Author Archives: Michelle Santiago Cortes

Final White Paper: AI Hallucinations Project

Leave a reply

The AI “Hallucinations” Project is an interdisciplinary initiative that transforms interpretations of algorithmic errors in the field of digital humanities inquiry and research. Fabrications or misrepresentations, otherwise known as “hallucinations,” produced by Large Language Models (LLMs) have been (mostly) treated as technical glitches. This initiative positions such errors as cultural artifacts. The project is a systematic documentation of errors made by popular LLMs –– ChatGPT, Gemini and Claude — in response to prompts about Black American and Puerto Rican histories of the 19th and 20th centuries. by building on the Black Knowledge Erasure Dataset (BKED) dataset.The AI Hallucinations Project is a dataset of 150+ “hallucinations,” their prompts, the models that produced them, and the accessible archival resource that provided the accurate response. It is a record of AI hallucinations not as bugs, but as cultural artifacts that reveal how algorithms distort or erase the histories of marginalized communities. It argues that every question about marginalized histories is an opportunity to learn from the archives. It provides a platform that is fundamental for critically analyzing how algorithmic errors and bias occur within the complex (algorithmic) systems of and behind AI-generated content.  

Recent, rapid integrations of AI technologies across various industries and environments, especially education, have facilitated the need for (and encourages) intense examinations of model reliability. University of Maryland Researchers  note the evolution of digital capabilities in telecommunicating misinformation as well as numerous real-life examples of resulting consequences, which is permeated by AI technologies. The researchers, driven by their interests in the implications of AI-generated falsities, find that any form of artificially generated content (text-based, audiovisual material) is designed to appear legitimately credible regardless of truth or accuracy. Such features are helpful to user experience when LLMs are carefully prompted to uncover and repair informational mistakes. They recognize that the abilities of generative AI to produce and mitigate misinformation mostly, as a tool, exacerbates or influences user behaviors (Park et al. 2025). Researchers at University Canada West have a generally supportive framework, arguing that Artificial Intelligence improves research efficiency within academia, thus highlighting its capabilities to sustain research competency in additional industries. They recognize, however, legal standards that are controversial due to ownership and creation, as well as the perpetuation of biases and inequities connected to historical oppressions and socioeconomic disadvantages (Madanchian et al, 2025). Lastly, scholars and research figures from the Stanford Institute for Human-Centered Artificial Intelligence examined the various effects of AI technologies on African American communities. It has proven useful in general education, medical care, as well as employment, while prone to algorithmic bias since bias is embedded within the data sources of many mainstream AI models (Djanegara et al. 2024). The diverse range of perspectives mentioned here represent extremely few academic voices, though connected to broader observations and critical  studies of Artificial Intelligence’s implementation across academia and various industries where research is imperative. Historical inquiry is not exempt, which is where the AI “Hallucinations” Project enters as a tool for observing informational errors and most crucially, specific examples of how errors were generated. 

Dartmouth professor Roopika Risam published New Digital Worlds: Postcolonial Digital Humanities in Theory, Praxis, and Pedagogy in 2019. Even though its publication is before recent ubiquitous expansions of public Artificially Intelligent models and services, her rhetoric describes the approaches undertaken in the development of the AI “Hallucinations” Project. She defines postcolonial digital humanities “an approach to uncovering and intervening in the disruptions within the digital cultural record produced by colonialism and neocolonialism.” Furthermore, remarks that DH praxis creates unique methodologies for literary, historical, and socio-cultural analysis, demonstrating a continuation of historical practices in humanities knowledge production promoted by interdisciplinary approaches, involving perspectives and practices from an array of academic backgrounds. These interventions in particular, open different routes for investigating questions of the authority held by mostly Western and US-based technology corporations, positioning respective practitioners to study the influences of associated ideologies on various digital cultural environments, notably the internet (Risam, 2019, pp. 3-4, 9-10). Risam’s explanations for postcolonial methods in Digital Humanities embodies a framework that is consistent with digital colonialism, which Toussaint Nothias defines as a kind of critique aimed towards technological corporations – mostly of Western origin. Moreover, digital colonialism rhetorics assert that prominent tech corporations operate in manners that prioritize chances of attaining revenue, even if it harms the interests and security of a user who could be vulnerable, marginalized, or otherwise arriving from an underprivileged background (Nothias, 2025). The AI “Hallucinations” Project, with our backgrounds, discussions, approaches, and goals considered, represents a postcolonial intervention as it outlined by Risam, supported by Toussaint Nothias.  

Introducing Roopika Risam is appropriate for understanding the decisions and processes undertaken to build the AI “Hallucinations” Project. Postcolonial digital humanities, according to the professor, involves a diversity of scholarly and practical backgrounds who are interested in critical examinations of the Generative-AI technologies put out to the public. Scholars involved in the further development of the project, after initial completion of the Black Knowledge Erasure Dataset, have experiences that range from documentation, public outreach, historical research, interface design, website development, and coding. Furthermore, we demonstrate commitment to the tradition of humanities knowledge production by merging these skillsets in a manner that enabled everyone to contribute efficiently to development, especially by remembering findings from research referenced above. It is also worth recognizing projects that partially inspired this initiative. More specifically, the AI Incident Database providing real-world cases of harms and mishaps involved AI technologies; additionally, TruthfulQA functioning as a benchmark for studying the degree of truth generated by artificially intelligent language models. The AI “Hallucinations”Project,” expands layers of perspectives about AI and presents a specific method for learning how information on real (past) experiences are generated as a result of generative models recollecting; in other words, analysis of algorithmic output. Furthermore, The Project encourages interventions in potential instances of epistemological manipulations, promotes pedagogies for studying AI distortions, and gives evidence of misattributions, factual errors, invented figures and other subjects, and other types of error generated by a Large Language Model, when inquired about experiences of historically marginalized communities. Considering that our project is concentrated within the humanities, the OER produced is not intended solely for academic analysis of AI legitimacy in retelling the past. Rather, potential studies referencing the published database can apply to environments where AI technologies serve members of the general public.  

Audiences

According to a recent survey by the Pew Research Center, more than half of US teens use AI chatbots for “information-seeking schoolwork,” while comparable studies on college students yield higher percentages. This is the age group that makes up the majority userbase of products like ChatGPT, Google’s Gemini, and Claude. The gains earned by Black and Puerto Rican activists seeking to have their communities histories properly acknowledged by the education system and taught in American classrooms stands to be undermined when these students turn to LLMs that rely on low-quality data, biased algorithms and misaligned value systems to do their history or literature homework. The AI Hallucinations Project, first and foremost, documents the extent to which these profit-seeking services are depriving today’s students from learning and engaging with these important histories. The students who are turning to these models to speedrun assignments are wanting in digital resources that address their needs and speak to their realities. Our project offers a self-guided intervention: Students can focus on close-reading the hallucinations and learning about accessible alternatives, or they can scroll through the Are.na board where memes and journal articles co-exist as complementary rhetorical tools. What our project does and LLMs fail to supply, is a sense of real agency that encourages personal judgment and discernment. 

Educators, our second audience, can turn to the AI Hallucinations Project as a tool for their own development, or an ongoing reference they can resurface in the classroom. While the AI Hallucinations Project is best navigated individually in an open-ended session where agency and curiosity lead the way, there is room for teachers and other educators to play a supporting role through class discussions or worksheets. We remain actively interested in connecting with educators who are curious about using our website, our Are.na board or our datasets.

But the Pew Survey’s findings regarding high school students’ use of AI chatbots for “information-seeking schoolwork” was conducted before Google announced its most significant search update in 25 years: A search box redesign that will integrate AI chatbots and turn all search queries into model queries. Which makes the topic of this project all the more relevant to the general population, or, at least, anyone who even casually uses Google’s search engine. Across all audiences, a successful interaction with the project would lead a visitor to internalize the questions the project raises.

Project Activities
Research and Fact-Checking

This project’s first order of business was to begin work on developing a new dataset dedicated to hallucinations about Puerto Rican and Diasporican histories of the 19th and 20th centuries. The Research Lead found that accessible online resources such as EnciclopediaPR and Puerto Rico in the American Century: A History Since 1898 (2009) were found to begin ensuring research organization, by setting up team members to seek out prompt ideas from known locations if needed. Next, Christian shared the links of up to 10 digital archives containing evidence of Puerto Rican experiences originating in the 19th and 20th centuries. Although it was ultimately decided that three main archives (the Center for Puerto Rican Studies, Library of Congress, Rutgers University Puerto Rican Archival Collaboration) would get explored during project development common set of standards were still needed for prompt engineering and identifying types of hallucinations generated. The first result is a set of guidelines that would standardize prompt writing by focusing on questions that begin with 5 Ws and 1 letter H (Who, What, When, Where, Why, and How),and discouraging prompts that would be considered unnecessarily complicated and would probably originate from an essay question on an Advanced Placement history exam. Once the prompts were executed, Christian examined data from the BKED to develop hallucination type categories and standardized definitions. The first round of fact-checking and categorization was split among the three team members before a final round of fact-checking and categorization was carried out by the Research Lead. 

Technical Development

The technical development of the AI Hallucinations Project focused on creating a robust pipeline for data collection, validation, and public dissemination. To achieve a functional Minimum Viable Product (MVP), the team’s developer implemented a tech stack centered on Python, GitHub, and standard web technologies. The core data architecture relied on Python scripts designed to automate structured query prompts sent to the APIs of major Large Language Models (LLMs), specifically GPT-5, Google Gemini, and Claude. These scripts (e.g., collect_records.py and append_records.py) captured raw model outputs as JSON files, which were then aggregated and transformed into structured CSV datasets for human annotation.

To maintain data integrity across the project’s two parallel datasets, the Black Knowledge Erasure Dataset (BKED) and the Puerto Rican history dataset, the team utilized JSON and CSV schemas for strict validation of metadata fields such as “error_type” and “verification_source”. GitHub served as the central repository for version control, issue tracking, and collaborative development, ensuring that the code architecture remained transparent and reproducible.

The project’s front-end was developed as a public-facing web archive using HTML, CSS, and JavaScript. The technical architecture was designed as a static site to host the explorable database, featuring interactive elements that allow users to filter over 140 records of verified hallucinations by demographic category and error type. Data visualizations, including comparative charts showing hallucination rates across different model families, were integrated using Chart.js to transform static data into accessible, empirical evidence. This streamlined infrastructure ensured high performance and reliability while meeting universal design standards without the overhead of complex back-end server maintenance.

While our initial proposal included stretch goals, such as implementing dynamic front-end frameworks like React or Vue, or integrating a dedicated back-end database, we scaled these back to ensure we met our project deadline. We determined that a static, minimally interactive site hosting structured CSV data was more feasible and perfectly aligned with our primary goal of making the data accessible. During the final deployment phase, we also adjusted our infrastructure by executing a domain transfer to Reclaim Hosting; our website is hosted on Reclaim Hosting courtesy of the GC Library and GC Digital Initiatives.

The technical milestones outlined in our initial Work Plan accurately reflected the project’s overall trajectory. However, the post-spring break phase proved to be the most logistically demanding, as technical development was strictly gated by the dataset verification workflow. We could not build the comparative hallucination rate charts or integrate the explorable database until the curation tasks were fully completed. Navigating this bottleneck underscored the critical importance of our collaborative team structure; continuous team communication and shared accountability ensured that our technical implementation seamlessly supported the rigorous research objectives, ultimately allowing us to test and deploy the live explorable database on schedule

Web Design and Community Building

The website’s overall design and layout were executed as planned. We chose to list the hallucinations on the main page, with a drop-down feature for each that included the prompt, model, and fact-checking details for each. Separate pages — all listed at the top of the website — each housed downloadable datasets and links to their corresponding GitHub repositories, methodology details, an About section, and links out to an Are.na board where books recommendations, journal articles, memes, articles, archives and other resources for additional resources were listed. 

Minor chances in design and web layout arose when we realized text-based pop-ups added visual clutter rather than easier points of entry. So the decision was made to populate the pop-ups with data visualizations based on the insights gleaned from the dataset, and to add introductory text to the main page to infuse it with narrative and ease in the visitor. This had a rippling effect in the rest of our public-facing materials: While the original plan included a series of 500-800 word “mini essays” that engaged in easy-to-read criticism about subjects ranging from the anthropomorphising effect of the term “hallucination,” generative AI’s parasitic relationship to knowledge production ecosystems, and the methodologies used to develop the datasets.These mini-essays would also double as content for future newsletters, but since the immediate needs shifted, we pivoted instead towards writing one long blog post-style text to send in a welcome email. Three total versions of our core statement were drafted: a “micro” version (40 words) to use when a two-sentence description is required, a 100-word version for the top of the homepage that can double as mid-length project descriptions to share for future appearances, introductory emails, or similar cases, and a long 500-word version (the blog post text in the welcome email). As a result, the mini-essays were not produced and the future email sends will be written on an as-needed basis.

Our original outreach strategy shifted away from distributing the website’s links across link repositories across the internet, and leaned, instead, on relying on word-of-mouth. We knew that for a project that primarily lived on a standalone website would rely on a relatively dormant, but hopefully long afterlife of visitors and engagement. We do not think this is an undesirable outcome, in fact, this is what successful outreach and dissemination look like. Our idea of success, as far as outreach is concerned, has not changed: Successful dissemination is when the link to the website earns its place among people’s personal collections of internet artefacts or among institutions’s list of recommended resources. The Are.na board is well-situated to reach its intended audience of designers, scholars, amateurs, hobbyists, autodidacts, and arts and culture workers.

However, we were hoping to identify 10-15 link repositories — user-generated spreadsheets, links compilations, etc. — where the website link might find interested audiences, and ended up with just under five. Additionally, technical difficulties during the domain transfer phase delayed our web development timeline and we were not able to have a shareable link until presentation day. As a result, we focused on developing our mailing list and newsletter workflow (more on that below), to ensure the project is shareable on a continuous basis.

Accomplishments

The culmination of our development phase is the successful deployment of the AI “Hallucinations” Project website, which directly achieves our proposed objectives by functioning simultaneously as a comprehensive digital archive and a critical analysis tool. This public-facing, explorable website systematically catalogues how prominent large language models fabricate, distort, or entirely erase the histories of marginalized communities. By fulfilling our quantitative and qualitative goals, we have produced a robust Open Educational Resource (OER) that equips students, educators, and scholars with the concrete empirical evidence required to actively challenge perceived algorithmic authority and neutrality.

The central affordance of the prototype is its highly interactive, explorable database. This interface empowers users to granularly filter AI “lies” by specific demographic categories, focusing explicitly on the 19th and 20th-century histories of Black Americans and Puerto Ricans. The search architecture allows visitors to isolate specific modes of algorithmic failure through a rigorous, controlled vocabulary of error types, which includes “Adjacent Error,” “Erasure By Omission,” “Factual Error,” “Geographical Error,” “Invented Figure,” “Misattribution,” and “Temporal Error”. Each entry within this database exposes the mechanics of epistemic erasure by displaying the original prompt, the unedited output generated by major model families (GPT-5, Gemini, and Claude), and a detailed, human-verified correction. Crucially, every curator-led correction is grounded in and verified against gold-standard historical repositories, such as the Schomburg Center for Research in Black Culture and the Center for Puerto Rican Studies.

Beyond the searchable database, the website features an array of dynamic data visualizations designed to synthesize and communicate our key findings. These include a primary visualization chart and comparative hallucination rate charts that illuminate how epistemic erasure and cultural distortions function differently across specific diasporas and distinct language models.

 

The site is structured with dedicated “About,” “Methods,” and “The Data” sections. These pages host short, critical essays that interrogate the anthropomorphizing effect of the term “hallucination,” examine generative AI’s parasitic relationship to established knowledge production ecosystems, and transparently document the methodologies utilized to develop the datasets. By reframing these technological glitches as significant cultural artifacts , the platform affords critical AI researchers, investigative journalists, and archivists the structured data necessary to conduct targeted audits of model behavior and report on the quantifiable harms of algorithmic bias. Furthermore, to ensure ongoing interoperability and reuse, the website provides direct access to our GitHub repository, allowing users to freely query and download the raw datasets for further interdisciplinary research.

In addition to the primary website, a supplementary digital product developed is the project’s dedicated Are.na board. The board has evolved into a central component of our organic outreach and dissemination strategy. The board serves as a publicly accessible repository of our intellectual framework. We curated foundational literature and annotations on algorithmic bias and AI ethics, including extensive excerpts from texts like Algorithms of Oppression and “On the Dangers of Stochastic Parrots”. This platform was strategically chosen over traditional, algorithmically-driven social media to deliberately embed the project within digital communities of researchers, designers, and technologists. By positioning our research materials here, we ensured the project’s resources could be seamlessly integrated into user-curated link repositories, digital toolkits, and academic bookmarks, effectively sustaining the project’s afterlife and extending its educational impact to highly engaged, “very online” users.

Evaluation
Responses and Feedback 

During the project’s dressed rehearsals, a consultation meeting with the Graduate Center Digital Initiatives (GCDI) fellows, and a one-on-one with Luke Waltzer (Director, Teaching and Learning Center), we received invaluable feedback that prompted a reassessment of both our technical outputs and conceptual framing.

Feedback on our prototype website and presentation highlighted the need for immediate clarity. Evaluators noted that the website’s visual design featured a scrolling background that was too “busy” and distracted from the data; we planned to remove this to ensure the misrepresentations were clear at first glance. Evaluators also suggested that our methodology section read too much like a technical manual and advised us to craft it into a “narrative” that tells a story.

In preparation for our public showcase presentation, feedback urged us to spend less time on the granular methodology and instead boldly state the project’s overall importance and goals on the initial slides. To improve audience engagement, we were advised to replace heavy text with a moving GIF demo of the website, describe the “less obvious” hallucination categories in greater detail, and conclude with clear takeaways, future directions, and an open invitation for collaboration.

Conceptually, our GCDI meetings challenged us to refine how we frame and categorize “hallucinations.” Evaluators asked a profound guiding question: How does AI that follows compliance continue to perpetuate harm?. This led to a reassessment and realignment of our error categories to better focus on the specific types of harms we wanted to analyze. We explored the “degrees of rightness/wrongness,” questioning if we could measure how confident the language of the AI’s response was, often referred to as “temperature error”. Furthermore, feedback on our prompt-writing process reminded us that historical truth is often subjective; we had to be highly conscious of the “assumptions within the questions,” noting that affect-loaded narrative questions can inherently spark debate.

Project Strengths & Weaknesses

The primary strength of the project lies in its rigorous, human-in-the-loop fact-checking methodology and its conceptual reframing. By treating AI hallucinations not merely as technical glitches but as “cultural artifacts,” the project successfully established a framework for evaluating algorithmic errors. We met our Minimum Viable Product (MVP) deadline by strictly adhering to our Project Work Plan and consciously scaling back our technical ambitions. Relying on a streamlined static tech stack rather than complex dynamic frameworks allowed us to focus our resources on data integrity and empirical validation against gold-standard archives.

Despite its successes, the project faced several logistical and technical limitations. One significant weakness was the bottleneck created during the post-spring break period, which became logistically demanding as dataset cleaning, annotation, and web development workstreams converged simultaneously.To address this, we made the decision to scale back on the Puerto Rican history dataset.

Technically, our decision to scale back the website meant we could not implement advanced database-driven search and dynamic filtering mechanisms that a dedicated back-end server would provide. However, as a functional compromise, we successfully implemented a localized search bar and filtering functionality on the front-end, allowing users to seamlessly sort the 147 records by error type and model directly on the site. 

What Could Have Been Done Differently? 

If the project were to be repeated or expanded, rather than simply incorporating new tests, we could have conducted a comparative analysis of our methodology versus “Red Teaming” strategies. Red Teaming is a hands-on exercise where participants intentionally test Generative AI models for flaws and vulnerabilities that may uncover harmful behavior. Historically, this practice has traditionally been carried out by major tech companies and specialized AI labs behind closed doors. By comparing this industry-standard practice against our own human-in-the-loop, archival verification approach, we could critically ask: “Are there things we are doing that they didn’t?”. 

We would also explore integrating an “Agentic AI Process” (such as using Claude desktop) to automate parts of the verification workflows. Finally, as a key area for future research, we would prioritize exposing the sparse training datasets behind the models. This would provide deeper context into why culturally specific “adjacent errors” and misattributions occur, ultimately helping us fulfill our main goal of providing a more robust and actionable critique of AI systems.

The Future of the Project: Continuation + Sustainability

One of the goals of the AI Hallucinations Project was to build a website that could live on as an argument, a work of criticism that doubles as a teaching tool. The language and narratives on the website position the work as an artefact of its time, anticipating the speed of change in AI product development (and the criticism that follows). Our work’s value comes from its rhetorical potential – it’s a collection of case studies that can be discussed in the classroom, or iterated on by whomever finds the dataset on Github or Are.na. Expanding and refining the core dataset in the future will certainly enhance this value, but the project does not stand to lose anything if these potentialities are not pursued. A rhetorical tool is only as good as the communities that use it and the discursive possibilities it engenders. The near future of the AI Hallucinations Project will be dedicated to reaching new audiences and capturing their attention in a sustainable way.

We optimized for longevity and ease of continuity by developing a modular design system and applying the same principles to other components, like copy. The first step taken in the name of maintaining and sustaining the project was to compile a “toolkit” of design assets, shareable links, and copy that can be applied to a range of needs and contexts, or passed onto other designers who will be saved the effort of having to guess or re-create design decisions made years prior. Fonts, color codes, logos (in a range of colors and sizes) QR codes (in various colors), were compiled in shareable formats on a shared Google Drive as soon as those decisions were made. Various versions of promotional language and website copy — of different lengths, for different audiences — were also compiled. External and shareable links to the website, GitHub and Are.na page were also compiled in the same “Post-Mortem” document, along with important internal details such as account credentials, API keys and more. If ever the need arose to create a new web page, design a new deck for a presentation, provide a brief description to be bundled along with conference materials, a new social media profile, or continue the work done in this phase in a new iteration of the project  — anyone with access to the “Post-Mortem” document will be able to do so in a way that is consistent with the project’s existing branding, visual language and ethos. 

The second step taken for the sake of easy maintenance involved spending down our budget. Our budget was exclusively dedicated to paying for services and subscription terms that would make the maintenance of our website and consistent engagement with our community as seamless as possible through — at least — May 2027. The terms of these services provided an intuitive length for the first phase of project continuation and served as its primary structure. The subscription lengths of our domain name (3 years), Linktree, mailing list, and hosting services (1 year, each) set the terms for how (and for how long) we would have to maintain the website, and we found that small upgrades proved majorly instrumental. We are using Linktree (as a simple way of packaging and sharing the project’s main offerings – the website, the dataset, the Are.na board. The Pro plan ($144/year) includes a Mailchimp integration that allows visitors to sign up for our mailing list directly in the Linktree. On the back-end, this plan allowed us to customize our QR code using our chosen colors and incorporating our logo, which prompted us to pre-design several iterations and added to our shared toolkit.

Through Mailchimp’s Essential plan ($130.65/year), we were able to set up an automated welcome email for anyone who joins the mailing list. Visitors to our website are prompted to subscribe to the mailing list via pop-ups, and anyone who clicks into the Linktree will be asked to do the same. Upon sign up, subscribers will receive an automated welcome email that summarizes the core tenants of the AI Hallucinations Project, lists each of the offerings (database on homepage, GitHub repository, Are.na reading list, visualized findings) and includes links. This will ensure the website is “saved” in people’s inboxes, and is easy to return and reference.

The primary challenge of sustainability will be a financial one: Once the year-long subscriptions plans to Linktree and Mailchimp lapse, we will have no way of capturing emails or sustaining the attention of those who find a link to the AI Hallucinations Project website via our Are.na board, an old QR code, or rediscover it in the welcome email they got in their inboxes. By May 2027, we will also have to decide if we want to pay for another year of web hosting via Reclaim or find an alternative. There are no immediate plans for where the project will be by then, and it’s possible that web hosting changes might not pose an existential threat to the project if it has evolved into a different kind or website or tool. And even if this is the end of the AI Hallucinations Project, the website has been archived on the Wayback Machine, and the link to that is on the Are.na board, where more people can find it.

Bibliography

Caroline Meinhardt, Daniel Zhang, Ezinne Nwankwo, Haifa Badi Uz Zaman, Gelyn Watkins, Michele Elam, Nina Dewi Toft Djanegara, Russell Wald, Rohini Kosoglu, Sanmi Koyejo, “Exploring the Impact of AI on Black Americans: Consideration for the Congressional Black Caucus’s Policy Initiatives,” The Stanford Institute for Human-Centered Artificial Intelligence (2024) https://hai.stanford.edu/assets/files/2024-02/Exploring-Impact-AI-Black-Americans.pdf

Mitra Madanchian, Hamed Taherdoost, “The impact of artificial intelligence on research efficiency,” Results in Engineering, Volume 26 (2025)
https://www.sciencedirect.com/science/article/pii/S2590123025008205#sec0022

S. Park and X. Nan., “Generative AI and misinformation: a scoping review of the role of generative AI in the generation, detection, mitigation, and impact of misinformation,” AI & Soc (2025). https://link.springer.com/article/10.1007/s00146-025-02620-3#citeas.

Immediate vs. Distant Future

Leave a reply

Is the website live? No. (Although we’re about to be on the other side of our domain troubles.) Is the dataset complete? Almost! Still, this is the week of finishing touches. It has to be. For me, as outreach and UX design lead, it meant writing website copy and preparing as much as I can for when we do launch the website early next week. 

Our plan was always to re-engage our community via email sends. Some research helped me decide that I wanted that to happen via Mailchimp, which would allow us to add a sign-up field to our website footer. The other half of our outreach strategy consisted of spreading our links — to the website, the GitHub, the Are.na board and (why not, I thought) the email sign-up — like spores in the wind with the goal of planting them in as many high-touch places as possible. If all a link can do is sit there and wait to be clicked, I want that to happen somewhere where it might be found by someone who would appreciate where it leads to. But these seeds needed a package, something to keep them bundled together as we passed around our feeds. I impulsively started throwing together a Linktree page during the last hour of our second-to-last class and I was pleased to learn that it also had a handy QR code generator that offered all the customization features I was looking for all semester. 

All of this was being built with Thursday’s showcase, and nothing else, in mind. I’d been obsessing over how to make the path from presentation to email sign up as frictionless and inviting as possible:The Linktree is the business card; the website is the storefront; everything else is a billboard. The end of our presentation will display our QR code, in our signature colors with our logo integrated. When audience members scan, it will lead them to our Linktree, a visually seamless extension of our website, that will prompt them to subscribe to our newsletter and include links to the website, the dataset repository, and the Are.na board. A welcome email will automatically plant more links into their inbox, for future reference or immediate use.

Turns out, a couple hundred dollars is the difference between future reference and immediate use. Free trials of premium features across both platforms – Mailchimp and Linktree – made sure we had everything we needed to keep it all together through Thursday. To consider the possibility of spending down the budget, meant picturing ourselves a year from now, when the annual subscriptions would lapse and we’d have to decide if the money was well-spent. Would we be telling people to subscribe to our newsletter in May 2027? Would the renewal notice remind us of something we abandoned a year ago?

As I tested the Linktree and added my teammates as admins to the Mailchimp, I documented the passwords, API keys, hex codes used, emoji Unicodes. I downloaded the png’s of the customized QR code in our two main colors into our shared Google Drive and wondered how any of us would be able to navigate it should we find ourselves in need of some link or bit of information at any time in the near or distant future. What kind of posterity was I planning for, and why did this question suddenly eclipse the urgency of being ready for Thursday’s showcase?

As I waited for the group chat to respond to these questions, I found myself adding a “Post-Mortem One-Sheet” to our shared document. One place that would make it easy to ensure that are links remain unbroken, that anything with the words “AI Hallucinations Project” has the right colors and the right emojis, that the QR code always works and that if anything needs to be refreshed or retrieved, it wouldn’t take any digging. All this busywork that required me to refer back to some decisions made over the last few weeks — retrieve links, approved color HEX codes, recall the typefaces we used for the website, etc., — made me realize the bulk of the project was already behind us. Despite the final-mile-stress, most of the work had indeed already been done and, at the very least, we have one long Google doc full of links, blog posts, and meeting notes to prove it. And I can’t think of a better time to start making plans for how to preserve it all.

 

Personal Blog: Stealing Company Time

On my suggestion, no, insistence, my team had a skill-share meeting where Sasha kindly walked the team through the technicals – the tools we needed to install on our computers and how we would use them to run our queries and store and share the results earlier this semester. My primary interest in this project was based on how it gave me the chance to work closely with a small dataset, nurturing it from its inception. I was also looking forward to a slow, critical encounter with LLMs. I was really eager to spend time with these tools and their outputs so that I can deepen my criticisms. 

But I did not make it to that skill-share on the day it was scheduled. We have a standing meeting on Fridays in order to keep up with our production schedule, that is where I do and share my share of the work on outreach and web design, and where we troubleshoot and make decisions as a group. Most of the time outside of class that I have spent on the project, has been dedicated to doing my part to support Sasha with web design and Chris with research and making sure the workflows I own – design, outreach, etc. — are well-tended to. I have to admit I’ve only been mildly successful in carving out extra time to dedicate to learning about the tools and methods we’re using to actually prompt the LLMs and document their responses. The few times I’d successfully done so was when I stole company time from my day job to spend a bit more time with the AI Hallucinations Project.

But I feel like I made up for it when we met at the library. While Chris was finalizing his prompts, Sasha kindly and generously walked me through the tools we would use: VSCode, Python, Homebrew, and the various APIs for each of the LLM’s we’d test. I realized she could’ve done all of the querying and documenting herself in a fraction of the time it took her to walk me through it all. But it was genuinely enjoyable to troubleshoot together. It also fulfilled the purpose of the project, and I suppose this class too. Being able to contribute to this part of the process ensured this experience didn’t feel like three contractors throwing their work into a Google Doc and onto a website. Now I can say I tested these LLMs myself, watched them slowly respond to my prompts one by one. And they tested me, I already know I will have to revise some prompts and run them again. The project would exist, and likely succeed, without my intervening in this part of the process. But I would’ve learned way less from it all if I hadn’t.

Update: AI Hallucinations Project

Excitement and anxiety are definitely rising as we close in on the final stages of this project. I might just be speaking for myself, but I have a feeling my teammates relate: As the time arrived where we would have to write out new prompts and actually produce the dataset, arguably the marquee feature of the AI Hallucinations project, we suddenly found ourselves oh-so-preoccupied with other things: Setting up appointments with the digital fellow and other advisors, adding to and refining the website, double-checking our archives and prompt-writing guidelines.

Reality had its own plans: The original plan was to each write about 15 prompts based on archival research on the Puerto Rican and Diasporican histories, query the models, and collect the data and then repeat the process for an aimed total of 100 prompt responses. But in the class before last, Sasha suggested we slash the size of the new dataset in half, so that we only write 15 prompts each for an aimed total of 50 responses. We were so focused on refining our fact-checking process and prompt-writing criteria that we soon realized we did not have enough time to break up the prompting process into two stages like we had originally planned.

Turns out, writing the prompts took up more time than any of us had originally expected. On Friday afternoon, we met at the main branch of the Brooklyn Public Library, with varying degrees of springtime allergies. None of us had finished writing our allotted prompts, as planned, so the first hour of our session was dedicated to finishing up that work together. Sasha and I finished writing our prompts quick enough that Sasha was able to walk me through prompting the models and organizing the output. We did not have time to fact-check any of the outputs together. So we spent the last 20 minutes of our time together reorganizing our schedule and priorities for the coming week.

Since we are further along on the website development side, we reprioritized as follows: We will each fact-check our own model outputs individually during the week, and come together to trouble-shoot in class. The other half of our time in class will be dedicated to making some final decisions regarding our web-hosting and domain because GoDaddy hates us. This is fact-checking week in more ways than one, as on Friday, we also confirmed two meetings with advisors: One with the Digital Fellow and another with Luke Walzer, both of whom have expressed an interest in going over our fact-checking procedures and methodologies. These meetings and our response fact-checking efforts during the week, should put us in a good position to start concluding the fact-checking phase of the project. After this coming week, we will focus on finalizing the website and preparing our materials for sharing and launch. 

I made a logo

I made a logo. I was not expecting to make one, and I didn’t think we needed one. But I was putting together the first batch of public-facing materials (a Google Form to gather emails, a QR code for Sasha to include in her presentation) and I caught myself wanting for a little glyph or icon that could add levity to all the text. This is usually when I look through the emoji library and do some selects of color-coordinated emojis and add them to the “brand book” along with colors, typefaces, etc. I was putting together for Sasha’s presentation, and would eventually become the DNA of our website and other visual assets.  But I couldn’t find any that worked. In other contexts, I would find or make new emojis and add them to the library I’m working with, and this made me wonder if I couldn’t customize one to match our color scheme and, boom logo! We’re going with an adapted, or perhaps “darker” version of the sparkle emoji that so many commercial AI projects use to align themselves with magic and instant satisfaction. So I ended up with a logo, I don’t know how I’ll use it but I’m glad we have it.

I can’t say I am fully satisfied with the work I’ve contributed to my project so far. Bills have to get paid and work got in the way of everything this month and my coursework has definitely suffered. I missed Sasha’s walk through of AI prompting and API usage, which I had asked for because my priority in joining this project was getting close to the datasets and developing my technical proficiencies. I won’t beat myself up over it, but during the upcoming break, I plan on making up for lost time: Sasha was kind enough to record her workshop and I will sit with it quietly and start trying my hand at prompting these models. This will be my first contact with LLMs and I plan on savoring it and really learning from it. In short, I hope to immerse myself in the world of the AI Hallucinations Project so that I can actually prioritize my intellectual curiosity and let go of my fixation on producing and delivering rushed contributions. Time, unfortunately, is the only thing keeping me from getting the most out of this project.

 

During the upcoming Spring Break, honestly, I am going to catch up on all the work I haven’t done for this project.

 

Outreach for AI Hallucinations Project

For our outreach and social media plan we are going with a strategy that focuses on building a solid and very engaged community around the project and deprioritizes mainstream platforms with fickle algorithms and unstable visibility criteria that require more trouble than their worth. It’s divided into three phases that correspond to the project’s own development.

Phase one is for “Behind-the-scenes” work and community building. It begins with in-network outreach where teammates are tasked with talking about the project with at least ten peers, friends, and mentors. The goal here is to turn this project into a real thing we are attaching our names and faces to, and to begin integrating community-based stakeholders into our thinking of who this project will serve. As part of this phase, we’ll also set up appointments with the Digital Fellow and our own mentors in order to formally get advising on the project and, again, solidify stakeholder relationships. In order to track community growth, we will ask folks for their email and collect them in order to build a mailing list through which we’ll launch the project. By the end of this phase we should have collected a total of 30 emails.

Because this project will take on its final (for now) form as a website, we’ve given considerable thought to how links are shared and preserved in a digital landscape ruled by the ephemerality of algorithmic platforms. We are thinking of ways to land our website’s link to people’s Bookmark folders, Notion pages, Resource guides, spreadsheets, and digital toolkits. We are imagining a user who is “very online,” uses generative AI at work and in their personal life, but otherwise takes great care to educate themselves on everything they consume and engage with. These are people who make and share spreadsheets for fun, and are always looking for new ways to organize their chaotic digital lives. This person is likely an Are.na user (a platform like Pinterest but for designers, artists, academics, “technologists,” etc.) So in our effort to spread this link as far as wide as it can go, this phase will also identify up to 10 link repositories from all over the internet and, in parallel, build an dedicated Are.na board for the project where we will collect research papers, notes, and inspirations for the visual identity of the project as a way of slowly and quietly embedding the project into a platform that will connect it to its eventual user. 

The Are.na board will also serve as a collection of inspirations and resources for the project itself, which will be used to develop a visual identity (colors, typefaces, imagery) for the deck Sasha will use to present last semester’s version of the project during NYC Open Data Week on March 25. For this presentation, we’ll also add a slide with a QR code that links audience members to a form where they can enter their email to stay up to date on the project’s future. By the end of this phase, we hope to have a robust Are.na board, a list of link repositories, and 50 emails that reflect our in-community outreach efforts.

We have decided that, for the state of this project, it is not worth building an Instagram profile or comparable social media platform. Future iterations of this project might benefit from building a social media presence, especially on Instagram. But, for now, the effort required to make it worthwhile is simply not commensurate with what we can expect to get out of it. A successful Instagram launch requires near-daily posting across Posts, Stories, and Reels. High-quality assets and constant engagement and we’d risk distorting the project in order to satisfy the platform’s needs.

Instead, we’ll focus on “planting” the website’s link all over the internet and setting it up to spread like spores in the wind. We will submit it to any and all relevant link repositories, email it to our mailing list of what we hope are at least 100+ interested parties, and treat each page link within the website as an opportunity to share the project: We expect to circulate the visualizations, literature excerpts and editorial components on Are.na, where the behind-the-scenes posting will have laid the groundwork of audience building. These include We will email or mailing list when we build a Coming Soon page, to invite folks to the final presentation, and on launch day. We might consider drafting a brief newsletter series adapting our web copy for the following topics: On Methodology, Hallucinations As Cultural Artefacts, “Hallucination,” The Information Ecosystem (which for web-dev purposes will be finalized by April 30th). Or, we might produce a print pamphlet version of this copy and visualizations — to be determined in the final quarter of the project’s lifespan. Our hope is that our professional lives and networks will thus earn us more opportunities to share and discuss this work, and that every phase can benefit from the last.

DMP for Life

I have to confess that this week got away from me, and I ended up being much busier with my day job and thwarted my plans to steam company time and join Sasha’s skill share session, a GitHub and VSCode walkthrough that would prove foundational to our data curation efforts further down the line. But Sasha was generous enough to record the session and include the links into our shared documents, so I am all set to walk myself through the workshop and touch base with questions for her next week. All of this to say the first data-management technique I need to apply to everything I do, is keeping a better calendar!

All jokes aside, from some creative coding classes and working in an editorial context where a piece of writing goes through as many as four versions, each of them shared among multiple parties and requiring strict version control, I had picked up on a few tricks over time: Smarter file naming conventions, copies spread out across multiple sources and locations, documentation (README’s especially), version control. But Steve Zweibel’s Research Data Management presentation was so useful I saved the link to all my workspaces. Just having access to the language of data management — types of research data (by origin or form), data life cycle, FAIR principles, etc. — gave some purpose and unification to what have long been a set of haphazard practices I use to work.

When I think about data degradation, I am more likely to think about link rot or bit rot, the actual deterioration and degradation of the mechanisms of saving information. But Michener’s “Data Entropy” diagram charts how distance from data’s original context and the people who created it are factors that data management is meant to mitigate. We take pictures, keep scrapbooks, memento boxes, travel souvenirs, and ask our elders to rehash the same stories so we can keep that immediacy that binds us to source of information. Like links, bits, and pictures, any form of data collection is subject to deterioration. So in truth, link rot is not too different from dry rot and in both cases, “future you is your first user.”

bio + contributor statement

I feel awkward about going into talk of myself with no preamble. So here are some lines. I notice how most people work around this shyness but referring to themselves in the third person. Like this: Michelle Santiago Cortés is a writer and editor living and working in New York… But once I get to the second half of most of the bios I’ve been asked to write, the first person feels more authoritative and immediate. But enough of that. The bio:

My name is Michelle Santiago Cortés. I am freelance writer and editor working in art and technology criticism. I regularly write for New York Magazine, ArtReview and work as a contributing editor for Lux magazine. I earned my B.S. in Magazine Journalism from Boston University in 2018, and have since worked in art publishing and digital media, writing any and all kinds of blogs, articles, features, essays and interviews. This extends into in-depth knowledge of distribution channels and techniques like SEO, newsletter, social media, and print. As an editor, I work primarily with artists and other “non-writers” to help them take small steps into public writing. My primary research interest is technology criticism (the history and practice of) and extends into art, theory and experimental publishing.

For the AI Hallucinations Project, I am taking lead on all things outreach and promotion. I mostly earn a living as an internet culture and technology critic, and will lean on almost ten years of connections and insights. My work — whatever shape it takes — is dedicated to promoting critical discourse around technology and emerging media. The AI Hallucinations Project, in a sense, is a work of tech criticism and it will be promoted to audiences with documented interest in such discourses. I am mostly drawing from what I learned in my self-publishing practice by targeting the online communities and distribution channels that offer a direct line to “our people,” leveraging word-of-mouth, as a way of building a community that will eventually gather around a more traditional platform like an Instagram profile, in addition to the website.

Participation and Promotion

I’ll be honest: I often look at the tech world’s obsession with “move-fast, break-things” work styles with suspicion. Why the fetish for being reckless and ignorant? I always wondered why one can’t move quickly and thoughtfully. I like knowing what I am going to do before I do it. And I usually trick myself into diving into projects by planning out every last detail until I’m suddenly in the thick of it.

But I welcome the changes this project brings, a trial-by-fire style of simultaneous thinking and doing. One Week, One Tool offered some assuring insight: You’re not suppose to know what you’re doing. Everyone had their own little bubbles to think about what they were doing, to meditate and check-in with themselves. I have a feeling that as long as I am doing something, it will be moving the needle somehow.

We met for some time last Friday. But that feels like time spent thinking together that in addition to the time we spent discussing the project in class, and we are familiar enough with each other from sharing a class room (or in my case with Chris, two classrooms) last semester. I think we have more warming up to do, it’s so hard to jump into conceptualizing and ideating a project cold that I can see how One Week, One Too’s approach to do-then-regroup is a more efficient in melding a group.

I am more excited to learn from this project than I am to contribute to it: I write tech criticism for a living, which includes of AI-hype bubble-bursting. (My first contribution to the project: open up the discussion on the merits and implication of using the word “hallucination.) But, shamefully, I have no hands-on experience working with chatbots and LLM’s. So I look forward to getting busy to do my share of data curation, to taking in, absorbing, learning instead of leading. Even in my my roles as documentation and outreach lead, I want to subordinate to the role of data curator. I think that is the ideal setup for any outreach or documentation lead — to be knee-deep in the work you are endeavouring to promote, or even “brand.” The promotion should extend naturally from the work being promoted and I have a lot of faith and enthusiasm for this project.

Skillset Post – What I can (and can’t, but want to) do

I am a writer and an editor, with a focus on internet and digital cultures, and art. My skills include:

Copy writing: I can write anything — web copy, social media copy, newsletters. Essays, presentations, critical papers, white papers. Anything! In English and Spanish (and French if I have a dictionary.)

publicity + social media management: I’ve been working in media for over seven years. I know the ins and outs of getting media coverage and fitting into the “discourse.” My specialty (and years of contacts) are deeply tied to all things art, culture and technology. I know the landscapes well and am confident I have the insights and contacts to help our project reach the people who would be most-interested in it. While I doubt we’d ever pursue media coverage in a traditional sense, I know how to build a community using social media and am familiar with all the platforms and alternatives ways of building engagement.

publishing: Newsletters, zines, printed matter. Social media is great but publishing is sometimes a more effective means of creating a community around a project and putting it in people’s hands. I know how to produce small-runs of printed matter as well as set up digital zines, newsletter platforms etc.

project management: I have experience managing groups and intricate workflows.

Research: I am working journalist (and budding academic), and I feel confident in my research skills: courthouses, archives, libraries, social media rabbit holes.

Development: Not very robust here, I know the bare minimum of HTLM, CSS, etc. I can read through a GitHub and get the gist of whats going on. Would love to develop these more in an assistive capacity. Don’t have the knowledge to lead anything here but I am eager to refine these skills and be of substantive assistance. I am particularly looking to get close to any data cleaning, dataset-curating, or database-related projects.