Ben Irving

9/17/23

Janus

        My breath is short. Each step feels like a hundred, the pack heavy on my back. It is the last stage of our 3 day journey through the Sierra. The granite, the sparse treeline. The deep blue of the lakes we climb past. These are familiar sights, comforting ones. I have moved through these mountains my whole life. Yet still the altitude weighs heavy. I smile, feeling roots carved into the narrow trail pressing at my boot soles. Brambles tickle my calves, the breeze a gentle breath in the setting sun. The valley spreads out below, dark granite cascading to a bed of forest and meadow. Rivers dance in the evening light, calm from far away, ribbons of blue easing towards the horizon.

        My mind is not on the trail. I am immersed in a journey, a daydream. Not of the valley below, but of a city at the edge of the sea. One that does not exist. Janus, I name it. The god of new beginnings. I plan it out, each new impracticality exciting a blossom of imagination.

The roads are below ground. Cars on the surface have always disgusted me, in their noise, their gracelessness. The petulance of the humans that drive them. Janus would shove any that are necessary below the ground, and remove the human component. Stairways into the underland would have to be everywhere for this to be in any way practical. Another astronomical expense, which would increase with the introduction of silent, autonomous trains in the ground and sky. Along with other public transport probably.

        The forests, rivers, and meadows would fill the space roads would usually eviscerate. There is a chaos in nature that I have always regretted the loss of, in our stark, organized concrete jungles. I step over another stream on the trail, watching the water rippling over the rocks. There is a speed to rivers, a destructiveness, in the way they cut through the earth, changing it. Peter out, burst over their beds. Allowing for such erratic behavior in an urban environment would lead to a lot of issues, obviously. But I am attached to that idea of change. Cities are far too stagnant, buildings in particular. So I imagine structures that shift and grow with the seasons, which move as the rivers rise and fall. I imagine systems to watch them, to monitor their health, their metamorphosis. Smooth, formless, more window than wall. Built to be shoved, to be poked. To be molded like clay. But strong enough to withstand things our skyscrapers today would crumble beneath.

        But I do have a soft spot for old things, and the whole place can’t be some hyper modern utopia. So in some of the outer districts, I imagine small alleys, quiet little streets, flavors of Provence, Kyoto, Venice. Pockets from my favorite places, blended and woven into the fabric of Janus, along its rivers, through its forests and meadows. Complete with the cuisine to match, of course. But without the grime, and with flavors of the new. Cleaner streets. Quiet bikes, scooters. Little driverless trams humming along. Small, autonomous, trash-collecting carts, picking their way along the paths. And a slow blend towards the inconceivable architecture of our future, as one gets closer to the center of town.

        Another thing I tend to despise about cities is their desecration of the night sky. The stars are an incredible gift. No towering monstrosity, awash with pulsing grotesqueries, is worth losing sight of the milky way. Candle light would return, by mandate or some sort of cultural expectation, after midnight. The rest would disappear. Safety issues would be handled by some sort of autonomous drone, or buggies on the ground, which don’t need a bright city light to spot a lost soul.

        But the joys of the late night can’t be lost. Sometimes one needs pulsing light and the deafening chaos of a hundred drunken degenerates. So my imagination carves out caverns below the Earth, huge, glistening with strange stalactite formations, glowing with purples, blues and reds. Deep winds from beneath the earth would keep a cool, balanced temperature, hovering around 65, 70 degrees fahrenheit. Massive underwater lakes, fed from aquifers buried into the rock. Huge walkways would stretch through the caverns, draped with bars, clubs, and other hedonistic delights, supported and monitored by autonomous stress detection systems, and some large-scale computer vision models. Maybe some lidar, to have a hyper-realistic mapping of the environment for better response time in case of incidents. Massive skylights would be cut from the ceiling, the only way in which the light of sin and pleasure would filter to the surface. Tartarus, the district is called. An occasional glowing reminder of the below for the pedestrians above, walking through the forested paths or along the meandering rivers towards quiet laughter, music, and streets from another age.

        As you might have picked up, humming through Janus, powering its public transport, supervising the wind turbines and river generators, solar panels and nuclear plants, directing its sewage networks, irrigation systems, and cleaning its trash, would be AI. Compute, designed to watch, to observe and moderate the natural balance of the city. A passive presence, which would reach into the lives of everyone. Janus would be alive, in body, and in mind. An organism, of which we would all be small parts.

        I think of the sorts of models I could even conceive of deploying to meet the vast needs of my fantasy. The scale would be immeasurable, as would the data required to make even the simplest water filtration reinforcement learning algorithm competent. These systems, whether it be a language model assisting in the classroom, a convolutional model serving as the eyes of the tram, or some not yet conceived of generative monstrosity trained to design and monitor structural supports in bridges city wide, would all interact in some capacity. They would benefit from each other's pretrained weights, aware of if not directly affecting one another's hyperparameters. And they would be built using the same infrastructure, language, run on adjacent hardware. Would the inference be local? Or through the cloud? Edge computing is leagues faster, so it would be cool to have it all on chip. You would want your farming robots to be somewhat independent of a cloud network, so that they could continue to produce food if the system ever came crashing down.

        But these specifics slip through my fingers as I round a bend in the trail, coming upon Max and Raf sitting at a bend in the path. My best friends, with whom I love to share insane visions of the future, of the past, of parts of the world foreign to us. I prepare to launch into my pitch for Janus, and the AI that would watch it, power it. Bely it.

        Since high school, that has been where my fascination with AI has stemmed. Its potential to provide the inanimate with an avenue to observe the world, and to occasionally poke at it. A benevolent in between, a step towards the omniscient, but a step away from the manner in which we think about gods. Not a builder, a breaker, or a ruler. Instead, a watcher. A listener. A silent guide, a deep silence. Which prods the world, quietly.

        I remember when I landed on CS. Going into senior year of high school, engineering and physics had been where I wanted to place my efforts. It seemed like the logical choice, a platform from which I could launch into any field. The math in physics was also one of the few things in high school that sparked my interest. I did love to code, however. Building games is where I started. But that didn’t seem like the path to the sort of diversity of subject matter I imagined. Engineering seemed to be more and more specific, forcing you to a major which would lead towards a smattering of possible industries (this turned out to be false, but I wasn’t doing as much intensive research as I should have been). So physics it was. But then I found ML. Andrew Ng is a Professor at Stanford, who pioneered the online AI education effort, through his free course on Coursera (a website he launched). Watching his grainy videos recorded in some dingy little study, coding through the simple exercises in the Octave programming language. That’s how I began. The gradient descent unit is emblazoned in my mind. A revelation, of sorts. An equation that learns, that changes dynamically in favor of performance? It was pretty cool, watching the loss decrease and the accuracy rise.

        As I read more and more about machine learning, I began to realize that this was the true launch pad. A road towards reaching into a thousand industries. Into every industry. I could build something to poke the world. So I decided on CS. An intersection, where I could pull at all the things I love. Where I could build datasets to map the health of mycorrhizal root systems in forests near home, design robots to pluck grapes from a vineyard, or fire rockets into the sky. To play with crisper, or the human brain. A field which could allow me to exercise huge degrees of dilettantism, to varying levels of effect. But all those different fields had a common strain: huge, complex systems, which no human can understand completely. We can try to learn as much as we can, to watch, and put our ear to the ground. But in the end, all we can do is make little attempts to influence these things in small ways, in the hopes that momentum will carry us towards a beautiful result. Which, funnily enough, is how most machine learning algorithms really work.

        The end of our hike is more of a stumble than a triumphant march. 11 miles, in one day. No lunch. Exhausted, we begin to hear the water taxis cutting across Echo Lake taking hikers back to the parking lot. Casuals. The echoes of civilization, reaching into the natural serenity of the afternoon. The birds chirp, the sun dips ever lower. I breathe deep. As more people come into view, the cars in the lot, the cabins nestled in the hills around the lake, I grow sad as I always do. The human footprint is a bit jarring when you spend any period of time away. But here, it is quieter. More thoughtful. Beholden to the land, a part of it in a way. The cabins are rustic, the wood faded and gnarled. As if they grow from the roots of the great trees. That is what I want for Janus, for the new beginning. For the cities of the future. I want humans to exist in the knowledge that we are born from nature, from evolution. We are not sacred beings standing apart from the natural world. A chaos of factors delivered us into this moment, and one day it will rip us free of it. We are a step on a long path, not the destination. It is the same with AI. It should not be thought of as some entity which will stand apart from humans. It will be born of us, and it will be a part of us, just as we are a part of it. We will grow together, change together. Both humans and AI will one day be unrecognizable to present day conceptions of both of those words. Perhaps we will not be separate at all, but beings born of both. And I hope the same will be said of the human relationship to nature.