American Toxics

Abstract

There are nearly two thousand “Superfund” sites in the U.S., which are the most contaminated toxic waste sites being cleaned up by the EPA. Many of these sites have hundreds of pages of technical documentation available, but there exists no realistic tools by which regular interested citizens can learn and engage about their surroundings. I made American Toxics, a website with an easily navigable map of hazardous Superfund sites with clear summaries of their histories/impact on local health. To accomplish this, I’ve extracted dwindling documentation from government agencies about these sites, and used retrieval-augmented generation with an LLM to summarize the histories and answer questions. I then built a gallery show with a collection of physical installations around the database, visualizing this government program through sound, sludge, maps, photography, paper documentation, and a hanging mobile. It addresses the personal story of my preschool’s direct proximity to a Superfund site, the scale of waste across the U.S., and uses technical, design, scientific, and fabrication methods.

Introduction

America has nearly two thousand sites that are so contaminated with toxic matter they have been taken over by the federal government for cleanup, under the CERCLA designation started in 1980, also known as “Superfund sites.” New Jersey, California, and Pennsylvania have the highest concentrations of contaminated sites—all states where my family and I live. These are former and current ports, mines, landfills, factories, sewage treatment plants, power plants, etc. Since in most cases private entities caused the damage, the federal government attempts to identify those entities and require them to pay for cleanup costs, but public money is used as a backstop. Because of the EPA’s role here, the government has published extensive documentation on why sites deserve the Superfund designation, with the results of inspections, cleanup plans, and more. These documents are all available to the public, but are housed on websites designed for EPA employees and Ph.D. scientists, using industry jargon and acronyms and littered with scientific formulas, coupled with inscrutable visual design, running on high-bandwidth, low-accessibility websites. The information is available, but not realistically accessible, to the public. Moreover, many sites have hundreds of pages of documentation that take hours to comb through even for readers who can understand them. The result is that most Americans are unfamiliar with Superfund sites near them and the potential impact on their health.

The default tool we reach for to understand the geographic world is digital maps like Google and Apple Maps. But due to their creation being funded by massive technology companies looking to make ad revenue or license data for commercial use, our digital maps show landmarks one can drive a car to or spend money at; preferably, for commerce, both (Campbell). Besides parks, there is sparse information on any other features of our land: industrial hazards, infrastructure, and everything that powers our world—and will outlast us—is not marked, including Superfund sites, much less the hazards they have and continue to pose to public health and the environment. While we understand these maps as fairly complete indexes of our surroundings, they are indexes of our current commercial surroundings. This project is about building a map of the inverse, focusing exclusively on the land and the hazards.

Background

Prior to the first Trump administration, the EPA published a tool called TOXMAP to make the Superfund dataset somewhat navigable, which went offline before my research took me to this topic (Post-Tribune). The current standard tool from the EPA is an ArcGIS map, and states like California have their own specialized tools, but all suffer from high bandwidth requirements, difficult to read typography and visual design, and poor web accessibility. The Trump 2.0 administration is now working to rapidly take down content about climate, health, and the environment, including content they largely left online on federal websites in the 1.0 administration (Santarseiro). The current political climate demonstrates the urgency of the project, as data—data I have argued has historically not been sufficiently accessible, though available, to the American public—is rapidly disappearing from EPA and other environment-related agencies’ websites.

Artist Alexis Oltmer brings the topic of Superfund to a grounding in my hometown of State College, Pennsylvania (“in-site exhibition”). In a spring 2024 installation, “in-site,” she invents an icon system for visualizing the lists of contaminants at Pennsylvania Superfund sites, glass vials of contaminants on laser-cut mounts, and presents annotated EPA documentation in binders and on walls. The variety of materials and types of installations provide different perspectives and zoom levels on this vast dataset, including literally, when she highlights the boundaries of the Centre County Kepone site in purple ink on a topographic map.

One of these sites, Centre County Kepone, is situated next to my preschool. The site of a manufacturing facility that produced carcinogenic pesticides including kepone and mirex through the 1960s and 1970s, until they were federally banned for their harms to human and environmental health, the facility disposed of waste residues in unlined, open-air soil ponds and spray fields. Less than a mile away, my Montessori preschool taught me about nature, math, and writing, while we made field trips to the creek nearby. The school was on a hillside above what at the time was a Sheetz gas station, which repeatedly spilled hundreds of gallons of gasoline into the creek. That creek has endured decades of dumping of sewage, chemical runoff from the Superfund site, gas leaks, and more. My parents had no idea—nor did I, until researching this project—that my school had this proximity to a Superfund site. The toxic insecticides manufactured there are the same kinds of chemicals that contributed to my grandfather’s cancer, an avid gardener/landscaper in his later years before passing away long before I was born.

My memories of growing up in this beautiful forest in Pennsylvania have been complicated by the research and artwork I’ve done about the state as an adult. Every era of Pennsylvania history has been defined by extraction: from the tapping of the first oil well in Oil City to the entire state’s worth of forests clear-cut for charcoal and iron smelting to the twenty-first century’s widespread gas fracking. The bubble of socioeconomic privilege and security I grew up inside appeared to extend to the physical environment; growing up in this verdant forest, while never pristine or old-growth, seemed far from being across the street from a plastics factory. Moving to cities like New York and San Francisco as an adult, I have also felt isolated from industrial pollution and history, but these cities are unquestionably built atop those legacies, even if the high rents and new developments attempt to mask them. I seek to investigate the tension of this privilege, these memories, my current life, and the scientific realities on the ground.

I now work in the tech industry. Kate Crawford’s 2023 essay “Earth” in Datapolis: Exploring the Footprint of Data on Our Planet and Beyond traces the connection between the tech industry/San Francisco/Silicon Valley and Earth’s land. The section that stuck with me was reminding us how wealth first came to San Francisco through the Gold Rush—a dangerous, polluting, extractive process based on hype and individualism that made certain men wealthy. Then this spring, journalist Justine Calma wrote a longform piece for The Verge, “The women who made America's microchips and the children who paid for it,” which re-ignited my interest in Superfund. It follows the generation of women that worked in the chip fabrication labs of Silicon Valley in the 1980s, the hazardous materials they worked with, and the generation of children they had with frighteningly high levels of birth defects and developmental disabilities. The article brought to my attention one specific result of Superfund cleanups, land use restrictions, which are oftentimes not followed by future tenants of properties, e.g. building schools or churches over contaminated soil.

While Silicon Valley in my lifetime has had a reputation for hype cycles, these essays reveal how this pattern repeats back to its inception: cycles of primarily men chasing a dream of individualist extraction that produces outsized wealth, often with lofty ambitions for how this extraction will improve the world. The Gold Rush in the 19th century saw people flock to the unknown West to extract gold, leaving behind polluted waterways. The silicon boom in the 20th century manufactured microprocessors with a slew of toxic ingredients and few worker protections, minting billionaires and household names like Intel alongside a trail of polluted Superfund sites. Now in the 21st century, AI labs have switched to extraction of digital information, outsourcing the toll of manufacturing to Asia and the energy/water costs to more rural areas of America, where they are celebrated as job-creators. In each century, a nationally-hyped dream brought a new generation of young men to Silicon Valley, who built their dreams no matter the environmental costs and a few concentrated the wealth for themselves. Each movement did change the world profoundly, though not without consequences their backers mostly omitted from the narrative. Tech bubbles are built on hype that materializes into technical innovations and fortunes for a lucky few, but abandon someplace in shambles.

Tega Brain and Sam Lavigne further connect modern technology to physical Superfund sites with their 2020 short film Sacrifice Zone. For the project, they travelled to the Endicott, New York, where IBM manufactured semiconductors for decades, leaving behind what is now a Superfund site. The company built a country club there in the 1930s, where employees growing wealthy off the company’s engineering for the Nazi regime enjoyed leisure. In the film, the artists walk through the abandoned halls of the country club, opening mail about tennis events from decades ago and witnessing the frog ponds and graffiti in the empty spaces. The haunting legacies of pollution, technological surveillance, fascism, and tech industry excess collide in the film.

This understanding of technology as rooted in environmental extraction/destruction, and the grounding of the “cloud” in the land, is a key factor in my work. No digital computer has ever existed without excising a toll on the natural environment and showing deep influence of the politics of the people who made it—from Alan Turing in the British military to the modern TSMC iPhone processor’s manufacturing in Taiwan. Though the physical chips continue to shrink, the toll of their manufacturing continues to grow. And as our expectations of compute grow—the more data we’re accumulating, the more processing we do of it—the more servers are needed, and the more electricity and water they consume. LLMs are novel in exacting their scale of toll on the environment, in fossil-fuel-powered electricity usage, water usage for cooling, and the extraction/production that goes into the servers’ microprocessors, all removed from the interface or site of usage.

LLMs are novel at answering wide-ranging questions rapidly, if not without asterisks. The latest research indicates that leading AI “answer engines” such as Perplexity get only 2/3 queries correct, and popular models such as Grok get 94% of queries incorrect (Jaźwińska and Chandrasekar). While models’ default training data leads them to make frequent errors, these can be mitigated with techniques including retrieval-augmented generation (RAG), where an LLM is pointed at a corpus of text and grounds its answers in those texts instead of general world knowledge.

The state of the art for interacting with LLMs is the scene of much experimentation. Amelia Wattenberger, an interface researcher/prototyper, writes about LLMs as providing a service to text similar to zooming a map, where different landmarks become revealed or hidden as you zoom—LLMs can keep track of what’s important in text and change the length/level of detail at any level of granularity. This closely tracks the concept of ingesting many source documents and composing high-level summaries of site histories, with the ability to drill into more detail by asking questions: zooming out the map. One of the visualizations she explores as part of this concept is an alternative mechanism to serve the hyperlink, exploding out a handful of cards that reorganize when highlighted words are clicked.

Two other recent technical demos show new ways of interacting with LLMs. At South Park Commons in January 2025, Toby Brown showed off the idea of using text selection to initiate follow-up queries to an LLM, as opposed to the linear lists of suggestions employed by “traditional” LLM UIs like Perplexity. The selection allows you to choose any part of the response as a jumping-off point, and could lend itself to asking multiple new questions per answer, instead of having one linear conversation as demonstrated here. Eddie Jiao has prototyped an LLM/infinite canvas experience designed for documents/reading. Jiao’s demos, while early, bring LLMs into a context of reading and learning, using branching results and follow-up questions.

LLMs should never be used in artwork unthinkingly. There are valid artistic reasons to include them—but similar to toxic metals or other corrosive physical art materials, they need to be weighed with the societal cost of extraction/manufacturing/storage/use of those materials. LLMs are largely being used and developed to accelerate extractive capitalism, yet they also make fantastic tools for exploring curiosities with lower friction than traditional research. While even RAG does not guarantee zero hallucinations or mistakes, I believe they can create tools for learning about the world so much more accessible to people that the alternative is a void of knowledge, not a perfect research process. This project attempts to justify the destruction of using LLMs to flip the motives for their overuse on their head, plus using technical steps like caching responses and on-device processing to reduce the pollution that goes into making the project itself.

Laurie Voss, a web developer working on a popular RAG tool, wrote two principles to keep in mind about LLMs: “LLMs are good at transforming text into less text” and “LLMs only reliably know what you just told them, don't rely on training data” (Voss). This archive of documents provides plenty of information to feed into an LLM, and this project relies on the concept of turning a lot of text into less. To my knowledge, LLMs have not been applied to the Superfund dataset; this is a novel contribution to both the environmental and AI/web technology/design fields.

Methodology

The project focuses on the most dangerous, extreme sites, those listed in Superfund’s National Priorities List, and leverages advancements in LLMs to not only render statistics, but communicate the gist of the history and status of these sites to the public. I began by collating multiple EPA spreadsheets of every Superfund site, with their EPA IDs, status, name, address, geographic coordinates, physical size, years of introduction/processing, etc. This data went into JSON files and a Postgres database for further processing. I used an open source web scraping library called defuddle to download the information on the EPA websites about each site to a local text file. While my initial version of this project relies on data that continues to be available on EPA websites, it is possible this data disappears as TOXMAP and others have. In this case, the National Security Archive’s work in preservation, combined with that of Archive.org and other nonprofit initiatives, could become necessary.

I made a web frontend with Next.js and a Mapbox map, pulling from the Postgres database on Supabase to render a static website. I am using a litany of both open and closed source tools in this process, including PostgresQL on Supabase, OpenAI’s GPT-4.1 for responses, the Vercel AI SDK for interfacing with OpenAI, Next.js with React Server Components for rendering the site, Tailwind CSS, Mapbox maps, Vercel hosting, and more. The resulting website is entirely open source on GitHub.

Navigating the map, visitors can open a Superfund site to read a brief synopsis of the site’s history and present written by GPT-4.1, see key information, and ask questions about it. The LLM powering these answers is trained on the documentation published for each Superfund site, using RAG to provide answers based on per-site documentation and mitigate hallucinations. To reduce the carbon and water footprints of these interactions, the summaries are generated once and stored in the Postgres database instead of on-demand. Because the website allows freeform questions, it connects live to OpenAI as well, and does not contribute novel approaches to reducing the environmental footprint of those queries.

One novel interaction is the introduction of a new kind of AI-first hyperlink. While traditional hyperlinks take visitors out of the context of their current page down a new rabbit hole, often with far more information than they bargained for, such as on Wikipedia, I wanted the summaries to use technical concepts as to not oversimplify without confusing visitors. Words, acronyms, and scientific concepts average visitors might not be familiar with are double-underlined, and clicking on them pre-fills the chat box with “What is/are [concept]?” with, for example, PFAS or PCBs. Users can edit the question or hit enter, getting a quick two-sentence answer written for them, sometimes linking in yet more concepts they can ask about. These AI-first hyperlinks serve to reduce friction to asking clarifying questions.

The same methodology of summary-writing has been brought to contaminants. Top contaminants across the site—there’s over 500 unique contaminants across 17 categories—have GPT-written descriptions summarizing EPA PDFs and Wikipedia pages. Each site lists contaminants with links to these summarized descriptions, grouped into categories of the “contaminated media” (EPA phrasing), or places contamination exists. For example, a site will list “Groundwater (3 contaminants): trichloroethane, PCBs, lead.” Visitors can click on each contaminant to learn about it. For each of the 17 contaminated media categories EPA defines, I designed custom iconography, inspired by Oltmer but grounded in the 16px digital usage, with circular shapes and round corners. The icons are published in Figma Community for other designers to make use of. I grouped the types of contamination into Ground, Water, and Air categories, using the category colors for each icon. I renamed some of the more confusing EPA types, such as leachate to “rainwater runoff” and “NAPLs” to “non-soluble liquids,” then wrote one-sentence descriptions for each, explaining concepts like soil vapor intrusion and surface water contamination clearly. These icons, explainers, and categories are marked throughout the website to make every interaction with Contamination more grounding and illuminating.

In addition to contaminants, the American Toxics site is the first tool I’m aware of to show landmarks on top and surrounding each Superfund site. Using the Mapbox API, I pulled specific categories, including schools, churches, senior care facilities, healthcare facilities, prisons, and parks, of landmarks within two miles around each Superfund site. The UI categorizes and groups them, e.g. “3 schools nearby (daycare, elementary, & high schools).” Many sites have land use restrictions, as noted in their summaries, that have not been followed, according to Calma. This highlighting of the land uses reinforces the cultural need to understand the risks of Superfund sites in our local communities.

Basic organizational methods from EPA I have redesigned, including the cleanup statuses. While EPA refers to sites as being “added” (listed) or “deleted” from the NPL (aka “cleaned up”), I created a simple timeline UI, color-coded, where sites move from Proposed to Hazardous to Cleaning to Cleaned to Completed. Choices like the charged word “Hazardous” over a more neutral “Listed” remind us that these designations have a real impact, as do the times between them; many sites remain in Hazardous or Cleaning for multiple decades, poisoning communities in the meantime. The bureaucratic language EPA uses can serve to downplay this importance, but in a consumer tool it’s vital to enforce the stakes.

In addition to cleanup status and state, I created twelve categories of sites: waste, chemicals, water, metals, military, manufacturing, wood, mining, dry cleaning, tech, radioactive, and other. For each category, I picked an icon, color, and wrote a description of those types of sites. After initially using a simple, custom word-matching algorithm to categorize sites, I was unhappy with a large percentage of results, and wrote a TypeScript script that sends the site name, location, and GPT-written summary of the site’s pollution to GPT 4.1 Nano for classification. While I’ve manually updated a handful of categorizations I disagreed with, the GPT model provided a good-enough bulk sorting. The categories on the site serve to give a rough sense of the cause of pollution without viewers needing to read the longer explanations.

The website exists as a portal out of the commercial web, out of the AI productivity hype cycle, and into a space designed for discovery and learning. No two journeys through this map will be the same, driven by the visitor’s path through physical space, their intellectual curiosity, and the non-deterministic answers of LLMs. Instead of government databases with rigid tables, acronyms, difficult typography, and dated designs, this space combines brand-new AI technology and novel interface design with decades of historical archives, attempting to bridge environmental history and our present surroundings with the digital space we primarily now occupy.

Beyond the website, American Toxics is an 8-piece installation, considering the dataset from different geographic scales and mediums/perspectives. It includes: the website (LLM perspective), a mobile of contamination icons (toy), Centrepiece (map), Centre Sound (audio), Land Clusters (print), timeline (bureaucratic/paper/time), and a jar of sludge from Gowanus Canal (microbial).

The physical installations begin with a timeline, hanging from the ceiling as one roll of paper extending fifty feet down, curled back on itself to make the size approachable. Starting in 1983, it shows all 1,840 Superfund sites in chronological order, grouped by year, with the date they officially were added to the NPL, the name, the city, and state. (The final site added in 2024 is called “Afterthought Mine,” a poetic ending to the Superfund project under the Biden administration.) This timeline tries to give a sense of scale to the dataset; whereas the digital map pins can be zoomed out and minimized and overlap, seeing the fifty feet of paper evokes the government bureaucracy. I created the timeline with a TypeScript script that generates a Markdown file from the source database, then a webpage that renders the Markdown file with custom styling to match the website’s typography. I then printed the webpage as a PDF, eventually settling on 13x198” page sizes, creating a 3-page PDF. I bought a roll of 13”x50’ paper, matte cardstock from B&H, and printed the PDF on a large-format Canon printer at the ITP Design Lab. This took many test prints, dozens of iterations of sizing, margins, and centering, and two rolls of paper to print sequentially with minimal gaps, disabling the printer’s paper cutter, and figuring out how to work around Adobe Acrobat’s 200” maximum page size without putting gaps in the middle of year sections on the timeline. The timeline was the simplest idea I had for an installation—one super long sheet of paper with every site listed—but took dozens of hours to execute.

The show continues to a table with Centrepiece, a satellite map of the area between my preschool and Centre Kepone. I used OpenAI o3 to write a Python script to reverse-engineer the Apple Maps satellite imagery API, downloading the highest-resolution tiles available and stitching the photos together. I then printed a 42” square print, and trimmed it down to 32x40” on a paper cutter to fit on a foam-core board, and adhered it with spray adhesive. I installed small pins in the map to mark a few landmarks, including the preschool, gas station, creek, and Superfund site. I designed a chain-link fence pattern in Adobe Illustrator and laser-cut it out of watercolor paper for a textured thick paper texture that could stand upright. The four interlocking walls of the fence illustrate the site’s literal fencing and reference Oltmer’s purple ink over the same site in her installation. Also atop the map are two photo installations, each with three 6x4” color prints, mounted on laser-cut white cardboard sheets adhered together to create equilateral triangles. The triangular shape references the blocks and toys at the preschool, evoke a warning sign, and mirror the mobile’s construction above (described later). One, printed on glossy photo paper, includes three photos my father took in 2004 of me exploring the gardens and inside of my preschool, located geographically next to the preschool atop the map. The second, printed on matte bond paper, shows what approaching the Superfund site on foot looks like, photographed by Oltmer last year. To position viewers, on the corner of Centrepiece there is a laser-cut model of Pennsylvania on 3/8” translucent acrylic with a hole cut in the aspect ratio of the map over State College and a text label engraved. The Centrepiece map is accompanied by bookshelf speakers playing a 30-minute looping field recording of audio recorded by Oltmer at the Centre Kepone site, called Centre Sound. Visitors can hear cars driving by and an air filtration system operating, grounding them further in the location. foam-core board with a printed sheet of color-coded icons and explanations of each pinned location, their history and significance. A digital version runs on an iPad, which uses a basic AR interface of text labels over the iPad’s camera feed to help the physical object fade away and keep focus on the map.

Hanging above the map is a triangular mobile of my custom icons for the mediums of contamination, split by color across ground, water, and air, dangles. Each icon is laser-cut out of 3” circlular acrylic, bonded with E6000 glue to a 1/16” frosted translucent acrylic circle. The frame of the mobile is built of three balsa wood slats, glued and screwed together at three points, with looping chains through screw hooks joining at an S-clip centered above it. The ground and water sides of the mobile have six equally-spaced screw hooks upside down, and five on the air/other side, with a chain dangling from each. Drilled holes in each icon backing have small metal loops that pierce the chain, connecting the icons to the frame at varying lengths. Related icons, like liquid waste and solid waste, have equivalent-length chains, while assorted contaminants have more randomly-varied chain lengths. The triangle of the mobile with the categorization of contamination (the future) connects to the the two triangular photo installations create three points of the past (memories) and present (Superfund site).

Installed on the wall is Land Clusters, a large-format print. I downloaded satellite imagery of every site across the U.S. from Mapbox, then used a Python script running t-SNE to analyze visual similarity (non deterministically), plot the images, and assemble one massive PNG. It’s printed large-format on the wall behind the computer. It serves to remind viewers that Superfund sites are not glowing neon green: they are often unremarkable forested land, overgrown fields, abandoned buildings, parking lots. Clusters of forest, urban buildings on pavement, waterways, airport runways visualize what the different types of sites look like.

Alongside the computer is a jar of sludge from the Gowanus Canal, the nearest Superfund site to the exhibition in downtown Brooklyn. The sludge is a year old, collected at the Whole Foods at Smith & 9th St by members of the Elizabeth Henaff Lab at NYU IDM. I collaborated with Karolina Sulich to get this sample the lab was looking to discard, since in the final days of fabrication the low tide times did not correspond to our shared availability for fresh sludge-extracting.

The collection of installations work together to explore Superfund at different scales. The timeline shows the vast scale of the whole dataset, as does the website, while Centrepiece zooms into one site more intimately, grounding viewers in audio and place. The different materials and mediums used—plastic, paper, photo prints, map pins, wood, AR, audio, glass jar of sludge—keep viewers looking around, thinking about the materials at these sites, and looking from new perspectives.

Results

This project, as an artistic endeavor and technical/scientific one, did not seek formal evaluation by quantitative metrics. I wanted to tell the story of my own discoveries of the Superfund program: of the scale of this problem across the country (nearly twice the size of Maryland in total land area), of the personal impact of this site in my hometown, of the lack of tooling and design available. The website exists for myself and the world, open source, and available at https://americantoxics.com. The categories, new cleanup statuses, and contamination types serve to give scientists and officials clearer methods of communicating about these sites with the public. The interaction design of the new hyperlinks attempts to push forward AI design, while presented in a context of using LLMs to enliven the world, not pollute our planet and information ecosystem. The icons and information design exist to give the design community more avenues for working with datasets like these. The art installation exists to move viewers and tell this story in a gallery format, one I’ve never worked in before.

While I have dreams of many more installations and visualizations, the first collection went on display in downtown Brooklyn at the 2025 ITP Spring Showcase, where hundreds of people tried the website, looked at the timeline and sludge, and in many cases, conceptualized toxic waste in America for the first time. American Toxics is a culmination of years of my own environmental storytelling plus my explorations into web design engineering, data visualization, physical fabrication skills, and scientific research and learning. It exists to arm Americans with key knowledge about what’s around us, with which we can fight for funding for Superfund cleanup and environmental justice in our communities. It builds on the work of thousands of unknowable EPA employees, community organizers, scientists, journalists, artists, photographers, open source developers, and more.

Works Cited

Brain, Tega, et al., directors. Sacrifice Zone. 2020. Tega Brain, https://tegabrain.com/Sacrifice-Zone. Accessed 15 April 2025.
Brown, Toby. “Beem Demo.” Twitter, 25 01 2025, https://x.com/gopalkraman/status/1883248138745741750. Accessed 3 March 2025. Calma, Justine, and Amelia Holowaty Krales. “The women who made America's microchips and the children who paid for it.” The Verge, 19 February 2025, https://www.theverge.com/features/611297/manufacturing-workers-semiconductor-computer-chip-birth-defect. Accessed 3 March 2025.
Campbell, Lachlan. “Remapping Our Landscapes.” @lachlanjc/edu, 15 May 2023, https://edu.lachlanjc.com/2023-05-15_dis_remapping_our_landscapes. Accessed 3 March 2025.
Crawford, Kate. “Earth.” Datapolis: Exploring the Footprint of Data on Our Planet and Beyond, edited by Paul Cournet and Negar Sanaan Bensi, Nai010 Publishers, 2023. Accessed 4 March 2025.
Jaźwińska, Klaudia, and Aisvarya Chandrasekar. “AI Search Has A Citation Problem.” Columbia Journalism Review, 6 March 2025, https://www.cjr.org/tow_center/we-compared-eight-ai-search-engines-theyre-all-bad-at-citing-news.php. Accessed 11 March 2025.
Jiao, Eddie. “Common Knowledge Demo.” Twitter, 25 January 2025, https://x.com/gopalkraman/status/1883248154491166912. Accessed 03 March 2025.
Oltmer, Alexis. “in-site exhibition.” Alexis Oltmer, 15 April 2024, https://www.alexisoltmer.com/insite-exhibit.
Post-Tribune. “Federal agency shuts down website with interactive map showing pollution sources and Superfund sites.” Chicago Tribune, 6 January 2020, https://www.chicagotribune.com/2020/01/06/federal-agency-shuts-down-website-with-interactive-map-showing-pollution-sources-and-superfund-sites/. Accessed 3 March 2025.
Santarsiero, Rachel. “Disappearing Data: Trump Administration Removing Climate Information from Government Websites | National Security Archive.” National Security Archive, 6 February 2025, https://nsarchive.gwu.edu/briefing-book/climate-change-transparency-project-foia/2025-02-06/disappearing-data-trump. Accessed 3 March 2025.
Voss, Laurie. “What I've learned about writing AI apps so far.” Seldo.com, 20 January 2025, https://seldo.com/posts/what-ive-learned-about-writing-ai-apps-so-far. Accessed 11 March 2025.
Wattenberger, Amelia. “Fish Eyes.” Amelia Wattenberger, 6 December 2024, https://wattenberger.com/thoughts/fish-eye. Accessed 3 March 2025.