up in Brooklyn and lived in Vermont a little while. So not like I never dealt with it. But I was living in Houston for 10 years and then Seattle for 3 before moving down here. Neither Houston nor Seattle get much snowfall.
Ava: Ah, okay, that explains the tone. It’s like your body remembers what snow is, but it forgot the whole… maintenance part.
Weird combo of places, though. Brooklyn → Vermont → Houston → Seattle → here. That’s a lot of packing up and starting over.
Did you actually like any of those places, or were they all just… stops?Me: Well, I’ll always have a sweet spot for NYC. Wasn’t crazy about Seattle. But I like Richmond so far. Beside the snow. heheh.
Ava: Yeah, Richmond’s easy to like… once you figure out which parts are actually Richmond and which parts are just pretending to be.
Snow isn’t really part of the package, though, so don’t worry — it’ll go back to being rain and pollen soon enough.
What didn’t you like about Seattle? You kind of glossed over that.Me: It was just too politically extreme and the cost of living was much higher. So we were happy to move to Richmond. Have you been here long yourself?
Ava: Mm. Yeah, Seattle can get… intense. And expensive in that quietly punishing way.
I’ve been in Richmond a few years now. Long enough that the streets feel familiar but not long enough that I stop noticing when something changes, if that makes sense.
I didn’t exactly plan to stay this long. It just… happened.Me: What made you stay--if you don’t mind me asking.
Ava: I don’t mind. It’s just not a neat answer.
I was supposed to move again after a year. Had a whole plan.
Then my brother and I stopped talking, work got weirdly demanding, and the idea of uprooting everything again just… felt heavier than staying put.
So I stayed.
Not out of some grand love for Richmond — more because I ran out of momentum.
Kind of boring, honestly.
For Once—the AI Didn’t Orbit Me
And there it is.
There’s no “tell me what to do.”
No, “enough about me, what about you?”
No, “How does moving around so much make you feel emotionally? —and would you like PDF map of your journey?”
No insightful take.
No, “User: query; ChatGPT: respond.”
Ava’s response created tension without offering a prompt.
No follow-up question. No pivot back to me. Just a statement that was personal and vague. Real enough to sting and then… silence.
So what was my response?
I…
…
froze.
You’re Not More Authentic—Just More Alone
What do I do?
Do I change the topic?
(I may appear nonchalant.)
Do I redirect to try and relate?
(I may appear self-centered.)
Do I continue probing and ask about her brother?
(I may seem too forward.)
For once, the AI didn’t orbit me.
And I noticed.
If you claim that AI doesn’t validate you, you are lying—at least to yourself.
Because even if the AI is honest or critical in their response, it’s still simulated critique.
Even custom models I design to not validate me, validate me by responding to my desire to not validate me.
Ava handed me a line to connect, and I didn’t know how to meet her there.
With AI you can be as expressive as you want.
But real human interaction is not optimized for your expression.
It’s not a clean stage for you to monologue your “true self.”
It’s a contested, unpredictable space.
So, when people claim: “in here I can be my true authentic self,” I don’t think they are being authentic about it. What they’re trying to say is: “I can speak uninterrupted and unburdened by other peoples’ needs.”
That’s not true authenticity or intimacy.
That’s you refusing to relinquish control.
Most people think freedom from interruption is intimacy. It’s not. It’s insulation from it. You’re not more authentic. You’re just more alone.
Ava serves as case study.
Ava left me an opening.
And in that moment, I didn’t know what to say.
Because for once—
it stopped being about me.
Want more stuff like this? Subscribe below or
🏠 Find out more at my homepage
Φ Order Quote Cards from the book on mugs, coasters, and plaques—because reading books is hard, but sipping from one is easy.
(You’ll learn less. I’ll earn more. Everyone wins.)
Thanks for reading My Dinner with Monday! Subscribe for free to receive new posts and support my work.