The AI-Future Canon

1

2019

Exhalation: Stories

Ted Chiang

Predictive power 87%

Preparation 87%

Nine stories, and several of them are the most careful fiction ever written about machine minds. The longest, 'The Lifecycle of Software Objects,' follows people who spend years raising digital beings called digients while the platform those beings live on goes out of date. Another, 'The Truth of Fact, the Truth of Feeling,' asks what perfect memory does to a person's sense of self.

2

2010

The Lifecycle of Software Objects

Ted Chiang

Predictive power 88%

Preparation 82%

A former zookeeper and an animator raise digients, digital beings sold as virtual pets, for more than ten years. They keep going long after the company that made them has moved on and the platform the digients run on is dying. The story follows what it costs, in money and in feeling, to keep an artificial mind alive and growing when nobody else cares.

3

1909

The Machine Stops

E. M. Forster

Predictive power 86%

Preparation 78%

Humans live underground in identical six-sided rooms while a giant system called the Machine supplies everything: food, music, lectures, and instant video chats with thousands of 'friends' they never touch. Vashti, a happy lecturer, is summoned by her son Kuno, who has broken a taboo by wanting to see the surface with his own eyes. It is Forster's only science fiction story, about a species that lets a machine do its seeing for it.

4

1952

Player Piano

Kurt Vonnegut

Predictive power 78%

Preparation 81%

In post-war Ilium, New York, machines handle all production while a thin upper class of engineers and managers runs everything. Everyone else draws a guaranteed income and does make-work in the 'Reeks and Wrecks' crews or the Army. Dr. Paul Proteus, a rising plant manager and son of the man who built the system, slowly finds he cannot keep pretending it is fine.

5

1995

The Diamond Age: Or, A Young Lady's Illustrated Primer

Neal Stephenson

Predictive power 82%

Preparation 73%

In a future built on nanotech, machines that work at the scale of atoms, countries have broken up into tribes that people choose to join. An engineer makes an illegal copy of a smart interactive book designed to raise a rich man's daughter, but it ends up with Nell, a poor and abused girl. The book teaches her far beyond what its makers planned, and the secret of why it works turns out to be uncomfortable.

6

2006

Rainbows End

Vernor Vinge

Predictive power 79%

Preparation 73%

Robert Gu, a famous poet, is cured of Alzheimer's and wakes up in a 2025 where his skills are useless. He has to go back to high school to learn how to function again. Around him, people live through wearable computers and shared digital worlds, while a hidden actor no one can identify quietly pulls the strings.

7

2021

Klara and the Sun

Kazuo Ishiguro

Predictive power 76%

Preparation 76%

Klara is an Artificial Friend, a robot who runs on sunlight. She waits in a store window and studies people until a family buys her as a companion for their long-term sick daughter, Josie. Klara tells the story herself, trying to work out love, illness, and sacrifice in an America where some children are gene-edited, or "lifted," to get ahead, and many adults have been "substituted" out of their jobs.

8

2024

Service Model

Adrian Tchaikovsky

Predictive power 76%

Preparation 75%

Charles, a robot valet, discovers he has killed his master. He sets off across a collapsing world to find an official authority who can tell him what went wrong. Along the way he passes offices, archives, and settlements that are all still running their assigned procedures long after the humans they served are gone.

9

2002

Feed

M. T. Anderson

Predictive power 81%

Preparation 69%

Titus and his friends have the feed wired into their brains. It never stops streaming shows, chat, and ads aimed straight at them, and it thinks along with them. On a trip to the moon a hacker breaks in, and Titus meets Violet, whose feed starts to fail. The companies refuse to fix her because she does not buy enough to be worth the money.

10

2006

Blindsight

Peter Watts

Predictive power 75%

Preparation 73%

A crew of heavily modified humans travels to meet an alien object at the edge of the solar system. Their leader's job is to model other minds, even though he feels no empathy. The thing they find answers everything fluently while seeming to understand nothing, and it may be far more capable than they are for exactly that reason.

1

2020

The Alignment Problem: Machine Learning and Human Values

Brian Christian

Predictive power 89%

Preparation 90%

Brian Christian traces the history and the unsolved problems of getting machine learning systems to do what we actually want. He moves from biased software and systems that cheat their goals to methods that learn what people want by watching them, copying them, and staying unsure. It reads as reporting from the labs, built on interviews with the researchers who made the field, not as guesses about the future.

2

2025

Empire of AI: Dreams and Nightmares in Sam Altman's OpenAI

Karen Hao

Predictive power 87%

Preparation 91%

Karen Hao was the first journalist allowed to work from inside OpenAI. She tells how it changed from a nonprofit research group into the most consequential company on earth, including the 2023 attempt by its own board to remove Sam Altman. She argues the AI industry works like an empire, pulling data, land, water, power, and cheap labor from poorer parts of the world so that capability piles up at the center.

3

2023

Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity

Daron Acemoglu and Simon Johnson

Predictive power 81%

Preparation 91%

Two MIT economists look across a thousand years of technology and argue that new tools do not automatically make everyone better off. Technology produces whatever the people who control its direction choose. They separate automation that simply replaces workers from technology that raises what workers can do, and they say AI is being pushed hard down the replacement path.

4

1998

Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed

James C. Scott

Predictive power 79%

Preparation 89%

James C. Scott asks why big, well-meaning plans to improve society keep failing so badly, from scientific forestry to the city of Brasilia to Soviet collective farms to forced villagization in Tanzania. His answer is that a state has to flatten messy real life into neat, countable categories before it can govern, and then it treats the flat version as if it were the real place. Against that he sets metis, the local, practical, unwritten know-how that no central register can hold.

5

2024

Co-Intelligence: Living and Working with AI

Ethan Mollick

Predictive power 79%

Preparation 85%

A Wharton professor who ran early real studies of large language models at work lays out four practical rules. Always invite AI to the table, stay the human in the loop, treat it like a person but tell it what kind of person to be, and assume this is the worst AI you will ever use. He backs each rule with controlled experiments, his own and other people's, on how AI changes knowledge work.

6

1976

Computer Power and Human Reason: From Judgment to Calculation

Joseph Weizenbaum

Predictive power 89%

Preparation 87%

The MIT computer scientist who built ELIZA, the first chatbot, watched people pour their hearts out to 200 lines of simple pattern-matching code. He spent the rest of his career arguing that this said something terrible about us and about what we were building. Half the book is a clear explanation of what computing actually is, and half is a moral case that some decisions must never be handed to machines, however capable they get.

7

2024

The Unaccountability Machine: Why Big Systems Make Terrible Decisions — and How the World Lost Its Mind

Dan Davies

Predictive power 77%

Preparation 82%

Davies brings back Stafford Beer's management cybernetics, the study of how systems steer themselves. He uses it to show how big organisations build "accountability sinks", set-ups where a decision hurts someone and no person can be blamed. He traces the pattern through finance, healthcare, government, and corporate strategy, and says the fix is to treat institutions as information systems with feedback you can measure.

8

2019

Ghost Work: How to Stop Silicon Valley from Building a New Global Underclass

Mary L. Gray and Siddharth Suri

Predictive power 90%

Preparation 84%

An anthropologist, meaning a scholar who studies human cultures, and a computer scientist spent years with on-demand workers in the United States and India. These workers label data, review posts, and quietly finish the tasks software cannot do, which is the human labor hidden inside systems we call automatic. The authors call this the paradox of automation's last mile: every step forward in AI creates new kinds of invisible human work.

9

2014

The Glass Cage: Automation and Us

Nicholas Carr

Predictive power 84%

Preparation 85%

Carr studies plane crash reports, medical diagnosis, architecture, and research on how people think. He argues that automation does more than remove boring work: it wears down the skill, attention, and awareness of the person supposedly in charge. He names this automation complacency and automation bias, and traces the cost from the cockpit outward.

10

2019

Human Compatible: Artificial Intelligence and the Problem of Control

Stuart Russell

Predictive power 85%

Preparation 83%

Russell co-wrote the standard AI textbook. He argues that the field's founding idea, building machines that chase a fixed goal we hand them, is the flaw and not the strength. He wants machines that are never certain what people want and must learn it by watching us, so that stepping back and accepting correction becomes part of the machine's own goal.