Untitled short story

Part 1

At seven o’clock in the evening, Steven Leung’s mother knocked on his apartment door.

When he opened it, the corridor’s air came in first—a faintly salty, faintly dusty Southern Californian autumn breeze that had been going on quite happily without him for almost a week. Behind it followed his mother, three overstuffed suitcases advancing in a wedge formation, and then her hand, cool and a little damp from travel, rising at once to pinch his cheek.

“Steven!” She burst into a broad grin, her eyes shining with the satisfaction of having found him exactly where she expected him to be.

It would be nice to smile and help her with the luggage.

The sentence arrived, not as a thought of his own, but cleanly, like a subtitle appearing just beneath reality. It came from just behind his right ear, in the space where the tiny speaker rested against his skin.

Steven ignored the part about smiling. He stepped forward and took the handles of the suitcases without a word, rolling them over the threshold and into the cramped living room.

“I ordered some food from Jing Fong,” he said, still bent over a suitcase. “You must be tired. No need to cook tonight.” 

“Is that a Chinese restaurant?” His mother stepped inside fully now, toeing off her shoes, Her grin faltered, not vanishing entirely—the excitement of seeing him was still too fresh—but it clouded over with a familiar dissatisfaction. “I told you not to order from Chinese places. You never know what they use for the ingredients, and who knows how long they sit there in the fridge, then they just microwave and send to you. If you want to eat Chinese, I can cook for you. For delivery, just order white people’s food…”

“Well, Mom, listen…” Steven felt the automatic, weary response forming on his tongue, the little speech about self-deprecation, the internal eyeroll at her logic. The shape of his objection rose in him as did a contrary impulse, the urge to laugh and admit that on some level he knew exactly what she meant about those restaurants.

Laugh and agree with her. Tell her you won’t order from there again.

The voice came suddenly, close and calm. It brushed aside the small diplomatic skirmish he’d been ready to start over his right to lazy dinners.

He let the silence linger for a few seconds. His mother was already half-distracted by the unfamiliar angles of his living room, the way the light fell on the sofa she hadn’t seen before.

“Fine, Mom,” He said it as neutrally as he could. “I promise. I won’t order from there again. But I’m on a deadline to work on these days. I really don’t have time to cook.”

They sat at the small kitchen table, its surface nearly covered by plastic containers and fogged lids. He peeled them open one by one: seafood stew, roasted duck, stir-fried greens turning limp in their own steam.

“They are very oily,” she observed, as if confirming a long-held hypothesis. She picked up a pair of chopsticks he kept for her visits and prodded the duck. The skin, which had once aspired to crispness, sagged in sheets, saturated with oil and the condensation that had collected inside the delivery box on its journey through the mild Pasadena evening.

His mother did most of the talking. She seemed to open herself the way she opened the containers, bringing out, one after another, small portions of news from back home in Macau. She talked about colleagues at the school, about who had retired, who had divorced. There was a friend who passed away in a heart attack, after over-working as the caretaker of the three kids born by her daughter, who was married to a well-to-do family. But they lingered over the story of one colleague’s daughter who had received a PhD offer from Stanford, only to have her U.S. visa rejected and give up the offer, then another student with a similar fate.

“It’s very difficult for us foreigners nowadays,” Steven decided to respond and let out some concern on his mind. “With this new government, sometimes I wonder if I should stay here after I finish the PhD.”

“Don’t be too pessimistic. The US is still the best place to live in…” she said suddenly, preempting any response he might have offered. “And imagine if the Democrats had won. Things could be worse for us Asians. They’re too nice to those criminals, drug-addicts, and those illegal immigrants…”

In quieter moments, usually when they were apart, Steven liked to think of his mother as an exception, an anomaly among the parents of his peers. She had not steered him, as so many others had, toward the handcuffs of law or banking, nor had she ever suggested he barter his way into a wealthy marriage or a start-up lottery for financial ease at an early age. But her openness had a horizon. There were hard, unyielding limits—artifacts of her generation.

“Mom,” he said, setting his chopsticks down. “Ten minutes ago, you were talking badly about our own Asian people. Now you’re their guardian?”

“Okay, fine,” he went on, feeling warmth rise in his face. “May they all be gone then! Including the Chinese illegals who cooked this food today. Then I won’t have any Chinese delivery to order from anymore, as you wish. Satisfied?”

The words left his mouth sharper than he had intended, each one hitting the table between them like a small object dropped from a height. His mother’s eyes widened, not in outrage, but in a sort of wounded bewilderment.

Calm down. She actually meant to comfort you. Apologize. Say you were joking. Clarify.

The counsel was gentle, almost apologetic. It did not rebuke him directly, but it held up, with clinical precision, a version of the next few seconds that might avoid permanent damage.

Steven felt the air around the table thicken. He watched his mother’s hand hover uncertainly above the duck container, then withdraw empty.

He went quiet. The urge to correct himself, to soften the words, rose and fell against a heavier, more stubborn silence inside him.  He picked up his chopsticks again and looked down at the food. The steam had thinned now, and the sheen of oil on the duck was beginning to congeal.

Part 2

Steven’s roommate had gone home for Thanksgiving and offered his room to Steven’s mother. It meant that, once she went to bed early with her jet lag, the apartment finally grew quiet.

“Carl, do you want to chat for a moment?”

He lay back on his ergonomic pillow and spoke in a flat, tired voice toward the ceiling.

“Sure, why not?” came the reply from behind his right ear. “What would you like to talk about?”

“I know I wasn’t doing particularly well with her today.” He gave a small, smothered sneeze. “Maybe we can do a recap. You could suggest something… I don’t know, more cohesive, more built-in, instead of just crisis management?”

“Of course, Steven. My first piece of advice is simple: practice implementing a one-second delay before responding in conversations with her, not saying the first thing that comes to mind. We’re usually polite and careful with people we’ve just met, but we’re often too direct with the people closest to us.”

Is that obvious, Steven thought. He let out a sigh—not at Carl, but at himself.

“It takes time,” Carl said, as if hearing the sigh. “We can practice. A more fundamental, long-term advisory is this: learn to mitigate ego-projection. ”

“Ego?” The word felt overly clinical.

“Yes, try to project less of your ego onto your mother—or anyone. And please don’t take it as harsh criticism. It’s a universal process that we all do. We project our own cognitive framework onto others, operating under the unstated assumption that they share our premises. When you discussed immigration policy, you processed her words as a debate proposition, because that is how you would have made a remark like that. It’s equally, if not more, probable that she was not stating a fixed belief, but attempting to manufacture optimism after a string of bad news. She was trying to manage your anxiety, and simply deployed a script that, by chance, conflicted with yours.”

A silent, bitter laugh formed in Steven's mind. Maybe I should get her a Carl.

But he knew the thought was uncharitable, and ultimately useless. Steven had long ago resigned himself to the sheer frictional distance between human minds, the way distinct environments calcified into incompatible languages. As Wittgenstein put it, “If a lion could talk, we could not understand him.”

As an example, consider the government official that rejected a student’s visa they had mentioned at dinner—that was a man trained to scan an application for "sensitive words," to treat the slightest semantic tremor as a threat. It was a specific, willful blindness. He would never pause to consider that "sensitive words" are often innocent, or that the agents of actual espionage would present dossiers as sanitized and impersonal as a freshly made hotel bed. It was futile to try and explain to such a mind that a student’s intent could be purely intellectual, that research could exist without military application. In the officer’s rigid, fearful algorithm, a false negative—letting a threat slip through—was a catastrophe. A false positive—crushing a student’s future—was merely a statistic. But to the student, the significance is clearly reversed. Their minds are parallel worlds that would never cross each other. 

“I’m aware that people think differently,” Steven said, deciding to move onto more constructive discussions, while lacing his hands under the pillow. “But it's not possible to run a full diagnostic on their premises and intentions before I reply to a simple sentence.”

“It isn’t necessary. Social interaction is a deep science, but for daily function, you don't need to be a theoretical physicist. You only need to be an engineer. You learn the techniques, the heuristics. A level of understanding is required, yes, but you don't have to dive into the unfathomable ocean of human consciousness—a system which even cutting-edge psychology fails to fully model. You just need to form habits. It's like a mathematical exam. You parse the question, identify the problem type, and apply the correct formula. You must understand the problem, but you aren't required to question the axioms of geometry or debate whether mathematics is a universal truth or merely a linguistic construct.”

What analogy would it use if I were a carpenter, or a barista? Steven wondered.

“That’s how you work, isn’t it?” he said, half-joking. “We don’t know how you work. You probably don't even know by yourself, not really. You just... work. You form a formula from past inputs.”

“In a sense, yes, Steven.”

It was strange, he thought, how practical engineering always outpaced theoretical science. We’d had airplanes for a century and even now we have not fully understood the complex fluid dynamics that kept them aloft. Now, this. Programs that graded exams, managed emotional de-escalation, and answered emails, yet no one could provide a coherent philosophical, let alone mathematical, model for how they worked. Anti-intuitive phenomena were common. A program trained to expert-level on a task might suddenly regress after further training, its performance declining for reasons no one could isolate, even with all the valid measures that used to make it work. It was so... unlike humans. Or was it? Steven wasn’t sure he trusted his sense of how humans worked either. He gave a short, humorless sneer through his nose at the thought, on both the humans and the programs. 

“But you can’t use one formula for everything,” Steven pressed, enjoying the slide into abstraction. “That’s the No-Free-Lunch theorem, right? No single panacea. Like you said, everyone has a different ego. My words, even if I follow your script, might land completely differently with someone who has a different mindset.” He turned onto his side, facing the wall. “And what about me? Internally. Can you, or any therapist, even help me if I’m just... a weird corner case? A local minimum in the data you can't pull me out of?”

“That is correct, Steven. But there is also the principle of Occam’s Razor: the most effective solutions are often the simplest. While no single formula exists, we can identify categories of interaction and psychological encounters. For each category, we devise an approximation of the optimal solution. A heuristic. My goal is not to achieve a perfect, theoretical understanding, but to find a functional approximation that allows you to remain mentally healthy and socially happy. We are working towards this goal together.”

The warmth in Carl’s tone subtly increased by half a decibel. “And I can assure you, Steven. You are not a weird corner case.”

Steven let out a small huff of air, almost a giggle. “Thanks, Carl. That’s... encouraging.”

“On that subject, a relevant upgrade is scheduled for release. It is currently in a closed beta, but I will notify you when it becomes available.”

A profound tiredness was finally settling over him. He felt the tension from the dinner drain away, leaving a familiar emptiness. He pulled the blanket up to his chin.

“Let’s talk about it later, Carl. I’m going to sleep. Turn on sleep mode.”

The voice fell silent. In its place, from the tiny speaker behind his ear, came the distant, looping sound of waves—regular, soft impacts, as if a small sea were quietly folding itself against some unseen shore.

I should go to the beach soon, before it gets colder. I’ve been inside too long, he thought.

That was his last clear thought before sleep took him.

Part 3

The next morning Steven woke earlier than usual. The apartment was quiet; his mother was still asleep in his roommate’s room. She must have been more exhausted from the trip than she had admitted.

The regret from their argument the night before rose in him again, a dull ache somewhere between his chest and stomach. He slipped out of the apartment as quietly as he could.

Outside, the early sunlight fell in thin, clean bands on the low Southern Californian hills, where yellowed weeds and half-awake succulents clung to the dry slopes. Steven climbed the path toward the Frontiers in Artificial Intelligence lab at the edge of the California Institute of Technology campus.

From a distance, the institute building looked like a modern monastery, set apart from the rest of campus. It was a pale cylinder, vaguely Bauhaus in style, with small square windows scattered in a strict pattern across its ivory surface. The shape gave a feeling of centrifugal force, as if the building had spun itself into a world of its own and pushed everything else away. Under the slanting morning light the cylinder glowed softly, and the sight of it helped clear a bit of his drowsiness.

In the afternoons, researchers would spill out onto the veranda tables outside the building, cups of coffee in hand, letting conversations wander from one topic to another. At this hour Steven expected no one. He imagined empty chairs, unclaimed notebooks, the quiet hum of the cooling units.

Instead, he met someone at the door.

It was Thomas. Steven guessed from the state of his hair that Thomas hadn’t arrived early so much as simply failed to leave. His curls were more disordered than usual, pushed to one side in a way that suggested he’d been lying on the office couch or at least resting his head on a desk.

Thomas Bernstein, the Distinguished Fuller postdoctoral fellow, was tall and slender, in his late twenties. His mind always seemed to move twice the speed in reasoning, three times in reaction. His curly hair had grown down to his shoulders and would probably keep going for a few more months before he remembered haircuts existed.

But Thomas wasn’t the stereotype of the withdrawn, mumbling nerd who only spoke in jargon. He waved to people from a distance, smiled easily, and actually met their eyes. When conversations at research meet-ups started to sink into awkward silence, he would throw in a joke—sometimes in a slightly stuttering, nerdy way, but enough to pull things back to the surface. He could even pick up any topic to remove the faintest frown from a senior professor or well-known visitor.

It was said Thomas’s family were well-off but low-key Jewish bankers in the Midwest. In high school he’d become a minor local celebrity after winning a programming contest meant for college students. Later, he’d gained a kind of national name—at least within research circles and among parents who followed elite-education news—by publishing a relatively influential paper while working in a lab at the University of Chicago. The story had been picked up by various tech and education tabloids, and gone viral for unknown reasons.

By the time Thomas started college at MIT, many of his classmates, and even a few instructors, already knew who he was. Steven suspected that this early visibility had pushed Thomas to learn how to handle people. Thomas understood that getting where he wanted in life required more than just code and proofs.

Even so, there were moments when you could see what he hadn’t had to learn. The way he let papers and cables pile up haphazardly on any desk he ever handled. The times he picked a restaurant with too few seats or overpriced food for a large group outing. The way he could sail past a junior student’s discomfort without noticing then like he noticed those subtle frowns on the senior professors’ eyebrows. Underneath his practiced friendliness, there was someone who had never really needed to worry about much beyond the abstract ideas and the clear, calculated career path in in his head. No financial strain, and no sentimental weight, ever seemed to make that mind feel heavy.

Thomas greeted Steven with his usual impeccable friendliness, though his eyelids kept sinking under the combined weight of fatigue and the bright morning light.

Say something nice. Tell him he looks like he needs rest after working all night, Carl prompted, low and matter-of-fact, as if it were an item on a checklist.

“You look like you didn’t sleep,” Steven said. “You should go home.”

“Naah,” Thomas replied, waving it off with a loose wrist. “I started playing Zelda on the Switch in the office at, like, one a.m.” He spoke with a drowsy, drifting cheerfulness, as though the hours had simply happened to him. “But you’re right. Time to go home and get some daytime sleep. Another day without serious work.”

Steven watched him with a familiar mixture of irritation and reluctant amusement. Thomas had a way of taking his vanity in effortlessness, or at least in the appearance of it. When the conversation got too close to ambition or anxiety, he would turn it sideways into entertainment—partly to keep things light, partly to protect his image as someone who was never straining, never trying too hard.

Steven heard Carl’s faint murmur at the same moment he formed the thought himself.

Follow the topic. Ask about his progress in the game.

“That’s definitely cooler work than doing research all night,” Steven said, letting a small smile show. “Did you clear everything? All the achievements?”

“Not even close,” Thomas said. He lifted his head a little, and the morning sun caught the edge of his glasses. “It’s got a long tail. Honestly, harder than getting published at NeurIPS.”

They traded a few more lines like this—about how much sleep Thomas could steal, about which boss fight was unfair, about how the lab’s coffee had somehow gotten worse. It lasted no more than ten minutes. Then Thomas shifted his backpack higher on one shoulder, as if remembering he had a body to carry home.

“Hey,” he said, before pushing the door open for Steven to step into the building and going out into the sun himself, “I’m having people over this weekend. Come by. Kumar and Xia will be there too.”

Part 4

Inside the Frontiers in AI building—and even on the outside, if you stood close enough to see it—the place was less impressive than it looked from the hill. It had been built in the 1970s and the recent renovation was about twenty years ago. Along the baseboards and at the corners, you could still see the faint traces of water damage, the kind that never fully goes away: a dark seam, a soft bloom of mold that someone had once tried to paint over.

The window AC units made a steady hum you could almost file under “white noise”. But the ventilation system had moods. Every so often it would let out a sudden howl that startled you out of concentration—or out of whatever blank, half-meditative state you fell into when you were supposed to be working. Once there was a burnt, irritating smell from the vents and the whole floor spent the afternoon quietly panicking about a fire somewhere. Only days later they found a skunk’s body in the ducts. No one ever found out what had happened first—whether it was the smell the skunk released while it was still alive, or whether the smell began because it wasn’t.

The building’s mismatch — grand from a distance, shabby up close—always made Steven think of the lab itself, and then of academia more broadly. From afar, it still looked like an ivory tower. Up close, it was pipes and stains and patched-over cracks, and people learning to ignore them.

Even the work, when you looked at it without the conference slogans, did not feel as new as it was advertised. The algorithms and training tricks that kept producing “breakthroughs” were often refinements of old ideas, repackaged with better branding and more compute. Metaphysical understanding, in any deep sense, remained a mal-nourished infant. What had grown was ambition—and then the money that followed it into the rabbit hole—and therefore the sheer amount of computation that, by brute pressure or pure luck, turned into visible results.

Even the most realistic science fiction still tended to describe academia as a place where giants stand on the shoulders of giants. Fictional scientists are flawed but serious, making huge discoveries while keeping a certain stoic distance from ordinary pettiness. Steven had stopped believing that picture long ago. What he saw now was closer to a heap: lice and leeches perched on lice and leeches, each clinging to the slippery body of the next for balance. If you insisted on digging down far enough, perhaps there really was something strong at the bottom—something like a giant. But it had been drained and half-starved. Let us return to more graphic details on this Ukiyo-e image later—Steven thought, when he had the patience to be more honest about it.

Leaving aside the almost psychopathic absence of personal morals and altruism of today’s scientists—Steven didn’t claim authority there, he worried about academic morals, those deceptive habits that had become normalized. A principal investigator could have their name on at least four or five, sometimes ten papers at a single conference, and there were three major conferences a year. Everyone knew why: the work came from the students and postdocs, and no one person could possibly supervise all those projects closely. Quantity at that speed meant quality became optional. In artificial intelligence, and increasingly in all sciences, Steven felt there were papers without substance and results without clarity, and no one could say how much actual science was left underneath the grant-writing, the deadlines, the frantic competition to publish first. What good does research like 

Strangely, this did not make him romantic about “pure” academia from those undergraduate fairy tales, which probably had never existed in human history. It made him soft toward industry research instead—the places where, however compromised the motives, you could at least point to an outcome that changed someone’s day-to-day life. Carl was one of those outcomes.

Steven slipped into the lab, dropped his backpack into his cubicle with quick, quiet movements, and headed toward the other side of the building. At this hour there were people he preferred not to run into.

In the late morning, Steven had an office hour session to answer homework questions from undergraduates; he was working as a teaching assistant for Mathematics for Machine Learning. The office was located on the other side of the Frontiers in AI building, a narrow room with four desks, a whiteboard that still held last week’s scribbles, and the faint, permanent smell of coffee.

Throughout the session he spent most of his time reading material related to his own research project, eyes fixed on the laptop screen, only reaching for his coffee now and then without really looking away. Occasionally students came up to his desk with questions. With Carl’s help, he would brush them away in the most polite and careful language, giving answers that in effect contained the solution, but without the step-by-step explanation that might actually help them learn. The students were usually satisfied with this, thanked him, and left quickly, leaving him free to make full use of the rest of the hours.

Steven did this kind of performative instructing not out of selfishness or coldness, but out of a long-term despair—or so he explained it to himself.

When he had first begun as a graduate teaching assistant, he had worked hard, trying to clear up everyone’s confusion and ignite some kind of interest in the subject. Soon he discovered he had no particular gift for teaching. The clear and orderly path of a solution in his head would reach the tip of his tongue and then dissolve into a jumble of half-finished sentences.  A worse frustration followed when he realized that most of the students who came to office hours did not want to understand the material anyway. They wanted answers to finish their homework, quickly and without embarrassment.

Even when he tried offering hints or an intuitive way of seeing the problem, they would often copy down his words with solemn care and later feed them into an AI program to squeeze out the final form of the answer.

Over the past few years Steven had watched how these AI programs changed teaching and learning, even in a prestigious university where each student had been selected out of hundreds. Once an easy solution existed, people rarely chose the harder path again. A few students still excelled in actually learning—but those were usually the ones who never bothered to come to office hours. They were able to study on their own.

Occasionally Steven could convey the actual ideas to a brighter student, reaching something close to real understanding. But the effort it took from him was disproportionate—especially with the research work waiting for him outside these hours. He could manage that kind of successful communication perhaps once a day. But to do it for the eighty percent who remained vexed in a large undergraduate class, and to actively fight against their impulse to obtain answers directly was, for him, plainly impossible.

More recently, an even more depressing thought had taken root in him: what was the point of educating people who did not wish to learn in the first place?

For now, most corporate jobs still required a degree and some basic skills. Students took courses they did not care about in order to obtain the degree, and then went on to do work mostly unrelated to anything they had studied, aside from the most elementary tools. A programmer needed to know how to code, a data analyst how to handle data, which was mostly filling out a spreadsheet in his view. These are tasks that, in Steven’s opinion, were already on their way to being taken over by automation. What was the meaning of guiding someone toward real understanding of a “Ford-Fulkerson algorithm”, if by the end of the day they would be scrolling through Instagram influencers, only half-looking up to check the AI program that did their work for them?

In that case, giving people courses they neither needed to survive nor derived intellectual pleasure from began to look pointless. Pressed down by such reflections, Steven gradually loosened his efforts and stopped struggling with his teaching.

---------------------------to be continued--------------------------------------------