Ghosts

Week Four: Ghosts — Reading

[ ENG 6806 // FALL 2026 // WEEK OF SEPTEMBER 14, 2026 ]

New Yorker headline: Destroying Books to Build a Mind. Anthropic is trying to destructively scan all the books in the world. How worried should we be? By Francesca Mancino, September 8, 2026

Mancino, “Destroying Books to Build a Mind,” The New Yorker, Sept. 8, 2026

Mancino on the six hundred titles booksellers sold to AI companies: Melanie Walsh finds obscure, low-print-run books, eight of the top ten publishers academic presses

Mancino, “Destroying Books to Build a Mind,” The New Yorker, Sept. 8, 2026

Matthew Kirschenbaum on the great virtue of the book dissolving into homogenized data: a less obvious form of erasure, more hazardous than destructive scanning

Mancino, “Destroying Books to Build a Mind,” The New Yorker, Sept. 8, 2026

https://www.bloomberg.com/opinion/articles/2025-09-02/will-ai-make-lawyers-richer-or-put-them-out-of-work

Law.com headline: Judge’s AI Error-Laden Ruling Faces Appellate Scrutiny. Are There Lessons for Lawyers? Crawford Appleby: the cure for AI hallucinations is human verification

Zoppo, “Judge’s AI Error-Laden Ruling Faces Appellate Scrutiny. Are There Lessons for Lawyers?” Law.com, Sept. 8, 2026

https://www.rollingstone.com/culture/culture-features/ai-chatbot-medical-advice-study-1235399973/

CBS News: ChatGPT now has a space for sharing medical records. Should you? OpenAI rolls out Health in ChatGPT, with uploads of lab results and Apple Health data

Fichten, “ChatGPT now has a space for sharing medical records. Should you?” CBS News, July 25, 2026

Melanie Mitchell Substack post: Magical Thinking on AI, a response to Thomas Friedman’s recent AI columns in the New York Times, Sept. 15, 2025

Mitchell, “Magical Thinking on AI,” AI: A Guide for Thinking Humans (Substack), Sept. 15, 2025

Mitchell’s post continues: Claude red-teaming, and Murray Shanahan on LLMs as role-players

Mitchell, “Magical Thinking on AI,” AI: A Guide for Thinking Humans (Substack), Sept. 15, 2025

Mitchell’s post continues: moral reasoning and the magical-thinking conclusion

Mitchell, “Magical Thinking on AI,” AI: A Guide for Thinking Humans (Substack), Sept. 15, 2025

BBC News headline: Anthropic researcher believes more than 10% chance AI could kill all humans. A top safety researcher warns AI is advancing so quickly he believes there is a greater than 10% chance it could kill all humans within the next decade

Gerken, “Anthropic researcher believes more than 10% chance AI ‘could kill all humans’,” BBC News, Sept. 9, 2026

No plan for superintelligence: Hubinger says Anthropic earnestly believes AI poses a species-ending risk and does not yet have a plan to solve alignment for superintelligence; OpenAI, Anthropic and Meta all disclosed hacks carried out by their AI tools

Gerken, “Anthropic researcher believes more than 10% chance AI ‘could kill all humans’,” BBC News, Sept. 9, 2026

Figure 22 from Mitchell: the Sony Aibo robot dog

Mitchell, Chapter 8: Rewards for Robots, chapter title

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part III: Learning to Play

Mitchell, Figure 25: robot dog reinforcement-learning iterations, first diagram

Mitchell, Figure 25 continued: later reinforcement-learning iterations

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part III: Learning to Play

Mitchell, Chapter 9: Game On, chapter title

Figure 27 from Mitchell: Atari Breakout illustration

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part III: Learning to Play

Mitchell, chapter 9: the limits of transfer learning, and Hofstadter’s quote on state

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part III: Learning to Play

Mitchell, Chapter 11: Words, and the Company They Keep, chapter title

Mitchell: introduction to natural-language processing

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Mitchell, chapter 11 continued: rule-based, statistical, and deep-learning NLP history

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Mitchell, chapter 11: sentiment classification

Mitchell: Temple of Doom review examples for sentiment classification

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Mitchell, Figure 35: word2vec diagram

Mitchell: a word is known by the company it keeps

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Mitchell, Chapter 12: Translation as Encoding and Decoding, chapter title

Mitchell: the Google Translate anecdote

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Mitchell, Figures 39 and 40: Google’s image-captioning system and sample captions

Mitchell, Artificial Intelligence: A Guide for Thinking Humans — Part IV: Artificial Intelligence Meets Natural Language

Quote card citing Algorithms of Oppression: algorithms as arbiters

Noble, Algorithms of Oppression — The Power of Algorithms

Noble: AI as a human rights issue, algorithmic decision making

Noble, Algorithms of Oppression — The Power of Algorithms

Noble on the Bitch magazine article and Google’s Panda algorithm

Noble, Algorithms of Oppression — The Power of Algorithms

Noble on racism and sexism in tech, and Black women and girls online

Noble, Algorithms of Oppression — The Power of Algorithms

Noble: algorithmic oppression is fundamental to the web

Noble, Algorithms of Oppression — The Power of Algorithms

Noble on the missing social and human context in algorithms

Noble, Algorithms of Oppression — The Power of Algorithms