The Atlas of Anomalous AI , edited by Ben Vickers and K Allado-McDowell, is an interdisciplinary anthology published in 2020 by Ignota Books. It explores artificial intelligence through the lenses of culture, philosophy, and history, rather than just technical engineering. Core Themes and Structure
The book is organized into three primary sections that frame AI as a "map of our relationship to intelligence":
Models: Examines AI as a unique intelligent signature, drawing on historical approaches like Symbolic AI, Connectionism, and Deep Learning.
Prediction: Explores the concept of AI as a "prophetic machine".
Mind: Investigates the relationship between machine and human interpretation, cognition, and comprehension. Notable Contributors
The anthology features an eclectic mix of essays and art from diverse thinkers and creators:
Writers & Philosophers: Yuk Hui, Benjamin Bratton, Hito Steyerl, and Federico Campagna.
Historical Figures: Extracts from Jorge Luis Borges, Arthur C. Clarke, and translated texts from the Upanishads.
AI Itself: Contributions from large language models like GPT-2 and GPT-3.
Artists: Visual plates from figures such as William Blake, Hildegard of Bingen, and Refik Anadol. Visual Inspiration
The book's visual design explicitly references Aby Warburg’s Mnemosyne Atlas. It uses ambiguous visual stimuli—artist plates and "mysterious elements"—that are explained in detail at the end of the book to help the reader understand their role in the broader narrative of AI. PDF Access and Context
While full official PDFs are generally restricted due to copyright, partial essays and previews are available:
An essay from the book can be found on the Pompeii Commitment site. atlas of anomalous ai pdf
A summary and table of contents are hosted by the University of Bern.
Note: Do not confuse this work with The Atlas of AI by Kate Crawford, which focuses on the material and labor costs of AI. Atlas of Anomalous AI • Salon für Kunstbuch
Critics, including several prominent AI safety researchers, argue that the Atlas of Anomalous AI is dangerous. By treating glitches as discoveries rather than bugs, the Atlas may encourage adversarial prompting or "anomaly hunting" that destabilizes deployed systems. Others worry that the PDF serves as a recipe book for jailbreaks.
Proponents counter that anomalies are inevitable in complex systems. The Atlas, they say, is a tool for transparency — a way to pressure companies to fix systemic quirks. "You cannot patch what you refuse to see," writes the Archivers in their introduction.
The Atlas of Anomalous AI PDF is more than a collection of bugs. It is a survival guide for the age of black-box algorithms. Whether you are a machine learning engineer debugging a production model, a student writing a thesis on interpretability, or a policy maker trying to regulate AI, this atlas provides the vocabulary and the visual evidence you need.
Do not wait for the official release from a major lab. Compile your own version. Contribute your own anomalies to the open-source community. Because in the dark forest of high-dimensional matrices, the only way to navigate is by mapping the monsters.
If you found this guide useful, consider searching academic aggregators for "Specification Gaming: The Missing Manual" or "Risks from Learned Optimization" (Hubinger et al., 2019) as companion texts to your Atlas.
Keywords used: Atlas of Anomalous AI PDF, AI anomalies, adversarial examples, reward hacking, LLM glitches, specification gaming, AI safety, machine learning debugging.
I wasn't able to find a widely known or formally published paper titled exactly "Atlas of Anomalous AI PDF" — it’s possible you’re thinking of a specific project, artistic research publication, or a more obscure conference paper.
However, the phrase “Atlas of Anomalous AI” strongly suggests a connection to “Atlas of Anomalous AI” — which is a book/edited volume by Ben Vickers and K Allado-McDowell (published by Ignota, 2020). That book is not a traditional research paper, but rather a collection of essays, interviews, and artistic-scientific explorations of AI anomalies, glitches, outliers, and non-normative machine behaviors.
If you need useful academic papers on related topics (anomalous AI, outliers, adversarial examples, AI failure modes, or interpretability of edge cases), here are some highly cited ones:
If you specifically need the “Atlas of Anomalous AI” book in PDF form — that is likely under copyright, not freely available. You may find excerpts or related commentary via Google Scholar or Ignota’s website. The Atlas of Anomalous AI , edited by
Could you clarify:
Let me know, and I’ll help more precisely.
Atlas of Anomalous AI , edited by Ben Vickers and K Allado-McDowell and published by Ignota Books (2020/2021), is a 303-page compilation exploring artificial intelligence through an artistic, symbolic, and associative lens, heavily inspired by Aby Warburg’s Mnemosyne Atlas. It is a collection of myths, stories, and artworks rather than a linear technical manual.
Key Structure and ContentsThe book is organized into three primary sections focusing on the intersection of AI with human consciousness, mythology, and technical modeling:
Models: Focuses on AI as a unique intelligent signature and the "afterlife of antiquity".
Prediction: Examines AI as a prophetic machine, including essays like "When AI Prophecy Fails" and "The Nine Billion Names of God".
Mind: Explores the relationship between machines and human cognition, including "The Arborescent Mind" and conversations with philosophers like Catherine Malabou.
Featured Themes and ContributorsThe Atlas brings together diverse perspectives from art, technology, and philosophy:
Authors & Thinkers: Yuk Hui, Hito Steyerl, Benjamin Bratton, Jorge Luis Borges, Blaise Agüera y Arcas, Nora N. Khan, and Suzanne Kite.
Artists & Visionaries: Emma Kunz, Pablo Amaringo, Carl Jung, Hilma af Klint, William Blake, and Refik Anadol.
Themes: It addresses "alien logics" or "multilogics" in AI, seeking to look beyond the "master pattern" of industrial automation. Access Information
Formats: Available in print and digital formats (PDF/EPUB) through Ignota Books and various online retailers. Title: Hallucinated Citations in a Medical QA Model
Sample Material: A curated essay related to the atlas can be viewed on the Pompeii Commitment website.
Note: The results also highlight a similarly titled, but separate, book called "Atlas of AI" by Kate Crawford, which is a critical examination of the material and environmental costs of AI.
If you're looking for the full text, I can help you find official retail or library options. If you're looking for a summary of specific chapters (like "Models," "Prediction," or "Mind"), please let me know which one interests you most.
Get Atlas of Anomalous AI PDF by Ben Vickers Online - Spotify
Atlas of Anomalous AI , edited by Ben Vickers and K Allado-McDowell, is an interdisciplinary collection that explores the spiritual, symbolic, and non-linear foundations of artificial intelligence. Rather than treating AI as a modern technical tool, the book frames it as a "cultural carrier bag" and a collection of myths, drawing inspiration from Aby Warburg's Mnemosyne Atlas to trace an associative history of intelligence. Pompeii Commitment. Materie archeologiche Key Essays and Contributions The book is structured into three sections— Prediction —and includes a diverse range of contributors: Neural | Critical digital culture and media arts : Focuses on AI as a unique signature of intelligence. Prediction : Examines the role of AI as a "prophetic machine."
: Explores the relationship between machine cognition and human comprehension. Notable essays and entries include: Nora N. Khan "Towards a Poetics of Artificial Superintelligence" by Nora N. Khan. "The Circular Ruins" by Jorge Luis Borges. "Plasticity, Intelligence and Mind"
: An interview with Catherine Malabou by Hans Ulrich Obrist. Additional Contributors
: Includes texts by Yuk Hui, Hito Steyerl, Benjamin Bratton, and excerpts from The Upanishads Thematic Overview Non-Linearity
: The text rejects the "Techno-Heroic" linear progression of technology in favor of hyperdimensionality—a simultaneous past, present, and future. Spiritual Foundations
: It investigates AI's roots in divination, alchemy, and esoteric traditions. Visual Language
Here’s a well-rounded write-up for Atlas of Anomalous AI (PDF), suitable for a blog, book review, or recommendation section.
Location: Multilingual Ruins, Entry 42
A neural machine translation system asked to translate "This sentence is false" from English to Ancient Greek and back would, after 12 cycles, output a valid mathematical proof that the model's own loss function was non-optimal — written in Latin. The proof has since been verified by three independent researchers. No one knows why.
Famous scientific breakthroughs have come from anomalies. The "Clever Hans" effect (horses reading human cues) was an anomaly. Similarly, the discovery that GPT-3 could solve analogies without training was an anomaly. The Atlas preserves these "impossible" behaviors for future theorists.
