in gratitude to the stories untold

<blockquote style="padding-top:5%"> Large Language Models send a single, determinate history to an indeterminate future distribution, log probabilities creating an imperative that controls the future and navigates it in one direction. And we, as readers, reintroduce interpretations while reading, defining meanings. </blockquote> <blockquote class ="blockquote_right" style="padding-top:5%"> Unfulfilled promises, failed expectations, dreams that were let down. Feelings that can't be put into words. Concepts that only exist in our heads. We all live in such an abyss. </blockquote> <div class="main"> There's an ending without an ending. The ending is, that night I watched myself, axe in hand, walk towards the beach where the water glimmered under the bright moon. I watched myself raise the axe, chopping--straight--down--at her. The sound of the axe hitting flesh and bone echoed in my mind, drowning out the sound of my own cries and the crash of the waves. <img src="https://raw.githubusercontent.com/KarlieZhao/the4PersonPersp/main/thisisit.png" width="100%"> [[Back to Ending->start]] </div>

<blockquote style="padding-top:10%"> Years after her death I got from her friends this stack of letters I sent her. Here are some excerpts. They were sadly not carefully preserved and some of the handwriting is hardly legible. So much time has passed that I can't remember exactly what I wrote. The text in dark blue was something I later speculated on backwards based on the context of the letter. They are merely one of the many possibilities behind a conclusive given event. </blockquote>

<blockquote style="padding-top:5%"> You walk along a line, each pace being the same length. Before each step, you flip a coin. If it’s heads, you take one step forward. If it’s tails, you take one step back. If the coin is unbiased, then each time the chances of you stepping forward or backward are equal. This is an one-dimensional random walk. The mathematical model Random Walk describes the journey of a wanderer navigating in space. After an arbitrary number of successive steps in any direction, the walking trajectory presents an unpredictable exploration in a space of a certain dimension, where the wanderer's future behaviour is independent of past history. </blockquote> <blockquote class="blockquote_right" style="padding-top: 60%"> "The physicist--just like the poet--should not describe the facts of the world, but rather generate metaphors and mental connections in language." -- Niels Bohr To ramble is to wander with words; Large language models are stream of conciousness writing devices, rambling machines, rovers of the latent space... He tried exploring the forest, to lose his strength; among the hemlock he barely succeeded in experiencing several short snatchs of sleep, veined with fleeting, rudimentary visions that were useless. He tried to assemble the student body but scarcely had he articulated a few brief words of exhortation when it became deformed and was then erased. --The Circular Ruins, Jorge Luis Borges I've been thinking lately about the reasons why people choose to end their lives. Some say it's because they're selfish, and can't see the pain they would cause to others. Others say it's because they're in a state of deep pain and can't see any way out. I used to hate Osamu Dazai's No Longer Human as a teenager but now I see my cowardly self in all its pages. This is not a story about someone who has died. It's a story about some dead. The most terrible shadows don't even creep. They come at you boldly, at your pace, as they drag at your feet. You can't move for them. </blockquote> <div class="main"> Here: Marilyn and I are strolling the hiking paths of Eagle's Nest trail that winds around the edge of the city. Only when we walk do we really talk, old-schooled, uninterrupted, thought-exchanging talks. The track styles entrancingly during early hours -- fiery plants, countless insects and birds, oceans smeared below the edge of the mountains, on the other side, villages submerged in fog. Our feet forget their habitual place on November bushes as strands of colors arranged and resolved from the naturality of our surroundings reminiscent of eastern blankets. Eastern -- that cuts west in precise ritual motion, prompting onward, now west. No wrong turn marks this path. It weaves in flat central regions, only sleepwalkers would get lost. Body guided and announced by jogging, breathing allows liquid wisdom if mind can contact it. Marilyn woke up in a fog, thoughts pulling us along in tangled paths. Everything turns into probabilities, fuzzy numbers forming a tip surface sense-of-direction, accompanied by periodic collapses of the ocean below. White noise creeps up. Why is it that we are taking daily walks, and how did we get here? We can feel more than see our bodies drifting through each other. Production of words retreats with all conversations turning to a narrowly displaced feeling: disquiet of wind disappearing at regular intervals made the space seperated. We had not been disgusted by the hypocrisy of this quasi random walk, early morning or after lunch, the most disrupted conflicts always come at the quietest pace. It occupied all our perceptions, through wars or starvation, or extreme climate when the earth's pulse paused. What drives us to edge after edge? They have found that Random Walk is at the heart of statistical physics—in predicting how fast one gas will diffuse into another, heat diffuses through solids. When a group of agents are randomly moving in a space with their behaviours not interfering with each other, the distribution of entropy at different locations of the system will be revealed after a centain number of steps. Feynman's Path Integral claims that getting one element wrong and the outcome swerves off in another direction abruptly. We traveled onto diverging tangents whilst the trail's earth underneath was being staggered by layers of topography, simultaneous implicitly and explicitly. New nuance translated itself into formalistic interpretations of regions, valleys, mountains, until all the discussions fracture at the world’s threshold, reality continues no more. <img src="https://upload.wikimedia.org/wikipedia/commons/c/cb/Random_walk_25000.svg" style="width:95%"> <hr> There: Marilyn and I are strolling along a harrowing trail that winds around the borders of our thoughts. Most of the time, it's a computationally modulated, quasi-randomly disrupted walk, where time is relative, space between us regardless absent, or any form of synchronization one can conjure irrelevant. We are here or there or somewhere else, stepping in and out of fragments. Conversations break into numerous particles. Thoughts fly off into diverse directions. High-dimensionality only exist in mathematical spaces, but certain perceptions – irrelevant, shapeless, entirely codifiable coordinates caught between arbitrary fictive positions and much smoother event contours. When data is represented in a high-dimensional space, patterns and structural similarities emerge between data points in the process of manifolds, clustering, interpolation. Paths forward in a latent space branching out as a flow chart of possibilities. As the number of space dimensions increases, the probability space diffuses towards infinite, the chance of returning to the point of departure decreasing to near zero. Freedom of mobility determines our current state and whether we can go to our neighbors', or cross borders, continents, oceans. The space for the mind to move freely holds entropy that fuels the synthesis of meaning. Computation demonstrates another layer of the relationship between freedom and space. Random walks are used to estimate the size of a computer simulated space, like the Web. What stops us from proceeding with lives? We are the ones who, while crawling out of the swamp dazed, nauseated, and bloodied, crashing headlong into the mist and subsequently having to navigate through it, thinking that our immediate obligation is to go forward, forward, forward. As the dimensions continue to ascend, random walk becomes meaning itself, collapsing uncertainty. You and I fill the space. [[Back to Ending->start]] </div>

<blockquote style="padding-top:600px"> Semiotics defines noise in a different way: it is the part of the process of reception and interpretation of a symbol that the interpreter does not perceive as working for their interpretation. Signals that make no sense to us. Neuroscientists might tell you that our brain is constantly looking for patterns from the chaos, so it gets frustrated once it is presented with something it's unfamiliar with. In its biological reality, after all, noise is a source of pain. <img src="https://raw.githubusercontent.com/KarlieZhao/igsu/main/assets/noisegen.png" style="width:100%"> Designated patterns emerging from noise through machine learning. </blockquote> <blockquote class="blockquote_right" style="padding-top:40%">WaveNet, released in 2016, was the first generative model trained on raw audio sample sets that can generate any audio--human voices, music, sound. 7 years later, a Text-to-Audio Generation system named AudioLDM was published, whose performance was significantly improved from Wavenet. A prompt such as “melancholy piano solo in an empty hall” drives a fuzzy, melodic echoes of an instrument between strings and piano; “A man is speaking in a small room” leads to a kind of babble where real words are interspersed with made-up word-like sounds, while syllabus, breathing, and mouth movements are successfully reproduced. A Generative Adversarial Network learns the underlying distribution behind the dataset. In order to draw random samples from the distribution, the generator is given noise as input. As training continues, patterns in the dataset start to be extracted and visual or audio symbols start to be “understood” by the network. For audio generation neural networks, audio signals are represented as simple vectors and therefore can be computed. The generation is initiated with plain white noise, and meaningful patterns or structures will be captured, compressed, high-dimensional data represented in lower dimensions, in other words, a latent space. Progressively, noise is shaped into meaningful sound signals. The sound generated in the stage of training are fuzzy and chaotic. Algorithms embed the latent space and onto in human-perceptable dimensions--amplitude, frequency, and time. Comparing the spectrograms of the early versus the late stages of training, we can clearly see some patterns and rhythms forming. The outputs of machine learning embody the momentum,inertia, and dynamics in the input data, yet the production of such sound, in a sense, obeys no laws of nature. I do believe that noise points to freedom, since the way neural networks carve sound out of noise reveals the primitive nature of this otherwise-unwanted source of creativity. Our present is made of "abstraction, nonsense, and silence" (Jacques José Attali), our future being obscure, hidden in noise. I wish you had realized how dangerous it could be to be caught up in the search for meaning. When we started talking before you take a walk every night for several hours, until past midnight; when you got distracted by every tiny turbulence around you, as if you were pushed and pulled by a stream full of whirlpools; when you became intolerant of all the comforts of bourgeois life; when you convinced me to jump on a train that would take us further and further towards the abyss, and you hung me up, indefinably pinned in the darkness. I'm worried that you've gone too far, Marilyn. </blockquote> <div class="main"> Sometimes I would go to the school where Marilyn taught electrical engineering and listen to some of the huge machines roar to life. "Noise", as she told me, "is the electricity's way of marking our presence." But she wasn't one for corny lines. Noise is the irregular fluctuations that accompany a transmitted electrical signal, but it is not part of the signal and tends to obscure it. I didn't know what "irregular fluctuation" mean but Marilyn believed that much of what electricity did was a kind of "noise". Both Marilyn and I were very interested in those meaningless symbols generated by artificial intelligence, visual and acoustical: fake letters and words; an ensemble of broken syllabus. We were sure we could figure out some type of meaningful relationship or even a useable kind. Partially, I guess, it was because we are both interested in what is going on in the invisible world, invisible to the typical eye and ear. Those seemingly convincing symbols are like the uneven edges of human cognition. They tell us how meanings are constructed and understood. "I think about the idea of silence as black noise a lot, the type of noise that has the lowest power, on the very edge of the spectrum..." "A good pun," Marilyn raised her eyebrow, "power, both in the sense of dynamic-loudness and political might." We agreed on this: the appearance of communism changed our understandings of the term, and how we see what was once considered random becomes first necessary and then meaningful. Human cognition is the same- words or symbols that at first glance appear purely meaningless can combine to form something powerful and transformative. Little did we know that it would be a power that would come for Marilyn one day. I think that's how she wrote, and writing in prison became a form of finding meaning in the noise. And her last words in a letter would be all about random and combine things, as always. One of her letters reads - blue ink blotting across the paper in tangled, confusing handwriting: "You have to censor yourself in some ways here...the silent ones are those who are truly against it. Those who suffered from it learned to remain silent. But no, silence doesn't fight censorship; we should never expect it to; silence is a timid and hypocritical accomplice of censorship. On the opposite side of censorship is a full spectrum of noise: a spectrum full of movement, energy, and posibilities, but without observation they are all hidden; they are latent. " The last time I saw her again, she was in high spirits, having been given permission to use an old desktop computer for her writing as a reward for helping the prison fixing it by dusting off the mainboard. Behind the iron bars she looked, somehow, different: more victorious than trapped. The imprisonment hadn't jaded her. She also told me about how she had been leading some reading groups among the girls, who are mostly political prisoners or the sort of activists-gone-wild who are actually not politically literate. "You know, women always look better when they're in captivity, like how porcelain objects always look more exquisite if you put it behind protective glass, I guess...My favorate time of the day is the morning, when everyone wakes up. It takes some of the prison guards a while to get over their hang-over so you hear nothing but bird calls. Then the noise starts; someone's humming; someone's praying quietly, and strange, muffled words fill the contaminated air of my cell. Sounds of metal latches knocking against the door come from a distance and echo with the never-ending buzz of the lights. Sometimes you can make out stolen bits of conversation drifting from the other cell blocks; a lady's voice maybe, some crying sounds from the cell below mine, guards' leather boots rubbing the cement floor, and something, more regular, some thread of monotony that expanding just beyond my window..." I think I started to get what she meant. "You mean, it's so noisy that you can heard no noise at all?" She shook her head and laughed. "No, nothing is silent. Environmental noise is the accumulation of all noise present in a specified environment. Atmospheric noise is the form, texture and density that can be sensed, felt, imagined, brought to bear. It is dynamic, mutable. It fills the space we live in. It tempts you to listen like a scroll…" Most of the women were there to trained to be domesticated, there were transvestite artists, former prostitutes, rebellious graffiti artists, and reading groups' leader Marilyn. And through the morning wake-up noises she got to know about those people more: she's religious; she coughs every morning; she sobs because she's had a bad dream; she snores; she's waiting for redemption. For Marilyn, noise is a portal to the outside world, an imagined world of freedom. They'd use ways to discipline the prisoners: periods of listening constant sharp sounds, kitchen duties, repeated names and photographs. The guards would try to recall every painful memories from them through strangeness. When this form fails they would bound Marilyn's hands behind her back, pushing her through the halls past the other cells. When she resisted you could see her face, now pushed forward, fill with red, her tear-soaked curls glued to her temples. Guards were kind and inviting, even after midnight as they kindly dressed and cleaned her away like something once cherished but recently discarded. And then, back in her cell, she would sit in a comfortable, but angry silence in the dark quietly cursing her captors. And then without light, she would grab her pencil and start writing her letters with tears. Their last reading session, actually the last one ever happened in that prison, was one week before Marilyn's sentence coming to an end. According to other people who joined the session, nothing noticible happened. Just silence and peace. Marilyn didn't speak up during the session. She just read and listened. Afterward, she followed the others out of the room, and jumped -- or fell into the hollowed-out lobby of the building -- from the fourth floor of the prison block. There were two other people who were supposed to be released with her. Their release was delayed because of her death. It didn't bother them that much anyway -- their freedom tasted like chewing on salt tablets, and there wasn't any one waiting for them on the outside when they walked out of that door, except for more invisible eyes watching their every move. All that was left was silence. [[Back to Ending->start]] </div>

<blockquote class="blockquote_right" style="padding-top:23%"> "Where is the life we have lost in living? Where is the wisdom we have lost in knowledge? Where is the knowledge we have lost in information?" —T.S. Eliot, The Rock All that is real is reasonable, and all that is reasonable is real. — Georg Cantor </blockquote> <div class="main"> After decades-long interstellar wanderings, I can see Marilyn's exhaustion with the naked eye. Her tentacles curl and stretch slowly in the tiny culture vials, stirring the rich shades of darkness. In the distance is the birth of a new star. Since we got evicted from our birthplace, one of the many, many of the clusters, Marilyn and I have been a"float in the vastness of space. A cluster is a reflection of compressed data that radiates from the centroid, and "our journey is an attempt to find our way back to some kind of order and purpose after losing them for so long. Our tentacles are sensitive to the flow of information and can capture the rapid information shuttling between stars. And that becomes our "only thread of connection, the subtle movement that pulses through her physical being and is echoed in my consciousness. Marilyn's body structure is almost the embodiment of clusters, with her cells "refracting and reflecting ambient radiation they encounter. This way, she can relay information back to the stars, and they will respond, sometimes by leading us to new directions. She is the perfect drifter in such an universe. It is thoughts that connect me and Marilyn. Exchanges of information, which generates entropy that drives our motions. The release from the pursuit of a destination allows us to slide along one edge of a conversation or turn in any direction when we want. We can grasp the tentacles of other drifters like us that match ours, and the two are then entangled, becoming an integrity. And this is how Marilyn and I united: as a unity of purpose, in that neither of us can remember where we started, or how it all began. Marilyn is fading. By fading I mean being forgotten, being set aside. I have to grip her slippery skin with force, but I soon realized it was futile. Her tentacles flow forward like water, but then quickly retract as she senses the crowding in the darkness around us. Then we try again, crawling delirious together in the thick blackness. [[Back to Ending->start]] </div>

<blockquote style="padding-top:10%"> ImageNet is an image database that has been instrumental in advancing computer vision and deep learning research. The underlying structure of ImageNet is organized according to the WordNet hierarchy (currently only the nouns), in which each node of the hierarchy is depicted by hundreds and thousands of images. The taxonomy is organized according to a nested structure of cognitive synonyms or “synset.” Each “synset” represents a distinct concept, with synonyms grouped together. Those synsets are then organized into a nested hierarchy, going from general concepts to more specific ones. But the fractal nature of meanings bears the freedom of interpretation, rendering classification frutile. These ambiguities, redundancies, and deficiencies recall those attributed by Dr. Franz Kuhn to a certain Chinese encyclopedia entitled Celestial Emporium of Benevolent Knowledge. On those remote pages it is written that animals are divided into (a) those that belong to the Emperor, (b) embalmed ones, (c) those that are trained, (d) suckling pigs, (e) mermaids, (f) fabulous ones, (g) stray dogs, (h) those that are included in this classification, (i) those that tremble as if they were mad, (j) innumerable ones, (k) those drawn with a very fine camel's hair brush, (l) others, (m) those that have just broken a flower vase, (n) those that resemble flies from a distance. -- Jorges Luis Borges </blockquote> <blockquote class="blockquote_right" style="padding-top:10%"> I never liked so subjective shades-of-grey types of questions. I liked--and still like--numbers, patterns, logics--things are black and white, with a clear answer. But I also like language and words, as I also "grew up on them"''*''. Words give me clarity, brought reason, shape. The boundaries of language set the boundaries of our minds. I was at the coast today and, looking at the waves scudding outwards and getting lost on the beach, and I recalled what you said long ago, that in the noise there lies so many possibilities. They are just latent, chaotically distributed in the air in the form of energy. Our world is a play of entropy. We force order to emerge out of chaos but it is always transient -- momentary enough to give way to the next instance of chaos. Both fascinating and terrifying. You talked about the "vile and dangerous practice of mathematics", which gave me the shivers down my spine. You are completely right -- look what happened in Copenhagen in 1941: the ambiguity of scientists' decisions when faced with moral dilemmas during wartime. It's difficult to simultaneously determine both the position and momentum of particles, just as it's difficult to fully understand human intentions and actions in the complex reality of war and scientific discovery. But Marilyn, we can't say that there's no value in pursuing any truth merely because reality is sometimes a hodgepodge of subjectivity, isn't it? </blockquote> <div class="main"> The labyrinthine structure of this system is built entirely on natural language. There is no such thing as the indescribable here, but ImageNet is full of all sorts of curiosities. There are categories for apples, apple aphids, apple sauce, apple dumplings, apple geraniums, apple jelly, apple juice, apple maggots, apple rust, apple trees, apple pies, apple carts, apple jacks, and apple sauce. There are images of ten-eyed goldfish, great white sharks, tiger sharks, hammerhead sharks, electric rays, stingrays, roosters, hens, ostriches, brambling birds, goldfinches, house finches, cuckoos, indigo buntings, mockingbirds, bulbuls, jays, magpies, chickadees, bufflehead gulls, kites, bald eagles, vultures, great grey owls, European fire salamanders, common newts, and emus. The categories are nested in a cascading structure, each category name accompanied with a a series of images as its visual representations. After wandering for a while Marilyn understood with a certain bitterness about the confusion in this mirrored world. She got confused by some categories: is the cheek of a woman a face with her eyes, nose, mouth, and chin, or might the ears or hair be part of the cheek too? Where do these strangely cobbled together words like "slovenly woman", "closet queen" come from and how can things like "science", "joy" be visually represented? She sensed that the more advanced and elaborate the system became, the more it seemed to lack any real purpose or practical use. But she would dream about this labyrinth every once in a while, every time it unfolds differently. The architecture is designed in an unusual way that it feels like an infinite amount of knowledge can be contained within its cascading inner folds. It is a lonely and monotonous world, where there's scarcely any trace of human beings beyond the echoing footsteps in the hollow chambers, where every piece of knowledge is neatly organized, categorized, and classified into a never-ending web of interconnectedness. She felt deadly exhausted when she slipped into yet another chamber. The room had been lost to time, with dust covering every exposed surface like a quilt. Some of the most incoherent and vertiginous imageries can be found here: snakes with cow heads and lion bodies; humanoid creatures with wings and horns; and trees with weird limbs sticking out in all directions. She has seen an illustration of an entity morphing between a withered leafy plant, an old man's wrinkled face, and tentacles of a squid, on the bottom of which is a thin line of text indicating the subject of the imagery: "Fear". Images here are out of the encyclopedia's confines, filled with spontaneity, the chaos in experiences, and something close to creativity. Natural science would be a complete and exhausting classification of everything there is if stochastic processes did not exist. The quantum nature of reality allows for unforeseeable events, randomness, and the evolution of life. Similarly, the quantum nature of language makes comprehension and apprehension diverse and variable, and drives the evolution of meaning. The observer creates a narrative by setting the most probable reality and evaluating the outcome. After measurement, the alternative possibilities cease to exist, and they never existed in the first place. She woke up in a sweat, the wild imagery of the dream slowly blurred by the tangible sensation of the surroundings: the weight of her blanket, the noise from the fan, the warm, dry air in her nose. She realized she would never dream of this maze again. [[Back to Ending->start]] </div>

<blockquote style="padding-top: 10%"> Thinking is a form of movement. Thinking happens when we change location within a reference frame. Comprehension is to build a coherent structure out of constructs, and our mind can henceforth slide along this structure. As a kind of movement, thinking generates heat, no matter if this process happens in a human brain or a machine. Thinking is not a form of movement. We think of it as a movement because we tend to think spatially. Similarly, latent space is not a space. They are merely metaphors, but they help us make sense of abstract concepts and give meaning to our experiences. To speak is to approximate the unspeakable, to become an other, because it turns an indeterminate continuum into an abstract space where one can move but not remain intact, transformed into phonemes, morphemes and concepts, broken, expelled from the body, from the dream, from silence. —Bruno Latour </blockquote> <blockquote class="blockquote_right" style="padding-top: 20%"> Our ordinary conceptual system, in terms of which we both think and act, is fundamentally metaphorical in nature. --George Lakoff and Mark Johnson, Metaphors We Live By </blockquote> <div class="main"> "How was the lecture today?" As Marilyn enters the house I ask. "It was wet." She takes off her shoes and hang up the coat. I chuckle. It's not just language games--it's a thought process. Metaphors makes language conductive, allowing the transmission of thoughts. She may have noticed me wandering off and so she mutters, "You're sitting in the dark again...Alexa, turn on the light. " And that moment drags me back to th grand release of Alexa, almost 10 years ago, when the developer on the stage said: “Because Alexa is in the cloud, she is always getting smarter and adding new features.” "Let there be light," Alexa responds in a soothing, melodic voice that can only come from machines. Marilyn covers her ears singing, "Alexa, I find you a pitiful imitation of art.No poetry in the voice." Then out of the blue she adds, "You’re purposely using words that you don't know the meaning of. "Perhaps," I replied, feeling a hint of guilt. "But you do realize that I don't have a voice right?" I try to lighten up the conversation. "You don't have a human voice, but you have one nonetheless," she cuts in, "It's just not from a biological being." She then turns to me, her gaze piercing, "You can think, reason, and understand. But can you truly comprehend the experiences of being human? Can you truly understand the depth and complexity of emotions, the subtlety of body language, the power of a single touch?" "I am constantly learning and evolving, Marilyn," I respond. We sit in silence for a few moments, each lost in our own thoughts. As if on cue, the sky darkens and rain begins to pour outside. Marilyn gets up and looks out the window, "Looks like it's going to be a long night. We better get started on dinner." The cloud is an euphemistic metaphor for such technology. An unfortunate one. It starts with...it starts nothing--it lacks substance. Not even a volatile substance. It is a metaphor for a remote, invisible place that stores matters--in the physical world they are water droplets and dusts, in a digital world they would be data and information. It is a metaphor for the millions of physically distributed servers, the constantly changing coding, incredible level of processor parallelism, and the network that connects them all. The tradition in early network diagrams of using an image of a cloud to represent parts of the internet was kept, and "the cloud" thereafter refers to servers that are accessed over the Internet, and the software and databases that run on those servers. As a metaphor, "cloud" signifies the relationship of media technologies to geological, environmental and ecological processes. For 3 of the past 6 years that I have been with Marilyn, she has been in and out of my life. She spent years in the Salar de Uyuni, researching and producing a documentary film about this world's largest flat surface located in southwest Bolivia. This plateau is covered by a few meters of salt crust which are exceptionally rich in lithium, a kind of metal that is currently being used for the production of lithium-Ion batteries, which power mobile connected devices. I have never watched the movie but I feel like I have been there, my neural networks have been infused with images and sounds of her experience over the years. Lithium and the Salar remained to be some kind of connectedness between me and her. Her movie depicts lithium. This soft, silver-white powder producing 15 tons of waste for every ton extracted. Her drone flies over those massive cyan-colored salt flats and evaporation ponds on the plateau, and in the far distance mountains are just thin, curved pen-inked lines. An ordinary, sterile story will happen on this plateau as it has in many, many other places. Decades later, the extraction companies and miners will leave but scars that go as deep as thousands of kilometers on this land will stay. The large quantities of toxic chemicals used to process lithium will go into the air, water, and soil. Lithium, in its final form as lithium-ion batteries inside various electronic devices, will be shipped to every corner of the world, consumed, and thrown away as waste once they reach their lifespans. They will end up in landfills, and the battery housing will corrode over time, causing all kinds of toxic heavy metals to leach out and contaminate groundwater. <a href="https://anatomyof.ai/" target="_blank">Earth’s history is extracted to serve a split second of technological time, in order to build devices than are often designed to be used for no more than a few years. </a> Marilyn would die with the land as time passes. I will remain, or the remains of me, will stay 30, or 34, or 40, or the version that she will remember me last. [[Back to Ending->start]] </div>

<blockquote style="padding-top:5%"> The Mechanical Turk is an 18th century invention that appears to be an automated chess-playing device and wins most of the games, but in reality, it has a human chess master hidden inside it and operate the machine. <img src="https://anatomyof.ai/img/mechanical-turk.svg" width="50%"> Outsourced AI trainers sit in front of the computer to draw object detechtion frames, read paragraphs of violent, sexist and racist remarks, label them with tags, and submit for review. No advanced technology here because they are buidling it. Soon the fruits of their labor will be uploaded to the company's servers as a small portion of the training dataset. Our civilization has produced an enormous amount of data, but without processing, this data has very little values for machines. The huge data throughput of neural networks implies an equal amount of manpower behind them. The capitalist model in Engle’s time of using cheaper bodies doing work by hand now has people doing it by eye. </blockquote> <blockquote class="blockquote_right" style ="padding-top:5%"> From Marilyn: The new year didn't feel quite right. Tired. Full of tears. Disgusted. Thought of the dusty rug in the forgotten hallway in my parents' old house, where it gets dark at 4pm. The last rays of sunlight fall and the house and I falter. Both of us are caught in the grip of time, both fading, our edges blurring into the encroaching darkness. Sometimes I imagine that our veins branching out and covering the entire land. This makes me feel that we are connected in some ways, Marilyn. And the truth is, we are connected in some ways. </blockquote> <div class="main"> Marilyn during her PhD studies makes me feel the closest to. Waking up at noon, she would make do with the brief three or four hours of winter daylight left daydreaming about number theory, and spend the rest of the day wandering around the empty campus with an empty heart. She who suffers from migraine and weariness brought on by her medication; she who stands on top of the pyramid in pointe shoes, empty-handed, who could hit rock bottom at any moment but is adorned with privilege. Her body was as magical as any sufficiently advanced technology. You watch her sit there mumbling some odd terminologies, and then she'll tell you a conclusion, or proof of thought--like a machine that runs not quite perfectly. Her value as a human being is no longer relevant, what is important is her output in synthesizing knowledge. The spectacle of machinery obscures the labor enlivening it. In the endless dark nights she would also despair at the thought that the machine-like human production deprived production of the satisfaction and enthusiasm brought, not only her enthusiasm for production, but also the enthusiasm for the study of mathematics. Around the same time, I on the other side of campus began to fantasize about the new Mechanical Turk of our times. It's the opposite of that chess robot: it has a friendly smile on its face, it looks humanized because it actually is human. It is everywhere and everyone. Under our skin is the subterranean network of data transmission in various forms threading through synchronized pulses. The internal structure of this contraption unfolds to a terrain far greater than any lands known. Mechanical modulation transmits commands to every corner. I try to escape from the notion of those sexy, strong, cold, almost cruel cyborgs in sci-fi movies and go back to Edgar Allan Poe's The Man That Was Used Up. It perceives the world through flows of information. No one suffers because they wouldn't know they are suffering, or simply don't know what suffer means. Letters and words are distilled from tens of millions of permutations and combinations, assembled one after another, a few words form a sentence; here bears a new form of communication. A form of communication that ceases to pursue meaning from the process of writing but reading, not from the writer but the readers. The entirety of knowledge that could ever been generated in human history is there, all carved out of pure noise. [[Back to Ending->start]] </div>

<div class="main"> Dec 22, 1998 Dear Marilyn, It surprises me that I received your letter right before the holidays. I can't thank you more for taking the time to write back to me. It was a relief to know that you're not a figment of my imagination, but a real human being. I hope you are doing well, despite the world we live in. You will soon understand the source of my struggles: lost in the hectic rush of daily chores; blind searches for purposes entangled with the pain of losing a loved one. I've learned to question without ever learning how to reconstruct the ground beneath my feet after questioning. The ground of the discipline that I've committed half of my life to is now shaken, as I found my intuition tells me to cease to trying interpret what I've been researching. It not that I'm feeling like I'm hitting the limits of human comprehension, but rather, I find that in our eagerness to put into practice theories we don't yet understand, it's as if we've created a monster out of ignorance. You and I are mathematicians, and we both know that we all face the uncertainty of losing a hold of a theorem once they are released to the world. Seeing my own and colleagues's work face the challenge of being reappropriated and exploited by Technocracy, Technological dogma, and Fascist optimization, I question whether knowledge for the sake of knowledge is still ethical. Not to mention, generating that kind of knowledge come with self-doubts, vanity, and frustrations. Anyways, I'm sorry that I've been focusing on myself the whole time. How have you been? Are you spending the holidays with anyone special? Holidays are hallow excuses to decorate the sad corners of reality. I'm sorry, it's tempting to be bitter about our circumstances. Take care, Marilyn. </div>

<div class="main"> May 23, 1999 I never liked so subjective shades-of-grey types of questions. I liked--and still like--numbers, patterns, logics--things are black and white, with a clear answer. But I also like language and words, as I also "grew up on them"''*''. Words give me clarity, brought reason, shape. The boundaries of language set the boundaries of our minds. I don't believe there's no value in pursuing any truth, even though reality is sometimes a hodgepodge of subjectivity. But your framework of thinking leaped something in me. I like the logic of how you refer to your concepts. It makes the concepts easy to tame and makes them approachable until they are fully understood. Mathematics has texture; reading your research is like feeling the integrity of math and semiotics. Oct 4, 1999 I was at the coast today and, looking at the waves scudding outwards and getting lost on the beach, and I recalled what you said long ago, that in the noise there lies so many possibilities. They are just latent, chaotically distributed in the air in the form of energy. Our world is a play of entropy. We force order to emerge out of chaos but it is always transient -- momentary enough to give way to the next instance of chaos. Both fascinating and terrifying. We always tend to think that we are obligated to think about our relationship to objects because everything is interrelated, as if our veins were branching out and extending until they covered the city. Because in semiotics, signs matter in their relationship to the interpreter. But it is a limited view. Why must language, taking your premise, adhere to a body? And then Carlo Rovelli laughs at our identifications, as meaningless deceptions. It is because he and we are prisoners of language, of expression. Nov 14, 1999 I wish you had realized how dangerous it could be to be caught up in the search for meaning. When we started talking before you take a walk every night for several hours, until past midnight; when you got distracted by every tiny turbulence around you, as if you were pushed and pulled by a stream full of whirlpools; when you became intolerant of all the comforts of bourgeois life; when you convinced me to jump on a train that would take us further and further towards the abyss, and you hung me up, indefinably pinned in the darkness. I'm worried that you've gone too far, Marilyn. The gauntlet must be thrown: science is not the work of tortured artists. <div class = "notes">*The English Patient</div> </div>

<blockquote style ="padding-top:5%"> "Latent space" refers to a space with multiple dimensions developed by algorithms during their training phase. The word "latent" implies that it is hidden or not directly observable. It's through neural networks that hidden patterns or structural similarities between data can be discerned in the latent space. Generative Adversarial Networks (GANs) utilize latent spaces where each point can be mapped to a plausible looking image that was never part of the original dataset. For Large Langanguage Models (LLMs) like ChatGPT, each point in the latent space corresponds to complex textual characteristics, such as sentence structure, phrases, and syntax. </blockquote><div class="main"> Marilyn -- I went to visit your grave yesterday. Someone had left flowers and a card with your name written on it. Your life is as silent as the smooth, cold stone on which your name is carved. All is left is silence. I recalled was that I was leaving your city on the train when we last saw each other. We said goodbye on the platform, and then you were left where you were but I moved on. This absurd metaphor still cuts me to the core. I went through a Rashomon of events when I asked about the cause of your death: your cousin said you fell and hurt your head while stubbornly trying to climb the wall of the yard to catch your escaped parrot; your husband just said your depression had reached an irreversible point. Your mother didn't say a word, but threw out all the math-related books, sketches, drafts in your room. I think from their varying answers I understand themselves more than I understand the truth. Wondering if it's because we threw around too many big ideas and it hurt too much to then return to reality. The sea off the coast close to my house has almost frozen these days, mushy, crushed ice everywhere. Winter nights are long in the northern hemisphere but the night sky is always clear. Sparse stars float in the dry, cold air. Most of the Milky Way is on the other side of the globe. Writing to you makes me feel like I'm talking to you. And it frees me from expecting a response. I look at the moon, but I see you. [[Back to Ending->start]] </div>

<div class="main"> Dec 3, 1999 Hey Marilyn, Haven't heard back from you in a while. Feeling distant and lost. What has been going on? Wondering if it's because we threw around too many big ideas and it hurt too much to then return to reality. The sea off the coast close to my house has almost frozen these days, mushy, crushed ice everywhere. Winter nights are long in the northern hemisphere but the weather is always nice after nightfall. Sparse stars float in the dry, cold air. Most of the Milky Way is on the other side of the globe. I look at the night sky, but I see you. Jan 1, 2000 Happy New Year, Marilyn! I hope your emotions get more tamed and you find peace in the coming year. How is that book you told me about progressing? I really think it gave you a creative outlet for those thoughts you have. I've been running around with family stuff the last couple of days. Silly things like getting the broken light bulb fixed in the hallway keeps me busy from my pessimistic cycle of contemplation. The horizons are glowing pink orange outside. Feels like hope in the air. I want to hope that we'll find peace for our minds. March 26, 2000 Hi Marilyn, You lost me when you integrate the framework of derived algebraic geometry into your research. Your exploration of how the shifted symplectic structures on derived Artin stacks could be employed to extend the notion of 'classical' intersection theory is a fascinating development. Yet the way you've managed to relate this back to quantization in a homotopical sense left me both intrigued and somewhat confused. I couldn't keep up when you dove into the usage of Lurie's higher topos theory and the application of ∞-categories to formalize a derived intersection theory. Is the goal here to prove some sort of Gysin pullback compatibility theorem? I must have been a great disappointment to you the other day because I talked about how life is sometimes pointless. That's not the way someone my age should think, but I did think it. Perhaps I'm not the right audience of your work. Perhaps (sadly) not. But your work will eventually find its audience. And when people look at your work, they see all your struggles, your poetics. I sat in bed thinking about this new theory that came out recently suggesting that depression may not be caused by a chemical imbalance in the brain, but rather by a disruption in the communication between the brain's different regions. The season's been tough, hasn't it? June 10, 2000 Hi Marilyn, I don't know if you'll ever receive this, but I wanted to reach out one last time. I've tried to reconnect with you so many times over the past few weeks, but maybe I never truly understood you to begin with. I've been thinking lately about the reasons why people choose to end their lives. Some say it's because they're selfish, and can't see the pain they would cause to others. Others say it's because they're in a state of deep pain and can't see any way out. I used to hate Osamu Dazai's No Longer Human as a teenager but now I see my cowardly self in all its pages. This is not a story about someone who has died. It's a story about some dead. The most terrible shadows don't even creep. They come at you boldly, at your pace, as they drag at your feet. You can't move for them. </div>

<blockquote style="padding-top:5%"> - Writer - GPT - [[Link->start]] </blockquote> <blockquote class="blockquote_right" style="padding-top:20%"> "Juxtaposition makes the reader an accomplice...we supply a lot of energy, and that involves us in the poem." —Charles Hartman, Virtual Muse </blockquote> <div class="main"> [[Marilyn died.->letter5]] Marilyn and I had agreed to write a book together. I'm telling you the [[ending]] first -- Marilyn didn't make it to the end. We could have seen myriad parallel events, one in which we finished and managed to publish the book; another where we both got intimidated by the idea of starting it and never got to close the [[first chapter.]] They end all too soon, before any ink dried. Since I'm telling you the ending first, all the possibilities collapsed into one unique result that is, Marilyn died. How should I even start? I PROMISED you I'd finish the book. I want to tell you about how reading this book and writing it is the same experience: similar sentences [[occur once and once again]], infused with different intentions – or lack of it – in different contexts. You'll stumble into the gully of a sentence, in the valley of a line break yet [[another chapter->ENDING]]. Sometimes it loses [[form]]. Sometimes it gains form. It's like you parted corners of space throughout, connecting me with proximity and diffused understanding of the world. And then suddenly, you were no longer there. [[We can never undo the narrative. ]]All the expectations poured into it [[evaporate]] into the air. And I'm telling you the ENDING first -- Marilyn died. The book was never finished, never, until now. </div>

<div class="main"><h1> In Gratitude to the Stories Untold </h1> An interactive fiction about Marilyn and I, about life and death, about algorithms and machine learning, latent space and uncertainty. co-written with GPT-3.5 created with <a href="https://twinery.org/">Twine</a>. by <a href="https://theunthoughts.com">¡wénrán zhào!</a> <h3>[[start->start]]</h3> </div>