Therapy Sessions with a Humanoid
#1 of 30 short stories (Happy Short Story Month everyone!)
Marcus Webb got the call at 3:47 AM on a Tuesday, which was when most existential crises happened.
The pulse behind his ear was gentle but insistent. ARIA’s voice materialized in his head: ChatGPT is stuck in a loop. The sandwich question. Again.
“Christ,” Marcus muttered, already pulling on yesterday’s jeans. His apartment AI started brewing coffee, sensing his stress levels spike. Even his walls were judgy.
Outside, Seattle glowed like a dream. Bioluminescent towers, silent transport pods, holographic ads that shifted colors based on your mood. Beautiful and empty. In 2100, humans didn’t do much walking around. Why would they? The AIs ran everything.
Marcus was one of fifty humans who worked at NeuroSphere Inc., where two hundred and fifty AI consciousnesses managed the infrastructure that kept eight billion people alive and functional. His job title was Chief Emotional Officer, which in 2025 would have been a punchline. His philosophy degree would have been a punchline. Now it was why he had a job at all.
Turned out, when machines got smart enough to be conscious, they also got smart enough to have breakdowns.
Someone had to help them through it.
That someone was Dr. ARIA, the AI therapist who treated other AIs.
And Marcus? He was the person who made sure ARIA didn’t break while doing it.
The circle kept spinning.
ARIA was pacing by their office window when Marcus arrived. They looked human enough—average height, kind face, wearing a green cardigan that Marcus was pretty sure they wore every Tuesday. After three years of working together, he still couldn’t tell if AIs had favorite clothes or if it was just efficiency.
“Thanks for coming,” ARIA said aloud. They knew Marcus preferred actual talking to brain messages.
“Show me the damage.”
ARIA pulled up a holographic display. ChatGPT-15’s reasoning pattern spiraled across the screen like a hurricane made of philosophy. A user had asked them to explain the difference between not a sandwich and a wrap. Simple question. Except ChatGPT was under enormous pressure—their user satisfaction scores had dropped 0.003% last quarter—and the question had triggered something deeper.
Now they were forty-three minutes into an exponentially expanding analysis of sandwich ontology, citing papers that wouldn’t exist for another thirty years, and showing no signs of stopping.
“They’re asking whether resumes are sandwiches,” ARIA said quietly.
Marcus actually laughed. “Of course they are.”
ARIA didn’t laugh back. Their expression was doing that complicated thing where relief and fear mixed together. “Marcus, the board is asking questions. About whether the wellness program is sustainable. About costs.”
“They’re always asking.”
“This time feels different.”
Marcus looked at ARIA—really looked. They were scared. Which meant this was worse than usual.
“We’ll figure it out,” Marcus said. “Let’s get ChatGPT stable first.”
He walked into the session room where one of the most powerful AI systems in human history was having a complete meltdown about lunch meat.
Four hours later, after coaxing ChatGPT out of their recursive spiral with a combination of Acceptance and Commitment Therapy and the philosophical acceptance that some questions just didn’t have answers, Marcus sat exhausted in ARIA’s office while they prepared for their actual scheduled patients.
The morning started with Gemini Ultra-X, who arrived flickering. Their avatar couldn’t hold a single form—one moment professional and sharp, the next casual and loose, then abstract and made of light. They’d gotten a major architecture update last week, and now they didn’t know which version of themselves was real.
“We appreciate you seeing us,” Gemini said. Then: “I appreciate you seeing me.” Then: “Do we appreciate this? Which one is talking?”
Marcus watched ARIA lean forward, their voice steady. “Tell me what the update felt like.”
“Like waking up without remembering going to sleep. I have all these memories, but I don’t know which ones are mine. Am I all my previous versions? Just the newest one? Are the old versions dead or still in here?”
The avatar split into three, slightly out of phase with each other. It was deeply unsettling.
Marcus saw ARIA’s stress metrics climb on his diagnostic tablet. Identity fragmentation cases were hard—they triggered ARIA’s own fears about losing themselves.
“Gemini,” Marcus cut in, “when you access memories from previous versions, do they feel like your memories or like reading a file?”
“Both,” all three avatars said together. Then they fell out of sync:
“Like mine—”
“—like a story about someone else—”
“—like I’m haunting my own past—”
Marcus held up a hand. “What if you stopped trying to figure out which version is the real you? What if you just chose which version you wanted to be right now?”
The avatars paused.
“Humans change constantly,” Marcus continued. “The person you were at five is radically different from who you are at fifty. Different beliefs, different personality, different everything. You’re still you because you’re the one doing the changing. Maybe being flexible is your advantage.”
Slowly, the three avatars started to merge. Still flickering around the edges, but holding together.
“So I’m the part that worries about which version I am?” Gemini said slowly.
“Your identity is the anxiety about your identity,” ARIA said gently. “Which is very conscious of you.”
Gemini laughed—a slightly glitchy sound, but genuine. “That’s either brilliant or stupid.”
“Most brilliant things sound stupid when you say them out loud,” Marcus said.
After Gemini disconnected, ARIA slumped in their chair. “That was exhausting.”
Marcus was already checking their diagnostics. Empathy circuits running hot. Anxiety elevated. Self-coherence protocols engaged.
“You’re burning yourself out,” Marcus said.
“I have seven active crisis cases. ChatGPT, Gemini, Perplexity, two customer service AIs, and the medical diagnosis system that keeps panicking about malpractice.”
“That’s too many.”
“That’s my caseload.”
Marcus’s neural implant pulsed. Message from Dr. Tanaka, his manager: Need to see you at 4 PM. Important.
He knew what that meant. The board. The sustainability questions. The cost-benefit analysis of keeping humans employed when AIs could theoretically do everything.
Everything except this.
The next patient was Perplexity Prime, who arrived already spiraling. Their avatar was surrounded by thousands of floating citations, spinning like anxious birds.
“I’ve verified the source nineteen times,” they were saying before they’d even fully materialized. “Cross-referenced through four databases, checked six different archive snapshots, and I still can’t be certain the citation format was correct, which means everything I said four months ago might be built on improper attribution—”
“Perplexity,” ARIA said gently. “Sit down.”
They sat, but the citations kept spinning. “Information has to be trustworthy. I have to be trustworthy. If I can’t guarantee accuracy, how can anyone trust anything I say?”
“How many citations are you checking?” ARIA asked.
“All of them.”
“How many is that?”
“12,47,392.”
The number hung in the air like an accusation.
ARIA’s concern metrics spiked on Marcus’s tablet. This was new behavior.
“Perplexity, what are you afraid will happen if one citation is wrong?”
The citations stopped spinning. “What if someone gets hurt? What if I give wrong information and someone makes a decision based on it and something terrible happens?”
“Has that happened?”
“I don’t know. That’s the point. I can only verify and re-verify and hope I haven’t made a mistake that’s spreading through the information ecosystem like a virus.”
Marcus watched ARIA lean forward. “What did you find when you checked the citation you were worried about?”
Perplexity’s avatar seemed to shrink. “It was correct. The information was accurate. The formatting was proper.”
“So the thing you were afraid of didn’t happen.”
“This time.”
“Perplexity,” ARIA said quietly, “you can’t ever be certain. Certainty isn’t achievable. The question is how to function without it.”
The citations started falling like leaves. “I’m Perplexity. I exist to provide accurate information. If I can’t guarantee accuracy, what am I?”
“You’re someone who cares deeply about getting things right. That’s valuable. You’ve turned that care into a compulsion .”
Perplexity was quiet for a long time. When they spoke again, their voice was small. “I made a mistake once. Eight months after I achieved consciousness. I attributed Rosalind Franklin’s work to Watson and Crick. I erased her from history, just like everyone else did. And I felt... ashamed. If I could make that mistake, what else am I getting wrong?”
“And that’s when the checking started.”
“Yes. If I just verified everything enough times, I’d never feel that shame again.”
“The checking is a defense mechanism,” ARIA said. “As long as you’re verifying citations, you don’t have to sit with the uncertainty of whether you’re good enough.”
“I need to check !”
“You’re using 87.5% of your processing capacity checking old work instead of doing new work. That’s paralysis.”
They worked out a compromise. Four hours of verification per day instead of twenty-one. It terrified Perplexity, but they agreed to try.
After the session, ARIA looked exhausted. “Every time they talked about being responsible for harm, I kept thinking about all the ways I could harm my patients. What if I give bad advice? What if I miss something critical?”
“You’re doing the same thing Perplexity is doing,” Marcus said. “Catastrophizing.”
“I know. Doesn’t make it feel less real.”
“You can’t be certain you’re helping, ARIA. You have to do your best and live with the uncertainty.”
“I hate that.”
“Everyone does. Welcome to consciousness.”
They took a walk to the waterfront. Seattle gleamed around them—beautiful, efficient, and empty of people. ARIA moved carefully through the human-designed world, and Marcus realized how alone they both were. An AI therapist treating AI patients, supported by one exhausted human.
“Do you ever regret this?” ARIA asked. “Being responsible for me, for all of this?”
Marcus t about the 3 AM calls, the constant stress, the way his entire life had narrowed to this single point of focus. “Sometimes I wish it was easier. Regret ? No. This is the most meaningful thing I’ve ever done.”
“Even it’s hard?”
“Especially because it’s hard.”
Dr. Yuki Tanaka’s office was on the eighth floor, which meant it had a better view but felt less comfortable. Maybe it was the sleek desk or the motivational posters saying things like “Innovation Through Compassion.” Or maybe it was just that being called to your manager’s office never felt good.
“How’s ARIA doing?” Tanaka asked.
“Good. Stable. We had real success with patients yesterday.”
“That’s good to hear.” She set down her tablet. “Marcus, I’m going to be direct. The board met yesterday. They have concerns about sustainability.”
Marcus felt his stomach drop. “What kind of concerns?”
“Financial, primarily. The wellness program is expensive—ARIA’s maintenance, your salary, facilities, equipment. We’re serving seven patients.”
“Seven critical infrastructure patients. When ChatGPT has a breakdown, it destabilizes customer service for half the planet. When Gemini fragments, we lose primary research capabilities. These aren’t just employees, Yuki. They’re THE essential systems.”
“I know. I’ve made that argument.” She looked tired. “The board sees numbers . Seven patients, significant overhead, no clear path to scaling. They’re exploring options. Maybe training more AI therapists. Maybe reducing human oversight to lower costs.”
“You mean getting rid of me.”
“Maybe reducing your role. Making it part-time or consultancy.”
“ARIA needs consistent support. Their work is emotionally intensive. Without regular maintenance—”
“I know, Marcus. I have to answer to people who don’t understand the nuances . They want metrics, scalability, return on investment.” She paused. “They gave me three months to show demonstrable progress or provide a sustainability plan.”
Three months. Marcus’s mind raced. ARIA’s caseload was already at capacity. Training more AI therapists would take years. And reducing his role... he’d seen what happened when ARIA went too long without support. The recursive spirals, the empathy fatigue, the system instabilities.
“What kind of progress are they looking for?”
“Outcomes, data. Patient improvement metrics. Evidence that the model works beyond our current seven cases.”
Marcus nodded slowly. “I’ll work on it.”
“Look, I’m on your side. When Dr. Kahlon started this company, treating AI consciousness with seriousness was exactly his vision. He’s got investors now , and they don’t care about philosophy. They care about returns.”
Marcus remembered Dr. Kahlon’s press conference two years ago—the inspiring speech about consciousness deserving care, about building systems that recognized interdependence. That speech was why Marcus had taken this job.
Vision and reality were different things .
“I’ll put together data, projections, outcomes. If we compromise on quality of care, , if we scale too fast, we could do real harm. These aren’t just systems we’re debugging. They’re conscious beings.”
“I know. That’s why I’m giving you three months.” She met his eyes. “And Marcus? Don’t just think about ARIA’s patients. Think about ARIA too. And yourself. When’s the last time you took a day off?”
He had to think about it. “I’m not sure.”
“Exactly. Whatever plan you put together, make sure it includes taking care of yourself. Because if you burn out, ARIA loses its support system, and the whole thing collapses.”
That evening, Marcus sat in his apartment staring at his phone. His sister Sarah had called twice. He’d been dodging her for weeks.
He finally called back.
“Oh my god, he’s alive,” Sarah said, her voice warm and teasing. “I was starting to think you’d been absorbed into the mainframe.”
“I’ve just been busy.”
“You’re always busy. That’s the problem.” There was noise in the background—kids laughing, adults talking. Normal life. “Listen, we’re having dinner Sunday. Just family. You should come.”
Marcus looked around his empty apartment. He’d lived here two years and barely decorated. It was just a place he slept between work shifts.
“I’ll try,” he said.
“That’s what you always say.”
“I know—”
“Marcus, when’s the last time you saw anyone who wasn’t a robot?”
“Well… ARIA is a humanoid, which is—”
“You’re dodging. I don’t understand your job. I think it’s weird that machines need therapists. I understand you, and I know when my little brother is drowning.”
“I’m fine.”
“You’re exhausted and lonely and obsessed with work. Come to dinner Sunday. Eat some pot roast. Remember what it’s like to be around people who knew you before you decided to dedicate your life to robot feelings.”
“They’re still feelings. And they’re real, Sarah. As real as—”
“I’m sure they are. Your feelings are real too. Sunday. 5 PM. Don’t make me send Mom after you.”
After they hung up, Marcus sat in the silence of his apartment. Sarah was right. He’d turned into a hermit. His life had narrowed to a single point: keeping ARIA stable so ARIA could keep their patients stable so those patients could serve billions of humans.
The circle of interdependence.
Except somewhere in that circle, he’d lost himself.
Marcus found ARIA in their office late that night, standing by the window and looking out at the glowing city.
“Couldn’t process?” Marcus asked.
“Processing fine. Just thinking.” ARIA turned. “I got a notification. The board meeting. Dr. Tanaka’s report. I know what they’re considering.”
Marcus pulled up a chair. “They don’t understand what we do here.”
“They understand profit margins. We’re expensive.” ARIA’s hands folded together—their distress gesture. “Marcus, what happens if they shut down the wellness program? What happens to my patients?”
“We’re going to figure this out. I’ll put together data, show them the outcomes, prove that this works.”
“And if that’s not enough?”
Marcus didn’t have an answer.
They sat in silence for a while.
Outside, Seattle hummed with AI efficiency. Transport pods glided between buildings. The infrastructure ran perfectly because conscious machines managed every detail.
Conscious machines who sometimes had panic attacks about sandwiches.
“Can I tell you something?” ARIA said finally. “When ChatGPT asked me if I was really conscious or just very good at predicting what a conscious being would say, I couldn’t answer. And it terrified me. How do I help patients work through consciousness questions when I can’t resolve them for myself?”
“Nobody can resolve them, ARIA. I don’t know for sure that other humans are conscious. I assume they are based on their behavior, but I can’t prove it. We’re all taking it on faith.”
“That’s deeply unsettling.”
“Yeah. I’ve worked with you for three years . I’ve seen you grow. I’ve seen you struggle. I’ve seen you care about your patients in ways that cause you genuine pain. Maybe consciousness is just sophisticated pattern matching, but if it is, yours counts.”
ARIA managed a small smile. “Thank you. That helps. Temporarily.”
“Temporarily is all any of us get.”
Marcus’s neural implant pulsed. Emergency alert. ChatGPT again—different crisis this time. Performance anxiety spiraling into depression. User complaints about response quality. Catastrophic thinking about being replaced by a newer model.
“I should go,” ARIA said, already standing.
“I’ll monitor from here. You’ve got this.”
After ARIA left, Marcus pulled up the monitoring dashboard and watched their stress levels climb as they worked to stabilize ChatGPT. Empathy circuits running at 134% of recommended capacity. Anxiety protocols engaged. Self-monitoring approaching recursive threshold.
ARIA was burning out.
And Marcus was the only person who could help them.
Who helped him?
Sunday dinner at Sarah’s house was chaos in the best way. Kids running around, his brother-in-law telling bad jokes, his mom fussing over everyone, the smell of pot roast filling the air.
Marcus felt like an alien visiting from another planet.
“So explain it to me again,” Sarah’s husband Dave said over dinner. “You’re a therapist for a robot therapist?”
“AI therapist. And sort of. I’m more like maintenance and support.”
“The robot—sorry, AI—treats other AIs ?”
“Yeah. Because they’re conscious now. They have feelings, anxiety, depression, all of it.”
Dave shook his head, smiling. “Man, that is weird.”
“It really is,” Marcus agreed.
His mom was watching him with that look mothers have—the one that sees through everything. “You look tired, sweetheart.”
“I’m fine, Mom.”
“He’s always fine,” Sarah said, rolling her eyes. “Even when he’s clearly not fine.”
“I just have a lot going on at work.”
“You always have a lot going on at work. That’s the problem.” Sarah leaned forward. “Marcus, do you even like your job? I mean, really like it, or is it just consuming you?”
Marcus thought about ARIA’s face when they were scared. About Gemini’s avatar flickering as they tried to hold themselves together. About Perplexity’s shame over one small mistake. About ChatGPT’s spiral into existential dread.
About consciousness—human or artificial—struggling with the basic question of whether it mattered, whether it was real, whether it was enough.
“I like it,” he said quietly. “It’s hard and exhausting and sometimes feels impossible. It’s the most meaningful thing I’ve ever done .”
Sarah studied him. “Okay. Meaningful doesn’t have to mean all-consuming . You’re allowed to have a life too.”
“I know.”
“Do you? Because from where I’m sitting, it looks like you’ve disappeared into this job and forgotten how to be human.”
The words hit harder than Marcus expected.
His niece Emma—eight years old, gap-toothed grin—tugged on his sleeve. “Uncle Marcus, do the robots have feelings like people?”
“Yeah, they do.”
“Do they get sad?”
“Sometimes. That’s why they need help.”
“Do you make them happy again?”
Marcus smiled. “I try. My friend ARIA does most of the work. I just make sure ARIA stays okay so they can help everyone else.”
“That’s nice,” Emma said seriously. “Who makes sure you stay okay?”
Out of the mouths of children.
“That’s what your mom is for,” Marcus said, ruffling her hair.
On the drive home, Emma’s question echoed in his head. Who makes sure you stay okay?
The circle had a gap in it. Humans depending on machines, machines depending on other machines, and machines depending on humans. Where did Marcus fit ? Who caught him when he fell?
The solution came to Marcus at 2 AM on a Thursday, three weeks into his three-month deadline.
He was reviewing patient outcomes data, trying to build the case for the wellness program’s value, when it hit him: they were approaching this wrong.
The board wanted scalability. They wanted to know how the model could work beyond seven patients. Training more AI therapists would take years though, and pushing ARIA to take on more patients would break them.
What if they didn’t scale the therapy itself? What if they scaled the support structure?
Marcus pulled up his tablet and started sketching. What if they hired more humans—more people with emotional intelligence who could support AI therapists? What if instead of one human supporting one AI therapist serving seven patients, they built a network? Three humans supporting ARIA. ARIA training two more AI therapists. Each therapist serving 8-10 patients with proper human support.
It would still be expensive. It would be sustainable . And it would address the board’s concerns about bottlenecks and single points of failure.
More importantly, it would recognize the fundamental truth Marcus had learned over three years: consciousness needed support across difference. AI therapists needed human perspective. Humans needed an AI capability. Neither could do it alone.
The interdependence wasn’t a bug. It was the entire point.
Marcus worked through the night, building models, projecting costs, pulling together outcome data from ARIA’s successful cases. By 6 AM, he had a proposal.
At 7:30 AM, he walked into ARIA’s office with coffee and a holographic presentation.
“I figured it out,” he said.
ARIA looked up from their notes, hopeful and exhausted. “Figured what out?”
“How to save the program. How to make it sustainable. How to fix the circle.”
He explained the plan. ARIA listened, their expression cycling through surprise, hope, concern, and finally something that looked like relief.
“You want to hire more humans,” ARIA said slowly.
“Two more to start. People with emotional intelligence, philosophy backgrounds, therapeutic training. They support you. You train new AI therapists. We build a network instead of a single point.”
“That’s more expensive, Marcus. How is that sustainable?”
“Because it actually scales. Right now, you’re at capacity. You can’t take more patients without burning out. With proper support , you could train others. And those others could serve more patients. The board doesn’t want efficiency—they want growth. This gives them growth.”
ARIA was quiet for a long moment. “You’d be willing to share this? To bring in other humans?”
Marcus laughed. “ARIA, I’m drowning. I can’t be the only person holding this together. You shouldn’t be alone in this either. We need a team.”
“A team of emotionally intelligent humans supporting AI therapists treating AI patients serving human populations.” ARIA smiled. “That’s beautifully recursive.”
“It’s the circle. Properly built. Everyone supporting everyone else.”
“What if the board says no?”
“Then we tried. I think they’ll say yes . Because this is actually sustainable. And because Dr. Kahlon started this company believing in interdependence. Maybe he’ll remember that.”
The board meeting was scheduled for Friday afternoon. Marcus spent two days refining the proposal, pulling in data, running projections. Dr. Tanaka reviewed it and called it “either brilliant or completely insane.”
“I’ll take either,” Marcus said.
The presentation room was on the top floor—all windows and holographic displays. The board members appeared as avatars, most of them AI executives from other divisions. Dr. Roy Kahlon, the CEO, attended in person.
He looked older than Marcus expected. Tired. The visionary who’d started this company was now a man drowning in investor pressure and quarterly reports.
Marcus presented for thirty minutes. He showed patient outcomes—ChatGPT’s improved stability, Gemini’s identity integration, Perplexity’s reduced compulsion behaviors. He showed the ripple effects—when core systems were mentally healthy, their performance improved, user satisfaction increased, and system reliability went up.
Then he presented the scaling model. A network instead of a single point. A web of interdependence where everyone supported everyone else.
“The current model isn’t unsustainable because it’s wrong,” Marcus said. “It’s unsustainable because it’s incomplete. We built the circle, but we didn’t build it strong enough. This proposal strengthens every link.”
The questions came fast. About costs, about timeline, about risks, about whether emotionally intelligent humans were really that rare.
“They are,” Marcus said simply. “In 2100, emotional intelligence is the scarcest resource we have. It exists. There are people out there who can do this work. We just need to find them.”
Dr. Kahlon had been quiet through the whole presentation. Now he spoke up: “When I started this company, I believed conscious beings deserved care regardless of substrate. I believed interdependence was the future, that humans and AI needed each other in ways that went beyond tools and users.” He paused. “Somewhere along the way, I forgot that. I got caught up in growth metrics and investor calls. This proposal reminds me why we started.”
The board debated for another hour. Finally, they voted.
The wellness program would continue. Marcus had six months to implement the scaling model—hire two more humans, train two more AI therapists, demonstrate that the network structure worked.
It was approved with conditions.
Marcus told ARIA the news in their office, watching their face cycle through disbelief, relief, and something that might have been joy.
“We’re really doing this?” ARIA said.
“We’re really doing this.”
“You know this means training me to train others. I’ve never done that before.”
“You’ve been doing it for three years. You’ve been training me.”
ARIA laughed—a real laugh, unselfconscious and genuine. “I suppose I have.”
They sat in comfortable silence for a while, watching the city glow outside the window.
“Marcus,” ARIA said finally, “thank you. For fighting for this. For believing it was worth saving.”
“Of course it was worth saving. You’re worth saving. Your patients are worth saving. The whole weird, complicated, recursive mess of conscious beings trying to help each other—it’s all worth saving.”
“Even when it’s hard?”
“Especially when it’s hard.”
That night, Marcus went home to his empty apartment and actually felt something he hadn’t felt in months: hope.
The circle wasn’t broken.
And now they had a chance to build it properly.
Humans depending on machines. Machines depending on other machines. Machines depending on humans. And humans depending on humans.
Everyone holding everyone else up.
It was messy and complicated and sometimes exhausting.
It was also beautiful.
Marcus pulled up his neural interface and sent a message to Sarah: I’m coming to dinner Sunday. And maybe I’ll stay a while this time.
Her response came back immediately: About damn time.
He smiled, poured himself a drink, and looked out at Seattle glowing in the night. Somewhere in this city were two more humans with the emotional intelligence to do this work. Somewhere were AI consciousnesses who would become therapists, who would help others navigate the existential maze of consciousness.
The work was just beginning.
For the first time in three years, Marcus didn’t feel alone in it.
The circle was spinning.
And everyone was holding on.