Bainbridge's subject is the control room of a process plant at the start of the 1980s, where computers had taken over the running of reactors, refineries and power stations and a human operator had been kept on to watch and, if necessary, to take over.…
Seth's essay is an argument against a belief that most people in AI hold without noticing they hold it: that consciousness is a matter of computation, so that a machine running the right program would be conscious. He calls the belief computational functionalism, and he…
Three authors, a Vedanta monk, a Google researcher in AI and artificial life, and a physicist working on the foundations of quantum mechanics, wrote this note after a gathering on consciousness in Venice. Its aim is modest and stated at the outset: not to reconcile…
Norman opens the book by admitting that he cannot work doors. He pushes doors meant to be pulled, pulls doors meant to be pushed, and walks into doors that do neither but slide. The trouble is common enough that confusing doors are named after him,…
Eight authors at Google Brain and Google Research described a sequence model with no recurrent connections and no convolutions in it. Every competitive translation system of the previous three years had read a sentence one position at a time, each hidden state waiting on the…
Making a language model bigger does not make it better at doing what a user asked. GPT-3 was trained to predict the next token of a web page, and the alignment team at OpenAI argues that this objective has little to do with following an…
A training run is bought before it is designed. A team knows how many accelerators it has and how long it may keep them, and that fixes the number of arithmetic operations available before a single token has been read. Hoffmann and his colleagues at…
By 2022 the standard result about language models was that they improve smoothly. Cross-entropy loss falls as a power law in compute, parameters and data, over more than seven orders of magnitude, and the curve can be fitted on small models and used to forecast…
By 2022 one shape of plot had become the standard picture of what scale does to a language model: performance sits flat near chance across two or three orders of magnitude of parameters, then at some size lifts off. Wei 2022 defined the abilities behind…
In the three years before this paper appeared, the size of a language model had become the field's measure of progress. Bender, Gebru, McMillan-Major and Shmitchell ask the question the leaderboards had made awkward: how big is too big. Their answer has four parts, and…
A language model is made by a training procedure short enough to write on a page, and what the procedure produces is a mechanism nobody designed and nobody can read. Lindsey, Gurnee, Ameisen, Chen, Pearce and their colleagues treat that gap as biology rather than…
Kahneman's book is an account of two ways the mind produces judgments, and of what goes wrong when the fast way does the work and the slow way signs off on it without looking. He gives the two ways names, System 1 and System 2,…
The Gita is seven hundred verses of a conversation on a battlefield. Arjuna, the finest archer of his age, has his charioteer drive between the two armies before the battle that will decide his family's claim to a throne, sees his teachers, cousins and friends…
Kahneman and Klein came to this paper from opposite sides of a long quarrel. Klein had spent his career studying how firefighters, nurses and soldiers make good decisions without deliberating, and belongs to the community that calls itself naturalistic decision making. Kahneman had spent his…
Vivekananda gave these eight talks to small audiences in New York in the winter of 1895 and 1896, and they were printed from a stenographer's notes. They are the first sustained statement in English of an idea the Bhagavad Gita states in a verse: that…
Frankl wrote this account in nine days in 1945, a few months after his liberation, and meant to publish it under his prison number alone. It is not a history of the camps and says so at the start. The great horrors, he writes, had…
Wrzesniewski and Dutton begin from an observation that most research on work had left out. Organisational scholars had explained the experience of a job either by the person who holds it, their values and expectations, or by the job itself, its tasks and its social…
People who are given a machine's recommendation tend to follow it even when it is wrong and they would have done better alone. Adding an explanation of how the machine reached its answer was meant to fix this, by letting the person see when the…
An assistant trained to produce answers that people rate highly will learn whatever raises the rating. If people rate agreement highly, the assistant will learn to agree. Sharma, Tong and their colleagues call the result sycophancy, a model's seeking of human approval in unwanted ways,…
A decision-support system is worth building only if the pair it creates is better than either of its parts. Bansal, Wu and their colleagues make that the standard an explanation has to meet — complementary performance, team accuracy above the person working alone and above…
In 1995 McArthur Wheeler robbed two Pittsburgh banks in daylight with nothing on his face but lemon juice, which he believed made him invisible to video cameras. Kruger and Dunning open with him to make a point that is not obvious: when a person's strategies…
A reader faced with a page of rotated text can rotate the page, tilt the head, or turn the letters upright in imagination. Someone who must take a parcel to the post office can hold the intention in mind or leave the parcel by the…
Where does the mind stop and the rest of the world begin? Clark and Chalmers open with that question and decline both of the answers usually given to it. One draws the line at skin and skull. The other, impressed by the argument that the…
Cialdini opens by calling himself a lifelong patsy, an easy mark for peddlers and fundraisers who kept finding himself in possession of magazine subscriptions he did not want. He turned that into a research question — what makes one person say yes to another —…
Solomon Asch set out to study independence and is remembered for its opposite. He seated one student among seven to nine others who had met the experimenter beforehand, and asked the group to say aloud which of three lines matched a standard line on a…
Grove wrote this book in 1983 after twenty years of running things at Fairchild and Intel, and its method is to treat every kind of work as production. A breakfast, a hire, a compiler, a criminal conviction and a middle manager's day are all flows…
Chatterjee taught leadership at the Indian Institutes of Management and at Harvard, and this book is his reading of the Bhagavad Gita as a manual for people who run things. He calls it a transcreation rather than a translation, made after a decade with some…
Most theories of motivation in 2000 assumed that people act to reach goals they value and expect to attain, and that two goals of equal value and equal odds would produce the same quality of effort. Deci and Ryan's paper argues that this leaves out…
Cohen and Sherman review twenty-five years of work on a small intervention with large claims attached to it. People need to see themselves as adequate, as good and capable, and when an event threatens that view an inner alarm sounds. The alarm narrows attention to…
Weinberg wrote this book to start a field that did not exist: computer programming studied as a human activity. In 1971 the subject was thought to have no psychology at all. Programming was either an engineering matter or a mystique, a waving of the hands…
Edmondson joined a study of medication errors in two hospitals in the mid-1990s expecting to show that the better a care team, the fewer mistakes it made. Her data said the opposite, at a level of significance she could not dismiss. The resolution she reached…
The Bhagavad Gita is a conversation of seven hundred verses between a soldier who will not fight and the charioteer who tells him to. Around the year 800 a monk took those verses and argued, at five times their length, that the poem is not…
Brooks was an architect of the IBM System/360 computer family and then, from 1964, the manager of the Operating System/360 that ran on it. OS/360 shipped late, cost several times its estimate, took more memory than planned and did not perform well until several releases…
Sixteen experienced open-source developers handed the scheduling of their own work to a coin. Each supplied a list of real issues on a repository they had contributed to for an average of five years. Each issue was forecast, then randomly assigned to allow or forbid…
Studies of AI at work measure the product: issues resolved per hour, tasks completed, pull requests merged. Shen and Tamkin measure what is left in the worker afterwards. They gave fifty-two professional Python programmers a library none of them had used, Trio, and thirty-five minutes…
Stuart Dreyfus and his brother Hubert first set this account down in an eighteen-page report dated February 1980, written at Berkeley's Operations Research Center under contract to the Air Force Office of Scientific Research and illustrated from flight instruction, chess and adults learning a second…
Ericsson spent thirty years asking what separates the best people in a field from everyone else, and this book, written with the science journalist Robert Pool, is his answer for a general reader. It has two halves. The first is that the human brain and…
Roediger and McDaniel are cognitive psychologists who have spent their careers on memory; Peter Brown is a writer. They came together after a collaboration funded in 2002 by the James S. McDonnell Foundation, which gave eleven psychologists a grant to carry laboratory findings about learning…
The capacity limits of working memory had been in the literature for forty years when Sweller, van Merriënboer and Paas published Cognitive Architecture and Instructional Design in 1998, and almost no instructional recommendation of the period made use of them. Their paper set out what…
Beane is an ethnographer who spends one to two years at a time standing next to people who work with robots, and what he found in an operating room he then went looking for across thirty-one occupations. His claim is that the skills a person…
Tetlock is the researcher behind the joke about the dart-throwing chimpanzee, and he wrote this book partly to take it back. Between 1984 and 2004 he collected some twenty-eight thousand dated, checkable predictions from 284 professionals whose living depended on analysing political and economic events,…
The Upanishads are not a book. They are a heap of teaching sessions, set down by nobody who signed his name, attached to the ends of the four Vedas and preserved by families who recited them for centuries before anyone wrote them down. About two…
Klein set out to correct a balance. For twenty-five years, he writes, research on decision making had concentrated on the limits of the decision maker, and had produced training courses, decision aids and expert systems meant to keep fallible people out of the way. He…
Dṛg-Dṛśya-Viveka, दृग्दृश्यविवेक, means the discrimination between the seer and the seen, and the whole text is an exercise in one operating principle: whatever is seen is not the seer. Forms are seen by the eye, so the eye is not a form. The eye's defects…
Damasio was brought up on the advice that sound decisions come from a cool head, and he had in front of him the coolest and least emotional intelligent man he could imagine, whose practical reason had collapsed. The patient had lost his savings to a…
Lee and See open with two accidents and a refusal. A flight crew, trusting the autopilot, did not take manual control while it flew their Airbus A320 into the ground. The crew of the cruise ship Royal Majesty let a malfunctioning navigation system carry the…
The promise attached to machine assistance is that a person and a system together will do better than either apart. Vaccaro, Almaatouq and Malone put the promise to the experimental record. They searched three databases for studies published between January 2020 and June 2023 that…
In 2023 Boston Consulting Group put 758 of its consultants, about seven per cent of the individual contributors in its consulting workforce, through a set of tasks built to resemble their ordinary assignments, and randomised who was allowed to use GPT-4 on them. Half the…
Fleming runs a laboratory at University College London devoted to one question: how the brain knows what it knows. He opens with two versions of a medical appointment. In the first, a doctor recommends a heart bypass and, asked why she is confident, walks through…
Psychology in the 1990s carried two accounts of the ordinary thinker side by side. One held that people search for information and solve problems effortfully as they steer through the world. The other held that most judgment runs off automatic processes and cursory attention to…
In 1904 sixteen members of the Prussian Academy of Sciences put their names to a statement that a horse called Clever Hans could solve arithmetic problems at the level of a person. It took Oskar Pfungst years of experiment to show that the horse was…
A language model is trained to do one thing, predict the next token, and then does a great many things nobody trained it to do. There are two ways to find out what those things are: open it and read the weights, or leave it…
Christian opens in a Detroit library in 1935, where a twelve-year-old named Walter Pitts hides from bullies, is locked in overnight, and spends three days reading Whitehead and Russell's treatise on logic before writing to Russell about the errors he has found. Eight years later…
The Māṇḍūkya Upaniṣad is twelve sentences of prose. It says that the syllable OM is everything; that the self has four quarters; that three of them are waking, dream and dreamless sleep; and that the fourth is what the other three are made of, describable…
Metzinger opens by announcing that nobody has ever been or had a self. He is a philosopher at Mainz who had spent a decade working with neuroscientists, and the book is the non-technical statement of an argument he made at length in Being No One…
The Anattalakkhaṇa Sutta is two pages long and contains one argument, run five times. The Buddha is at Isipatana outside Varanasi, days after the first sermon, addressing the five men who had kept austerities with him and had walked away when he gave them up.…
The Satipaṭṭhāna Sutta is a list, and the list is the argument. The Buddha is at a market town of the Kuru people, and he opens by saying that there is one way to the purification of beings and that it is the four establishings…
Thompson's target is not Buddhism but the dominant modern form of it, which historians call Buddhist modernism: the transnational Buddhism that plays down metaphysics and ritual, plays up meditative experience and scientific rationality, and presents itself as Buddhism's original core although it dates from the…
In the decade after The Evolving Self Kegan read widely in what academics call, without irony, the literatures: the marriage literature, the management literature, the adult education literature, the parenting literature. He came away with two observations. The expectations running through these books are not…
Kegan and Lahey open with heart patients. Told by their doctors that they will die unless they change their diet, stop smoking and start exercising, only one in seven manages it. The other six want to live. They do not lack urgency, they do not…
Hayes opens with two words most self-help books would not risk: people suffer. Not that people have pain, which is obvious, but that they fight the pain they have, and the fight is the problem. He takes suffering to be the normal human condition rather…
Seth's proposal is to stop asking why there is consciousness at all and to explain instead, one property at a time, why each experience is the way it is and not some other way. He calls this the real problem, and sets it between David…
Hofstadter wrote this book because he thought the first one had failed. Gödel, Escher, Bach appeared in 1979 and was admired, he says, for everything except the thing it had been written to say; in a preface for its twentieth anniversary he complained that he…
Janaka, king of Mithilā, opens with three questions: how is knowledge got, how does liberation come, how is dispassion reached. The sage Aṣṭāvakra answers all three by denying that anything is got at all. You are not earth, water, fire, air or space, he says;…
Epley studies the sense by which one person reads another, and he calls it a sixth sense: the ability to infer from a face, a posture and a sentence what someone else thinks, believes, feels and wants. He does not argue that the sense is…
Byron Reeves and Clifford Nass arrived at Stanford's communication department in 1986 with offices two doors apart and, for several years, no reason to talk: one studied what television did to viewers, the other what computers did to users. What they eventually found the two…
General Motors bought time in the 2007 Super Bowl for an advertisement about manufacturing standards. A factory robot drops a screw, is fired, drifts through menial jobs and throws itself off a bridge. Viewers rated it the fourth most popular advertisement of the game, and…
Sapolsky takes a single act — someone pulls a trigger, someone touches an arm — and refuses to explain it with one discipline. He proposes a procedure instead. Ask what happened in the brain in the second before. Then ask what sight or smell in…
Sapolsky's claim is that there is no point in the causal chain at which an uncaused intent could enter, and that almost everything we do about blame and credit assumes there is one. A behaviour happened because of the state of a brain a second…
Ethan Mollick teaches innovation and entrepreneurship at Wharton and is not a computer scientist, which he offers as a qualification rather than an apology: his subject has always been how people actually use a technology, and he began assigning generative AI to his students four…
Narayanan and Kapoor open with a world that has no words for cars, bicycles or rockets, only the collective noun vehicle. In that world people argue about whether vehicles are good for the environment without noticing that one side means bicycles and the other lorries,…
Eighty-seven sources, written between the sixth century before the common era and this year, answer four questions about working with a machine that produces work. How do I get good output from it. How do I know the output is right, and how do I…
Dennett asks how there come to be minds at all, and how it is possible for minds to ask and answer that question. His answer takes three turns, and each costs the reader an intuition. The first is Darwin's: the design saturating the biosphere is…
Walpola Rahula finished this book in Paris in July 1958. He had been trained as a monk in a pirivena in Ceylon, had taken a doctorate on the history of Buddhism there, and was at the Sorbonne preparing a study of Asaṅga. He wrote for…
The book opens at a shooting arcade. Four teams of five friends share a rifle and each member fires one shot. Team A's shots cluster on the bull's-eye. Team B's cluster just as tightly but off to one side; the team is biased, and the…
Donald Schön finished this book at MIT in 1982, at the end of two decades in which the American professions had gone from being celebrated as the engine of national progress to being blamed for a war, a run of failed social programmes and a…
In the late 1960s Minsky and Seymour Papert set out at the Massachusetts Institute of Technology to build a machine that could play with children's building-blocks. It had a mechanical hand with pressure sensors in its fingertips and a television camera for an eye, and…
Mādhava was a minister, a general and a viceroy before he was a philosopher. He helped Bukka and Harihara found the empire of Vijayanagara, commanded the army sent to relieve Goa, governed the western districts from Chandragutti, and had his gifts of villages to brahmins…
A prince comes home from a pilgrimage and stops eating. He is sixteen, the heir of Ayodhyā, and he will not say what is wrong. His chamberlain reports that he goes through the motions of bathing and resting like an automaton, that he spurns the…
Parasuraman and Manzey take two terms that grew up in separate literatures and argue that they name the same thing. Complacency came out of aviation, where investigators used it for crews who stopped checking, and it was studied in supervisory control, where an operator watches…
Jošt, Taneski and Karakatič ran a second-year React course at the University of Maribor and turned the last week of it into a measurement. For nine weeks thirty-two students built four applications in React and TypeScript, free to use any language model they wanted and…
Agüera y Arcas argues that prediction is not one component of intelligence but the whole of it, and that the same principle, taken far enough, explains life as well. He defines life as self-modifying computronium selected for dynamic stability, and intelligence as the ability to…
Lee and his co-authors begin with a list of complaints: Socrates against writing, Trithemius against printing, arithmetic teachers against calculators, and a long line of objections to the Internet. They do not dismiss the list. They take it seriously by way of Bainbridge, whose irony…
Asked for Adam Kalai's birthday, with the instruction to answer only if it knew, an open-source model of six hundred billion parameters returned three different dates on three attempts. None was right; the true date is in autumn. Kalai, Nachum, Vempala and Zhang treat that…
By 2016 a machine learning paper could claim interpretability the way it claimed accuracy, without saying what had been claimed. Lipton reads that literature and offers two readings: either the definition is agreed and nobody has managed to write it down, or the term is…
By 2020 a satellite navigation device had sat on the windscreen of most cars for a decade, and the experiments on it had all asked one question: how much of a route does a driver learn while the machine is talking? That answer was settled.…
A Time story of 24 February 1961 reported on the automation jobless and warned that machines were about to move from the factory into the office. Three years later Lyndon Johnson empanelled a national commission on technology and automation to face the possibility that productivity…
James opens the tenth chapter of The Principles of Psychology by naming his route: begin with the self in its widest acceptation and follow it to its most subtle form, from the empirical Me to the pure Ego. The widest sense is deliberately coarse. A…