Atlas explores the explores the notion of authorship pertaining to the perception of place. History would have sounded different to us if it were recorded from a different perspective, but with this piece I wanted to question how the history of humanity would have looked to us if an AI had analyzed sets of data in the past. In a recent Stanford project, computers scanned millions of pictures of parked cars from Google maps to predict geographic information in cities of the United States. The computers were fed millions of snapshots from Google Street View and shockingly, accurately predicted income, race, education level, voting patterns and even pollution levels at the ZIP code and precinct level in cities across the country. Through the computers analysis, the AI came to the conclusion that Chicago has the highest level of income segregation in the nation. Such information is imperative to political constituents. It has the ability to raise awareness of the disparities in the Chicago community, and lead to political reform. According to the New York Times, it would take a person 15 years to analyze the data a computer processes through AI learning in two weeks.
I’ve walked through and photographed the areas projected on the screen in Chicago before looking at them in Google Street View, and my version of the location looks entirely different from what Google Street View cameras captured. If I were to show my photographs of, or write about Chicago, my viewpoint would be influenced by biases formed in my upbringing, life experiences, and intersectionality would greatly influence my description and point of view. My understanding of my surroundings will always be different from another human experiencing the same moment. But there are potential dangers the advancement of AI technology might hold due to biases the AI “trainer”. Some scientists believe that, with enough exposure to widely representative databases of people and places, algorithms' bias issue can be eliminated. Much like a person is more likely to hold less bias if they have been exposed to other ways of living, and different types of people. As AI technology advances, it is imperative to humanity to responsibly raise it almost like a child in development in order to avoid bias in the analysis of data that will shape future generations views of the world.
I’ve walked through and photographed the areas projected on the screen in Chicago before looking at them in Google Street View, and my version of the location looks entirely different from what Google Street View cameras captured. If I were to show my photographs of, or write about Chicago, my viewpoint would be influenced by biases formed in my upbringing, life experiences, and intersectionality would greatly influence my description and point of view. My understanding of my surroundings will always be different from another human experiencing the same moment. But there are potential dangers the advancement of AI technology might hold due to biases the AI “trainer”. Some scientists believe that, with enough exposure to widely representative databases of people and places, algorithms' bias issue can be eliminated. Much like a person is more likely to hold less bias if they have been exposed to other ways of living, and different types of people. As AI technology advances, it is imperative to humanity to responsibly raise it almost like a child in development in order to avoid bias in the analysis of data that will shape future generations views of the world.