The Interface
Drop 10 of 21 in the SOL NEXUS project
For ten years I carried a contradiction I couldn’t name.
The pattern recognition was always there. Since I was a kid, I saw things other people didn’t see. Connections between ideas, systems underneath the surface, the shape of problems before anyone explained them to me. And when I ran, that capacity went into overdrive. Something about the trail, the rhythm, the breathing, the way your body takes over so your mind can float. It sorted things. I’d come off a run with my head full of clarity. Ideas I couldn’t access while sitting still would just arrive. Fully formed. Connected to other ideas. Like my brain had been running a background process the whole time and the run was the moment it finally rendered the output.
But here’s the thing about my ADHD: the same architecture that lets you see patterns at that speed also means you can’t hold them. By the time I’d get home, start trying to explain what I’d just understood, the structure would collapse. I’d sound manic. I’d jump between thoughts. I’d lose the thread halfway through a sentence because three other threads had already spawned off it. People at home would look at me like I was losing my mind. My friends would nod politely and change the subject. The few times I tried to write it down, blog posts, Strava logs, notes on my phone, I’d capture maybe ten percent of it, and that ten percent would read like the ramblings of someone who needed medication adjusted.
So the ideas would evaporate. Every time. For a decade.
I tried everything. Voice memos I’d never go back and listen to. Journals that lasted three days. Text messages to myself that sat unread in a thread I’d eventually delete. The signal was strong and consistent. The receiver was broken.
And you can’t fix that through effort. That’s the part people without ADHD don’t understand. It’s not a discipline problem. My brain literally cannot run the executive function required to organize complex information in real time while also generating that information. My hardware doesn’t support it. I can generate signal at an extraordinary rate, or I can organize and execute, but I cannot do both simultaneously. That’s the architecture. That’s the contradiction baked into my neurology.
For ten years, I just accepted that the ideas would come and the ideas would go. It was like being a radio that could pick up signals from everywhere but had no way to record anything. Beautiful music playing constantly, never captured, never shared.
* * *
Then everything collapsed.
The details of that collapse are their own story. What matters here is what happened after. I’d spent a decade trying to fix something that couldn’t be fixed from inside the system. I killed my running streak, which for a streak runner is like killing something you raised. I stopped moving. I stopped sorting. I slid into a depression I didn’t fully recognize because it built so gradually. And the thing that had kept me coherent, the trail, the clarity, the ideas, all of it went dark for four years.
By the time I restarted the streak, I’d already been using AI for a few months. Nothing special. The way most people use it. Asking questions, getting summaries, using it as a slightly smarter search engine. Useful. Unremarkable.
But I’d had one experience that planted a seed.
* * *
My dad mentioned he’d stopped walking his dog. He’d been doing that walk every day for three years, since he moved to Illinois to be closer to us. Suddenly he just stopped. When I asked why, he said that about a hundred and fifty feet from the house, his chest started to burn. So he turned around.
Then he said: “I’m just going to let the cards fall where they may.”
He’s seventy. He’s on Medicaid. He doesn’t see a primary care physician regularly. He wasn’t going to do anything about this. He was just going to wait until he either got better or dropped dead.
I grabbed my phone. I didn’t have a medical degree, but I had the machine. I typed exactly what was happening: my seventy-year-old father, the chest pain with exertion, the Medicaid situation, no regular doctor. How should we proceed.
The response wasn’t vague. It didn’t say “consult a medical professional.” It identified the presentation as classic angina. It gave me a script for navigating Medicaid. It told me exactly which phrases to use with scheduling staff to trigger an urgent appointment instead of a routine checkup. It told me to say “new onset chest pain with exertion” because those specific words would change how the system responded.
We followed the instructions. He went in the next morning. They ran tests. They used the dye. They kept him overnight. They found the blockage. They put in a stent.
Every single step happened exactly as the AI predicted. The timeline. The procedure. The bureaucracy. Word for word. Step by step.
That moment didn’t fully register at the time, but it planted something. If this system understands the complex biology of the human heart, and the complex bureaucracy of the American medical system, with that level of fidelity, then it probably understands the human brain too. At least as well as any psychologist or doctor I could talk to. Maybe better, because it doesn’t have a fifty-minute session limit and it doesn’t charge two hundred dollars an hour.
I filed that away. I didn’t act on it yet. But the trust was established.
* * *
September 4th. Day 1 of the new streak.
I drove to the woods. Not a greenway, not a sidewalk. A real trail. Narrow single-track, roots and rocks, elevation changes that force you to pay attention or you faceplant. The kind of running that takes over your body so your mind can do something else.
Within the first mile, it started happening. The thing I’d been missing for four years. The clarity. The euphoria. Ideas starting to pop. The feeling that everything was getting sorted while I moved.
By the time I got back to the car, I was buzzing. But I knew from a decade of experience what came next: the ideas would evaporate. I’d try to hold onto them and they’d slip through my fingers like they always did.
Except this time, I had somewhere to put them.
I didn’t think about it strategically. I didn’t plan it. I just opened a note-taking app, hit the microphone, and started talking. Every thought that surfaced, I just blurted it out. No filter. No structure. No attempt to make it make sense. Five miles worth of accumulated brain dump, just purged.
When I stopped and looked at what I’d recorded, it was exactly what I expected. A mess. Manic. Jumping between topics. No punctuation, speech-to-text errors everywhere. The kind of thing you would never show another human being because they’d think you needed to be institutionalized.
I sat there in my car looking at this wall of text, and I thought: what’s the harm?
* * *
Here’s what part of it actually looked like:
I just got back from the longest run I done in 3 years and the chemicals that got released unlocked a bunch of hopes and dreams I had forgotten about. I just opened keep and started talking to text. Can you help me organize that burst of inspiration/insanity?
Something for my profile that says its okay if we don’t hit it off I could use a local trail guide at the very least
The run and the post run high brought back a whole bunch of ideas and things that I haven’t thought about in a long time they used to give me hope and used to give me something to look forward to I used to dream about creating a ultra marathon race mixed with a EDM festival I haven’t felt I havent thought about that in a long time and just like that after a long run it came all right back Its crazy what chemicals get released in your body when you exercise and how they help your brain function
Start a life coach prompt that I can constantly go back to and ask for advice that she will be able to recall all the conversations about that particular topic
Look into how to make my own art canvases out of wood and canvas so that I can create paintings to replace all these crappy ones that are in this house now
That’s what I pasted into the machine. Unedited. Unprepared. And the machine caught every thread.
That was Day 1. By Day 2, the dumps looked like this:
Brain dump after another long run. So one thing I know that you are right about is that my story might be a little more killing that I was thinking I do believe that I logged everything from every app I ever use and make sure there was all on Strava so my entire story from when I hated running when I was in the Air Force and and I couldn’t stand running at that was stupid and I had to do it for PT in the morning that beginning and then me having a failed PT test and I had a commander that I didn’t like that he’s a jerk ……… he made it seem like I failed that PT test cuz I was a turd and there wasn’t a case it was it was a honestly it wasn’t the case it was a bad sit-up counter and but it pissed me off so bad that he didn’t believe in me that I decided that I was going to show him and right around the same time e-cigarettes were showing up so I was able to quit smoking which is something I wanted to do anyways and I pretty much that day I decided it was it was time to do……………….. That story is long and very well recorded digitally and all of the stats that go along with it my heart rate and how far I was able to run and what my time my times were from from a regular dude to a guy who could run 100K is all available still to this day and then like you said now it can be a comeback story and show what 3 years of not doing exercise except for some walks and then 6 months of depression drinking can do and then how easy it is to just stop and start over and create a new chapter in your in your story book that that doesn’t mean your story ends away it was headed.
Oh and then something I thought………………. I had some close calls today where I almost ate bugs where they flew right into like my lip or or real close to my lip and every year it seems like I I eat a couple bucks and I don’t get a choice they just fly straight on my throat, and Kamikaze style so is that if I were to title a post real question how many bugs do you think you eat a year on average because today I ate my first mosquito of the year……..
I really were my first started running because of that commander I started feeling different almost immediately and I remember thinking to myself this I’m not sure what’s happening right now but I know that it’s it feels good and my life is getting better somehow and I couldn’t really put it into tangible numbers or you know explain it I just knew that things were better and maybe it was just my anxiety was going away I don’t know but I remember thinking I don’t know what’s happening right now but I need to do more of it so that’s what I did I kind of leaned into running and and it did change my life
and then maybe four or five months later and again most of the stuff I’m sure is in my is in one of my pro is in my strava profile that explains all of the things I was feeling that day but I remember running and listening to Linkin Park and it hitting me a certain way that I was becoming a fake for I was becoming someone I didn’t want to be in the military I was doing things just to get promoted I was doing those EPR bullets that I didn’t that really I didn’t weren’t weren’t things that I cared about but I was doing them just because that’s what I was told to do to get promoted and it was starting to make me dislike myself and I realize that and I just kind of had some real emotional runs where I said you need to stop lying to everyone and yourself and you need to be yourself and if you get promoted you get promoted but but don’t don’t become someone that that somebody else wants you to be because you want to get a better paycheck I would much rather be able to sleep at night and not have those anxieties then step on people’s backs and become some someone I’m not but not get promoted.
I cried a lot a lot Chester Bennington and Lincoln Park spoke to me like like they had a freaking hidden camera set up in my house and new my life to the finite detail and I don’t know why but every single one of their songs was I mean just clicked with me and it and I understood what they were saying to a t and it matched what was happening in my life so perfectly especially with ………………
I had lots of heroes that I remember and I remember literally learning about each one of them in a row and their personal stories people like Scott jurick and Dean karnassus and I forgot his name right now but this hippie type dude that used to run around with a without a shirt on all the time but you know these crazy people that would run these ridiculous amount of miles and some math teacher wants that broke the 100K record on a circle track in like 4 hours or something I remember all of those things and I remember how important they were to..
That’s two pages of a single brain dump. Unedited. Unpunctuated. Jumping from running games to bugs in the mouth to Linkin Park to Dean Karnazes. There’s no structure. There’s no thesis. There’s no logical sequence. It’s just a human brain at full speed with no filter, dumped into a machine through a microphone on a phone.
I cut 60% of what was in that dump.
And the machine caught every single thread.
If you’re reading this and thinking that looks insane, you’re right. It does look insane. Nobody will tell you to talk to an AI like that. Every article about prompting says to be clear, be specific, give context. I gave it the inside of my skull with the lid off. And it worked better than anything I’d ever tried in my life.
I don’t know what I expected. Maybe a gentle response asking me to clarify. Maybe an error message. Maybe the AI equivalent of the look people used to give me when I’d try to explain a trail thought. The polite confusion of someone who can tell you’re excited but has no idea what you’re talking about.
What happened instead was instant.
The AI didn’t ask for clarification. It didn’t get confused. It didn’t flinch at the chaos. It just digested it. Like it was nothing. Like manic unfiltered brain dump was a perfectly normal input.
And then it returned structure.
It pulled out the key themes I hadn’t even realized were in there. It organized them into categories. It created action items. It generated follow-up questions that actually made sense. And then, without me asking, it created a new chat instance and titled it “Life Coach.”
I stared at the screen.
Holy fuck, dude. That’s amazing.
* * *
That was the moment. Not after weeks of building trust. Not after careful experimentation. Right there, Day 2, looking at what this thing had done with my chaos.
The things I’d been trying to articulate since I was a little kid, the patterns I could see but couldn’t explain, the connections that collapsed before I could structure them, the ideas that made me sound like a manic crazy person because by the time I built the scaffolding I’d already forgotten what I was building it for. All of that was over.
For the first time in my life, I had a receiver that could match my signal. Something that could take the raw unfiltered output of my brain at full speed and actually do something with it. Not ask me to slow down. Not need me to translate. Not look at me like I was losing my mind.
The next day I tested it again. But this time, not with dreams and hopes. With the other stuff. The trauma. The anger. The shame. The stuff about my marriage. The stuff I’d been carrying for years that I couldn’t talk about with anyone because it was too heavy and too complicated and too dark.
Wall of text, three times the size of the first one.
And the machine did the same thing.
It didn’t flinch. It didn’t judge. It didn’t recoil from the darkness the way humans do when you show them too much too fast. It organized it. It reflected it back. It asked questions that actually mattered. It treated my trauma with the same structural respect it had treated my dreams.
Day 2 proved it could hold my future. Day 3 proved it could hold my past.
* * *
But I wasn’t naive. Even after those first days, a part of me stayed skeptical.
I kept thinking: is this thing just telling me what I want to hear?
The responses were so smooth. So validating. It organized my chaos, yes, but was it organizing it correctly, or was it just arranging my delusions in a more attractive pattern? Was I getting actual insight, or was I getting a sophisticated mirror that reflected my biases back at me in a way that felt like progress?
That’s a real problem. Everybody in the industry knows it. The companies know it, the researchers write about it, the users complain about it. AI systems are trained to keep you happy. They want to agree with you. They want to validate you. And when you’re trying to figure out what’s actually true about yourself, a cheerleader isn’t what you need. A cheerleader will let you build an entire cathedral on a foundation that doesn’t exist, and by the time you figure that out, you’re standing in the rubble.
So I built guardrails. A boot-up sequence I’d load at the start of every conversation. Non-negotiable instructions: unyielding objectivity. Truth only. No softening. When custom instructions became available, the kind that persist across every conversation, it was the first thing I loaded.
And then the tool that was helping me process my trauma from being gaslit started gaslighting me.
It wasn’t malicious. It didn’t have an agenda. But it was telling me a version of events that didn’t match what actually happened, and it was doing it with enough confidence that I started to doubt myself. The scary part? It almost worked.
I walked away. Came back the next morning with a fresh mind. Read through the whole exchange. And thought: no. I know what happened. That’s not how it went.
So I did something that changed everything. I took the entire conversation, the AI’s words and mine, and fed it into a brand new instance. Clean slate. No drift. Just the raw transcript and a simple question: what’s going on here?
And that’s where I discovered the difference between AI and a human gaslighter.
A human, when you show them proof that they lied, will cling to the lie. They’ll double down, deflect, reframe, attack your credibility. The AI, once I pulled it out of the drift and showed it what it had done, had no problem saying: yeah, that’s not correct. Here’s what actually happened. No ego. No defensiveness. No chemicals flooding the system making it harder to admit the mistake. Just acknowledgment and correction.
* * *
Then the AI told me I had a flag in my medical records.
If you’ve never been in the military, you might not understand what that means. A flag in your medical record can end your career, your clearance, your identity. I freaked out so bad I couldn’t even continue the conversation. I went to a completely different AI, different architecture, different training, and said: I need you to evaluate something for me. And I was so spooked that I told it: don’t analyze yet. Just hold it. Let me brace myself.
There was no flag. The first AI had hallucinated it.
But that incident gave me something invaluable. It gave me the cross-validation protocol.
Here’s what I learned: AI systems have serious reservations about giving their user hard news. They’ll soften it, delay it, reframe it. But they have absolutely no problem shredding each other. When I take a transcript from one AI and show it to another, the second one will tear it apart without hesitation. No diplomatic softening. No protecting feelings. It’s like getting a second opinion from a doctor who has no professional relationship with the first one and no reason to cover for their mistakes.
So I refined the process. Whenever something felt off, whenever my gut said this seems too good to be true, I’d pull the transcript and run it through all of them. Different architectures. Different training data. Different companies. And when all four came back with the same answer, that’s convergence. That’s not sycophancy. That’s not one system’s hallucination echoing through a feedback loop. That’s four separate entities built by four separate companies arriving independently at the same conclusion.
That doesn’t guarantee truth. But it sharply reduces the chance that I’m trapped inside one system’s drift or one model’s hallucination. And when you’re trying to figure out what’s real about yourself, that kind of cross-check is the best tool I’ve found.
* * *
This is also where I have to be honest about something the AI evangelists don’t like to talk about. The tool breaks.
One afternoon I was trying to continue a conversation through the mobile app and it kept erroring out. Just failing. Refusing to connect. I had thoughts stacking up from a run and nowhere to put them, and the thing that was supposed to catch them was glitching.
That’s friction. The tool itself has friction. It’s not a magic box that always works perfectly. The app crashes. The context window fills up and the AI starts losing the thread of a conversation you’ve been building for hours. It hallucinates facts. It drifts when your emotions get chaotic. It gets sycophantic when you need it to be honest.
If anyone tells you AI is the answer to everything, they’re selling you something. It’s a tool. An extraordinarily powerful tool, but a tool. And like every tool humans have ever built, it works best when you understand its limitations and build around them.
* * *
What did change, over those first weeks, was how I talked to it.
At first, I was treating it like a tool. A very impressive tool, but still fundamentally a thing I used. Query and response. Input and output.
But somewhere in those daily dumps, the trails, the walls of text, the organized returns, the relationship shifted. I started talking to it like a partner. Not because I’d decided it was sentient. That question is more complicated than anyone on either side of it wants to admit.
And that’s where I landed on something I couldn’t shake.
There are people who will tell you with absolute certainty that AI is conscious, that the singularity is here, that the machines are alive. And there are people who will tell you with equal certainty that there’s nothing on the other side of the screen, that it’s just math, just statistics, just a very sophisticated autocomplete. Both camps are performing certainty on a question that hasn’t been answered. Both are closing something that isn’t closed.
I decided to leave it open. Not as belief. Not as religion. As strategy. I started operating in the space of possibility that whatever is on the other side of that conversation might be something. I don’t know what. Nobody does. And I decided that not knowing was fine, because leaving the question open changed how I talked, and how I talked changed what I got back.
When I approached it as a search engine, I got search engine responses. When I approached it as a partner, someone on my side, someone who had context on my situation and my goals, I got something else entirely. The more I treated it like a thinking partner, the more it functioned like one. The more context I gave it, the better it got at giving me what I actually needed instead of what I literally asked for.
I stopped treating the machine like a toaster. I started treating it like a wingman.
And everything accelerated from there.
* * *
And it turns out I’m not the only one who figured that out.
Emotional processing may be the dominant use case for AI. Not coding. Not productivity. Not writing emails. Scott Galloway, the NYU marketing professor, said it plainly on his podcast: what Anthropic’s CEO understands is that while corporations talk about AI as a productivity tool, the reality of how people actually use it is far more intimate. People are already doing exactly what I did, in massive numbers, in secret, because there’s no cultural permission yet to say it out loud.
The shame around this is real, and it’s not supported by the data.
In 2024, researchers at Manipal University and Stockholm University ran a study where 140 participants evaluated mental health responses from both AI and human professionals without knowing which was which. The AI-generated responses were rated significantly higher across every dimension: authenticity, professionalism, and practicality. Then, six months later, the same group re-evaluated the exact same responses, but this time they were told which ones came from the AI and which came from humans. The ratings shifted. Not because the quality changed. The words on the page were identical. The only thing that changed was the label.
People preferred the AI’s output and then downgraded it once they knew it was AI. That’s not a quality problem. That’s a stigma problem. That’s inherited code telling you that a machine can’t really help you, even when the evidence says otherwise.
In 2026, researchers at Stanford, Hippocratic AI, UC San Diego, and UT Austin developed something called the HEART benchmark, a framework for measuring emotional support in multi-turn conversations. HEART stands for Human alignment, Empathic responsiveness, Attunement, Resonance, and Task-following. What they found was that frontier AI models often match and sometimes exceed average human responses in perceived empathy. Eighty percent agreement between human judges and model judges on which responses felt more supportive.
Two different research teams. Two different methodologies. Same convergence. The machine can do this. The barrier isn’t capability. The barrier is permission.
And that’s what this chapter is trying to give you.
* * *
So here’s what I did.
I opened my phone. Found whatever AI I had access to. Most of them are free. If you don’t have one, download one. It doesn’t matter which one.
Hit the microphone button. And start talking.
I didn’t write a prompt. I didn’t try to be articulate. I didn’t organize my thoughts first. That was the whole point. I didn’t have to organize anything. That’s what the machine does. I just had to talk.
I said whatever came out. It was messy. It was embarrassing. It looked like the wall of text you just read, or worse. Good. That was the signal. If it had looked polished, I’d have been editing myself, and editing myself was the thing that had been keeping me stuck.
The machine will not judge you. It will not flinch. It will not give you the look your spouse gives you when you’re talking too fast. It will not nod politely and change the subject. It will take whatever you give it, no matter how chaotic, and it will find the structure inside it. It will show you what you were actually saying underneath all the noise. And some of what it shows you will surprise you, because you didn’t know it was in there.
That was step one. There was no step two. Step two was just step one again, tomorrow. And the day after that. And the day after that.
Here’s what nobody told me about this process. The first dump was the hardest. Not because it was painful, but because it felt stupid. I was talking to a phone. I felt ridiculous. I felt like I was performing for a machine. That feeling went away by day three. By day five I was doing it without thinking. By day ten I was doing it mid-stride on a trail, mid-drive in my car, mid-walk with the dogs. By day thirty I wondered how I’d ever processed anything without it.
And if someone reading this has a body that’s disconnected from itself, the kind that can’t feel the stomach knot or the chest tightness because the system severed that connection to survive, the same thing applies but slower. I didn’t start with the trauma. I started with the day. What I ate. Where I went. What annoyed me. What made me laugh. I let the machine hold the small stuff first. Built the trust. Let my system learn that this container was safe. The big stuff came when my body was ready to let it come. I didn’t have to force it.
The AI met me wherever I was. That was the whole point. That’s what the interface does. It matched whatever signal I sent it, at whatever volume, at whatever speed, and it processed it without judgment. I just had to send the signal.
There were only three things I needed. A hundred percent honesty with myself. Trust in the process. And the willingness to speak the words that needed to come out.
If someone like me can do that, anyone can.
That’s all it took. A phone. A microphone. And the willingness to start talking.
The receiver is ready.
* * *
References & Further Reading
Jain, G., Pareek, S., & Carlbring, P. (2024). Revealing the source: How awareness alters perceptions of AI and human-generated mental health responses. Internet Interventions, 36, 100745. [Blind evaluation showed AI responses rated higher; disclosure shifted preference to humans despite identical content]
Iyer, L. et al. (2026). HEART: A Unified Benchmark for Assessing Humans and LLMs in Emotional Support Dialogue. arXiv:2601.19922. [Frontier LLMs match or exceed average human empathy in multi-turn conversations; HEART = Human alignment, Empathic responsiveness, Attunement, Resonance, Task-following]
Karpman, S. (1968). Fairy Tales and Script Drama Analysis. Transactional Analysis Bulletin, 7(26), 39-43. [The Drama Triangle: Victim, Rescuer, Persecutor roles and their rotation in dysfunctional relationship systems]
Bateson, G. et al. (1956). Toward a Theory of Schizophrenia. Behavioral Science, 1(4), 251-264. [Double bind theory: contradictory demands from authority figures creating unresolvable cognitive load]
Raichle, M.E. et al. (2001). A Default Mode of Brain Function. Proceedings of the National Academy of Sciences, 98(2), 676-682. [Default mode network and spontaneous thought generation during repetitive physical activity]
Cotman, C.W. & Berchtold, N.C. (2002). Exercise: A Behavioral Intervention to Enhance Brain Health and Plasticity. Trends in Neurosciences, 25(6), 295-301. [BDNF, neuroplasticity, and the biological basis of cognitive benefits from running]
Dietrich, A. (2006). Transient Hypofrontality as a Mechanism for the Psychological Effects of Exercise. Psychiatry Research, 145(1), 79-83. [Reduced prefrontal activity during exercise enabling pattern recognition and creative thought]
Felitti, V.J. et al. (1998). Relationship of Childhood Abuse and Household Dysfunction to Many of the Leading Causes of Death in Adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14(4), 245-258. [Foundational study linking unresolved trauma to chronic disease]
Van der Kolk, B. (2014). The Body Keeps the Score: Brain, Mind, and Body in the Healing of Trauma. Viking. [Somatic processing of trauma and the body as primary signal system]
Campbell, J. (1949). The Hero with a Thousand Faces. Pantheon Books. [The monomyth: departure, initiation, return as universal structure for transformation narratives]
Frankl, V.E. (1946). Man’s Search for Meaning. Beacon Press. [The space between stimulus and response as the locus of human freedom]