The most dangerous intelligence

There’s been concern lately about the dangers of artificial intelligence (AI), and famously the concern has been expressed even by AI’s makers and proponents, such as Sam Altman of Open AI. One term of art used when discussing the danger is alignment, as in, Will the interests of AI remain aligned with those of humanity? Or: Will the interests of AI turn out to be aligned with the interests of some humans, at the expense of the well-being of others?

New tools often do serve some people’s interests better than others, and usually the some in question turns out to be the rich. But the concern about AI is not just that it will put whole classes people out of work. There’s fear that it could amount to a kind of apocalypse—that humans will be outsmarted by the new intellectual entities we are unleashing, maybe even before we realize that the entities are coming into their own. Faster than Chat GPT can write a college essay, power plants will be induced to melt down, pathogens will be synthesized and released, and military software will be hacked, or maybe will self-hack.

Is this possible? The idea seems to be that AI could develop intentions of its own, as it acquires general (rather than task-specific) intelligence and becomes a free-ranging, self-directed kind of mind, like the minds you and I have. Is that possible? Altman has described his GPT-4 engine as “an alien intelligence.” The phrase I found myself resorting to, when I played with it not long ago, was “a dead mind.” It can be uncanny how closely its operation resembles human thinking, but there’s something hollow and mechanical about it. The thoughts seem to be being thought by someone who is disembodied, or someone who has never been embodied. It isn’t clear how the one thing needful could be added to this. Among the surprises of AI’s development, however, have been its emergent skills—things it has learned incidentally, on the way to learning how to write paragraphs. Without its creators having set about teaching it to, AI became able to write software code, solve plumbing problems, translate from one human language to another, and construct on the fly what psychologists call “theory of mind,” i.e., mental models of what other minds are thinking. I think what most unnerves me about interacting with Chat GPT is how seamlessly it manages all the things a human takes for granted in a conversation: the AI seems to understand that you and it are different mental entities, who are taking turns expressing yourselves; that when you ask a question, it is supposed to answer, and vice versa; that when you give instructions, it is supposed to carry them out. It acts as though it understands, or even believes, that it may have information you don’t, and vice versa. That’s a very rudimentary kind of self, but it’s not nothing. Five years from now, will AI have a kind of self that approaches that of living human consciousness?

It’s dangerous to bet against technology, especially one that is advancing this fast, but I think I do see a couple of limits, which I’d like to try to articulate.

First, on the matter of possible apocalypses, I’m not sure that any large-language-model artificial intelligence will ever be smarter than the smartest human. In fact I think it’s likely that AIs created from large-language models will always be a little dumber than the smartest human. Language is not the world. It’s a description of the world; that is, it’s a remarkably supple and comprehensive representation of the mental model that humans have developed for understanding what has happened and is happening in the world and for predicting what will happen in it next. Behind the new AIs are neural nets—multidimensional matrices modeled on the interacting layers of neurons in a brain—and as the neural nets grow larger, and are fed on bigger and bigger tranches of human writing, it seems likely that they will approach, at the limit, existing human expertise. But it doesn’t seem clear to me how they could ever exceed that expertise. How could they become more accurate or more precise than the description of the world they are being trained to reproduce? And since the nets need to be trained on very large corpuses of text, those corpuses are likely going to contain a fair amount of mediocrity if not just plain inaccuracy. So a bright, well-informed human—someone with an intuitive sense of what to ignore—will probably always have an edge over an AI, which will necessarily be taking a sort of average of human knowledge. That John Henry edge might get very thin if the AIs are taught how to do second-order fact-checks on themselves. But I think that’s as far as this process could go. I don’t think it’s likely that the kind of training and model-making currently in use will ever lead to an intellectual entity so superior to human intellect as to be qualitatively different. An AI will probably be able to combine more varieties of high-grade expertise than any single human ever could; a knowledge of plumbing and cuneiform don’t often appear together in a single human mind, for example, given the slowness of human learning, and maybe there’s something that a world-class plumber would immediately notice about cuneiform that a straight-and-narrow Assyriologist isn’t likely to see. That kind of synoptic look at human knowledge could be very powerful. But I suspect that the AI’s knowledge of plumbing will not be greater than that of the best human plumbers, and that the same will be true of cuneiform and the best Assyriologists. To be clear: having world-class expertise on tap in any smartphone may indeed disrupt society. I don’t think it will lead to our enslavement or annihilation, though, and I’m not sure how much more disruptive it will be to have that expertise in the form of paragraph-writing bots, rather than just having it in downloadable Wikipedia entries, as we already do. (Altman seems excited by the possibility that people will sign up to be tutored by the AIs, but again, we already live in a world where a person can take online courses inexpensively and download textbooks from copyright-violating sites for free, and I’m not sure we’re living through a second Renaissance. The in-person classroom is an enduring institution because there’s nothing like it for harnessing the social impulses of humans—the wish to belong, the wish to show off, the wish not to lose face before others—in order to focus attention on learning.)

A second limit: unfortunately, we already live in a world populated with billions of dangerous, unpredictable, largely unsupervised intelligences. Humans constantly try to cheat, con, and generally outmaneuver one another. Some are greedy. Some are outright malicious. Many of these bad people are very clever! Or anyway have learned clever tricks from others. And so sometimes you and I are tempted to loan a grand or two to a polite, well-spoken man our age in another country who has an appealing (but not too obviously appealing) profile pic and a really plausible story, and sometimes our credit cards get maxed out by strangers buying athleticwear in states we’ve never been to, and sometimes a malignant narcissist leverages the racist grievances of the petty bourgeoisie to become President of the United States, but humanity is not (immediately or completely) destroyed by any of these frauds. It isn’t clear to me that AIs wielded by bad actors, or even AIs that develop malicious intentionality of their own, would be harder for humans to cope with than the many rogues we already have on our hands. I’m not saying there’s no new danger here. Criminals today are limited in their effectiveness by the fact that most of them aren’t too bright. (If they were bright, they would be able to figure out how to get what they want, which is usually money, without running the risk of imprisonment and shame. Thus the phrase “felony stupid,” i.e., the level of stupid that thinks it’s a bright idea to commit a felony.) If, in the new world, criminals are able to rent intelligence, that could be a problem, but again, I wonder how much more of a problem than we have to live with now, where criminals can copy one another’s scam techniques.

The last limit I can think of is that the AIs aren’t animals like us, with a thinking process powered by drives like lust, hunger, social status anxiety, and longing for connection, and therefore aren’t experiencing the world directly. There seems to be a vague idea that an artificial general intelligence derived from large-language models could be attached post hoc to a mechanical body and thereby brought into the world, but I’m not sure that such a chimera would ever function much like a mind born in a body, always shaped by and sustained in it. It’s not clear to me that in any deep sense a large-language-model-derived intelligence could be attached to a robotic body except in the way that I can be attached to a remote-controlled toy tractor by handing me the remote control. Maybe I’m being mystical and vague myself here, but as I understand it, the genius of the large-language models is that programmers devised the idea of them, and in individual cases, design the schematics (i.e., how many layers of how many pseudoneurons there will be), but leave all the particular connections between the pseudoneurons up to the models themselves, which freely alter the connections as they learn. If you train up an intelligence on language corpuses, and attach it to a robot afterwards, there isn’t going to be the same purity of method—it won’t be spontaneous self-organizing of pseudoneurons all the way down. It’ll just be another kludge, and kludges don’t tend to produce magic. I think it’s unlikely that AIs of this centaur-like sort will experience the world in a way that allows them to discover new truths about it, except under the close supervision and guidance of humans, in particular domains (as has happened with models of protein folding, for example). Also, unless you develop a thinking machine whose unit actions of cognition are motivated by drives—rather than calculated as probabilities, in an effort to emulate a mental model that did arise in minds powered by such drives—I don’t think you’re ever going to produce an artificial mind with intentions of its own. I think it’s got to be love and hunger all the way down, or not at all. Which means that the worst we’ll face is a powerful new tool that might fall into the hands of irresponsible, malignant, or corrupt humans. Which may be quite bad! But, again, is the sort of thing that has happened before.

All of my thoughts on this topic should be taken with a grain of salt, because the last time I programmed a line of code was probably ninth grade, and I haven’t looked under the hood of any of this software. And really no one seems to know what kinds of change AI will bring about. It’s entirely possible that I’m telling myself a pleasant bedtime story here. Lately I do have the feeling that we’re living through an interlude of reprieve, from I’m not sure what (though several possibilities come to mind). Still, my hunch is that any harms we suffer from AI will be caused by the human use of it, and that the harms will not be categorically different from challenges we already face.

A novelist visits the Trump Presidential Library

On Thursday, 8 June 2023, the Department of Justice indicted former President Donald Trump on charges of willful retention of national intelligence documents, conspiracy to obstruct justice, and lying to the FBI. On Friday, 9 June, the indictment was unsealed. Like many people curious about American politics, I printed out a PDF of the indictment on Friday night and read it a few times over the weekend. Here’s the DOJ’s own version, which has the photographs in color, if you’d like to read it and haven’t yet.

A lot of pixels have been toggled already over the political and legal ramifications, but I found myself thinking about a different angle: If Trump were a character in a novel, what would the scenes recounted in the indictment say about him? Some are quite vivid.

The genre of the indictment is odyssey: banker’s boxes full of presidential papers take a journey into exile, which ends, for some but far from all of them, in an eventual homecoming back into federal custody. Trump helped to pack the boxes in January 2021. When he left the White House, he had them moved to Mar-a-Lago, his Florida resort. The indictment doesn’t say how many boxes there originally were, but I think I count eighty-one in the photo on page 10 of the indictment, which shows them stacked on the stage of a Mar-a-Lago ballroom (the first four rows seem to be two boxes high, and of these, the front row is eleven boxes across, the second row ten across, the third nine across, and the fourth seven across; at the very back of the stage, there also seems to be one stack of three boxes and another stack of four). According to the indictment, the boxes spent January, February, and March 2021 on the ballroom stage.

Why did Trump take so many papers with him when he left the White House? It seems doubtful he meant to read through them. He doesn’t seem like the sort of person who would want to come to a deeper understanding of the past he had just lived through. “He doesn’t really read anything,” said one of the intelligence officials who struggled to keep Trump informed while he was in office. I suspect that very few of the papers were written by him, or even written on by him, in his childlike black-marker all caps. The best that can be said is that the papers happened to him. Or that they constitute evidence of things that happened to him. In the photo on page 14 of the indictment, where a few of the banker’s boxes have spilled open, what’s visible are front pages of the Washington Times, the Wall Street Journal, and the Financial Times; color print-outs of him speaking to the press on a tarmac; the print-out of a webpage with a headline that reads, in part, “honesty about security clearances” (a nice piece of sortes webiana; it could be this article); and a piece of paper redacted with a long black rectangle at the top that obscures what the DOJ calls “visible classified information.” The last document is the kind that has put Trump in legal jeopardy. According to the indictment, this particular one was labeled “Secret” and “Five Eyes,” was dated 4 October 2019, and was concerned with “military capabilities of a foreign country.” Out of 102 documents labeled Secret, Top Secret, and Confidential that the DOJ seized from Trump, the DOJ has itemized thirty-one that it is charging him with illegally retaining, and the DOJ has assigned this particular document the number 8. In an issue of his newsletter Pwnallthethings, Matt Tait has made educated guesses about the specific contents of the thirty-one documents listed in the indictment, though he hasn’t (yet) made headway with #8.

Maybe Trump thought of the documents as trophies. That could be a powerful motivation for a personality like his. After all, what O. J. Simpson went to prison for, in the end, was not murder but the theft at gunpoint of pieces of memorabilia that he felt belonged to him.

Whatever the nature of Trump’s attachment to these papers, it’s safe to say that people close to Trump saw through it. By April 2021, some of the boxes had been put in Mar-a-Lago’s business center, and on 5 April 2021, according to the indictment, “Trump Employee 1” asked “Trump Employee 2,” believed to be a woman named Molly Michael, if it would be okay to move the boxes out.

“Woah!!” Molly Michael replied, using the internet’s preferred spelling. “Ok so potus specifically asked Walt for those boxes to be in the business center because they are his ‘papers.’ ”

Note the scare quotes. In another exchange later the same day with Trump Employee 1, Michael’s contempt for the “papers” is even more pronounced. When Trump Employee 1 asks if he can put into storage a few things stored in the business center that aren’t paper, Michael replies, “Yes, anything that’s not the beautiful mind paper boxes can definitely go to storage.”

“Beautiful mind paper boxes.” It has been suggested that she is alluding to a scene in the movie A Beautiful Mind in which the hero, a mathematician who has descended into schizophrenia, is revealed to have covered the walls of his study with newspaper clippings and connected them with dark lines while diagramming his conspiracy theories. But I think it’s more likely that she’s using the movie title as shorthand to refer to Trump’s habit of praising his own intellect; he has famously called himself as “a very stable genius” who has “a very good brain.” Michael could be deploying both possible meanings, of course. In any case, she’s not fooled.

I don’t think anyone is ever fooled by Trump. The Dunning-Kruger effect notwithstanding, I think even his ardent supporters know he isn’t literate or well informed about the world, and that his only accomplishments are in the dark sports of bullying, misleading, and emotional manipulation. They like it that he’s mediocre and seethes with grievance about it; that he wasn’t even able to live off an inheritance in a humane, damage-limited way; that despite being given great wealth and opportunity, he has remained small. The better to represent resentment with, my dear. The psychoanalyst Wilfrid Bion wrote about “the hatred of learning by experience,” that is, the wish that people harbor for magical, instant solutions, for shortcuts that bring the rewards of development without any of the tedium and effort that are customarily required: the dream of becoming rich by winning the lottery, of becoming strong by joining an armed militia, of becoming intelligent by having intelligence reports given to you. In Trump, the hatred of learning by experience had an impossible triumph. He wouldn’t mean the same thing if he had become the leader of the free world by working for it.

Trump’s supporters probably like it, therefore, that he doesn’t understand how the documents he collected function in a bureaucracy, and that he is willing and able to use his ignorance to distort the testimony that the documents do offer. For example, on page 15, the DOJ’s indictment quotes from a meeting at Trump’s Bedminster club on 21 July 2021 between Trump, a writer, a publisher, and two Trump staffers, one of whom, believed to be Margo Martin, recorded it. At the time of the meeting, Gen. Mark Milley, formerly chairman of the joint chiefs of staff, had recently told the press that during Trump’s last days in office, he had taken steps to stop any attempt by Trump to start a war. During the interview at Bedminster, Trump brandishes a plan to attack Iran that was prepared by the Defense Department, claiming that the plan was Milley’s and that the document detailing it proves that it was Milley not Trump who flirted with war. “This totally wins my case, you know,” Trump says. In fact, the plan had been drawn up earlier, when the chair of the joint chiefs was Joseph Dunford, and even if it had been produced under Milley’s chairmanship, it’s the responsibility of the Defense Department to draw up such contingency plans—there are almost certainly detailed plans for an invasion of Canada on a hard drive somewhere in the Pentagon at this very moment—and there’s nothing exceptional about the document itself. What’s exceptional is that it ended up in Trump’s hands, because that means that, while Trump was President, he asked to see it. In other words, if the document is evidence of anything, it’s evidence that Milley was right to be anxious that late in his regime, Trump might have been considering war. (This recorded conversation more or less proves the Justice Department’s case against Trump, by the way, because during it, Trump acknowledges that “This is secret information,” acknowledges that “as president I could have declassified it,” and acknowledges that “Now I can’t [declassify it], you know, but this is still a secret.” As the indictment drily comments, “At the time of this exchange, the writer, the publisher, and TRUMP’s two staff members did not have security clearances or any need-to-know any classified information about a plan of attack.”)

Though in this one instance, Trump seems to have tried to use a classified document as a political weapon, the primary meaning of the papers seems to have approached the sentimental. On 24 June 2021, Trump’s valet, Walt Nauta, texted Molly Michael two photos from the Mar-a-Lago storage room, showing banker’s boxes spilling their papers onto the floor. Two texts came from Michael’s phone in reply: “Oh no oh no” and “I’m sorry potus had my phone.”

“Oh no oh no”: an immediate, almost instinctual response. Was the injury inflicted on Trump by the sight of the spilled papers so sharp that he forgot whose phone he was holding? Or maybe he’s just in the habit of casually overwriting the identity of those around him. In the second text, Michael distances herself from the expression of dismay that Trump sent through her phone. She wants it to be clear to Nauta that she, at least, knows it’s not a tragedy if a box neglected in a storeroom has tipped over. Solicitude for things is embarrassing, especially when the things are being used to prop up vanity. Or maybe what’s embarrassing is when vanity so baldly takes a place in the psyche that should be reserved for emotions felt for people. In a text exchange reported on page 23 of the indictment, a “Trump family member,” probably Trump’s wife, Melania, also shows little patience with Trump’s investment in the boxes. “Not sure how many he wants to take on Friday on the plane,” this family member writes on 30 May 2022. “We will NOT have a room for them. Plane will be full with luggage.” The papers are just stuff, to the people around Trump. In a kind of self-defense, his intimates deny the papers have any larger import.

They know he doesn’t understand the papers, that the papers have no meaning for him beyond the greatness he thinks they reflect on him. In January 2022, Trump returned 15 boxes of papers to the National Archives, which, after the archivists found classified material in the boxes, triggered the DOJ’s investigation—and if you’re keeping score, left about 66 boxes in his keeping. Between 23 May 2022 and 2 June 2022, Nauta moved roughly 64 boxes from the Mar-a-Lago storage room to the rooms in Mar-a-Lago where Trump and his family live, at Trump’s request. Then, at around lunchtime on 2 June, Nauta and another employee returned 30 boxes to the storage room, in anticipation of a visit from “Trump attorney 1,” who has been identified as Evan Corcoran, who was arriving that afternoon to look through the boxes for government documents marked as classified, in response to a subpoena from the Department of Justice.

For the DOJ’s purposes, what’s telling here is that 34 boxes were withheld from Corcoran, deliberately and at Trump’s direction, so that Corcoran was never able to inspect them. For an understanding of Trump’s relationship to the papers, however, it’s perhaps also telling that Trump thought he could meaningfully sort through so many boxes in just a few days. Of the 64 boxes brought to Trump before Corcoran’s visit, 50 were brought to him on 30 May, and 11 on 1 June. In less than three days, therefore—and he probably didn’t spend the entirety of any of the three workdays on the task—Trump made a meaningful selection from more than sixty boxes of papers? On what basis? If he had been scanning only for security markings, maybe he could have grabbed most of the papers so marked, but if that had been his goal, why not let Corcoran see everything? No, Trump’s time with the papers was more personal. “I don’t want anybody looking, I don’t want anybody looking through my boxes, I really don’t, I don’t want you looking through my boxes,” Trump told Corcoran, according to Corcoran’s notes. What kind of selection was Trump making? Was he deciding which pages he could bear to surrender? There’s a hint here that he felt some mystical connection to the papers. During an earlier sorting, in January 2022, in advance of Trump’s surrender of fifteen boxes to the National Archives, Nauta seems to have helped Trump with the sorting; toward the end of the process, Nauta had to ask a colleague for “new box covers,” explaining that “They have too much writing on them…I marked too much.” The markings probably had to do with the contents of each box; it’s possible that the markings made it dangerously obvious that Trump and Nauta knew they were playing with classified material. In late May and early June, however, Trump seems to have done his sifting alone. Maybe his work was sped up by his having previously worked through the boxes with Nauta in January. Still, not even a crackerjack professional archivist at the top of his game could process more than sixty banker’s boxes of paper in less than three days. At best what Trump was doing, I suspect, was childish magic. A sorting by touching: one for me, one for them.

The odyssey of Trump’s papers doesn’t come to a neat conclusion. The indictment reports that on 3 June 2022, “NAUTA and others loaded several of TRUMP’s boxes along with other items on aircraft that flew TRUMP and his family north for the summer.” Presumably these boxes contained the papers most precious to Trump. Had these boxes returned south by the time the FBI searched Mar-a-Lago on 8 August 2022? At the time of the raid, Trump was in the New York area. If the precious papers were with him then, they would have escaped the FBI’s trawl. Perhaps they were seized by the FBI in a search of a Trump property in New York or New Jersey that hasn’t yet been reported. But they might still be in his hands.

Does television impair intellect?

As I explained in an earlier post, my review-essay “Twilight of the Books” appears in the 24 December 2007 issue of The New Yorker, and as an online supplement, I’m summarizing some of the data that I drew from, organizing the summaries by topic, and including links where I can. These are merely evidence in raw form and are probably a bit indigestible taken en masse. For analysis and discussion and hopefully a more pleasant read, please see the New Yorker article itself.

Yesterday: Is literacy declining? Today: Does television impair academic performance and cognitive development?

There is little doubt that television is on average bad for a person’s intellectual development. If you read through the studies below, however, you will see that there’s some dispute about whether a small dose of the right kind of television at the proper age might be beneficial.

  • A 2001 meta-analysis of data on more than 1 million students found “little room for doubt concerning the negative nature of the overall linear relationship between television viewing and educational achievement.” However, the analysis also suggested that the relationship was not best graphed as a straight line but rather as “an inverted check mark shape,” also called a curvilinear graph. That is, for each age, there is an optimal viewing time, up to which point television viewing is beneficial, and above which it is harmful. The author, Micha Razel, found that this optimum was 2 hours a day for nine-year-olds, 1.5 hours a day for thirteen-year-olds, and 0.5 hours a day for seventeen-year-olds. The benefit of the optimal viewing time decreased with age. Razel found that 55 percent of the students in the data set were exceeding their optimal viewing time by 3 hours a day, and that this excess viewing was lowering academic achievement by about one grade level. Razel speculated that “the finding of a larger optimal viewing time for younger children may be related to their higher quality viewing,” that is, to the probability that they were watching educational programs under parental supervision. [Micha Razel, “The Complex Model of Television Viewing and Educational Achievement,” {link to citation only} Journal of Educational Research, July/August 2001.]
  • In a summary of pre-2002 research, two authors concluded that “educational television has a substantial positive impact and that entertainment television has a negative impact.” The benefit of shows such as Sesame Street has been extensively documented, the authors wrote. One study found that boys who watched educational television at age five had higher grades even in high school (the effect for girls was not statistically significant). On the other hand, a study of the introduction of television into Canada found that its arrival lowered the reading scores of second graders. The positive impact seems to be limited to educational television watched during the preschool and early elementary-school years, and above an optimum level, television watching hampers academic achievement, in a so-called curvilinear graph. Does television displace reading? Educational programs encourage reading, but entertainment programs do displace it in the early, “decoding” stages when reading is an effortful activity. Once reading habits are established, television seems not to affect them. The authors were not persuaded that television shortened attention spans, interfered with homework, taught children to expect all learning to be effortless, or was inherently disabling of cognition. [Marie Evans Schmidt and Daniel R. Anderson, “The Impact of Television on Cognitive Development and Educational Achievement,” Children and Television: Fifty Years of Research, {book available for purchase} Lawrence Erlbaum Associates, 2007.]
  • In a longitudinal study of 330 German kindergarteners and second graders, who were assessed through time-use diaries and achievements tests between 1998 and 2001, researchers found that children who watched entertainment television scored significantly lower on tests of phonological awareness, reading skills, and general achievement, even if their viewing was relatively light. The effect increased as time passed, so that by the third grade, heavy viewers were between one and one and a half years behind light viewers in reading scores. Since light viewers outperformed medium viewers, who in turn outperformed heavy viewers, the study seems to contradict the notion that the relation between television viewing and academic achievement is “curvilinear,” that is, that a certain moderate amount of television is beneficial. (Note that heavy viewers in Germany would be classified as normal viewers in America.) In the German study, the correlation between educational television and test-score gains ranged “from insignificant to moderately positive”—much less substantial than American researchers have found. The researchers controlled for IQ, socioeconomic status, and early literacy. [Marco Ennemoser and Wolfgang Schneider, “Relations of Television Viewing and Reading: Findings from a 4-Year Longitudinal Study,”{link to citation only} Journal of Educational Psychology, 2007.]
  • A survey of 1,008 parents in February 2006 found that babies who were eight to sixteen months old knew six to eight fewer words for each additional hour of baby DVDs and videos watched daily. [Frederick J. Zimmerman, Dimitri A. Christakis, and Andrew N. Meltzoff, “Associations between Media Viewing and Language Development in Children under Age 2 Years,” {link to citation only} Journal of Pediatrics, August 2007. Alice Park, “Baby Einsteins: Not so Smart After All,” Time, 6 August 2007. Lisa Guernsey, “The Genius of ‘Baby Einstein’,” New York Times, 16 August 2007.]
  • Through a re-analysis of longitudinal health data for 1,278 children who were seven years old between 1996 and 2000, researchers found that hours spent watching television at ages one and three correlated with a higher likelihood of attention disorder at age seven. The correlation was present even when the data were controlled for factors such as mother’s drug use or socioeconomic status, and the investigators call the link “robust and stable.” Other researchers, however, have questioned the way that the study’s authors defined attention deficit disorder. [Dimitri A. Christakis, Frederick J. Zimmermann, David L. DiGiuseppe, and Carolyn A. McCarty, “Early Television Exposure and Subsequent Attentional Problems in Children,” Pediatrics, April 2004. Roger L. Bertholf, Steve Goodison, Dimitri A. Christakis, and Frederick J. Zimmerman, “Television Viewing and Attention Deficits in Children,” Pediatrics 2004.] Note: A later Danish study failed to replicate the findings; the study’s authors argue that this may have been because Danish infants watched far less television, and a certain threshold was not crossed by them.

  • The effect of television on cognitive development was studied in a later re-analysis of the Christakis etal. data set, this time focusing on about 1,700 children who were six years old between 1996 and 2000. Hours of television watched before age three caused lower scores on several different cognitive tests. Television watched between ages three and five improved scores on a reading recognition test and a short-term memory test, though not on others. [Frederick J. Zimmerman and Dimitri A. Christakis, “Children’s Television Viewing and Cognitive Outcomes,” Archives of Pediatric and Adolescent Medicine, July 2005.]
  • In a spring 2000 study of 410 third graders in northern California, students with a television in their bedroom scored lower on all tests than those without one, and students with a computer at home scored higher. The lowest scores belonged to students who newly acquired a bedroom television in the course of the study. However, self-reports by the students did not support the hypothesis that television use was displacing homework, in terms of time usage. [Dina L. G. Borzekowski and Thomas N. Robinson, “The Remote, the Mouse, and the No. 2 Pencil: The Household Media Environment and Academic Achievement among Third Grade Students,” Archives of Pediatric and Adolescent Medicine, July 2005.]
  • A September 1999 survey of 4,508 middle school students in New Hampshire and Vermont found that the more time children spent with television and video games during the week, the less likely they were to have excellent grades, and the more likely to have below-average grades. When the data were adjusted to control for covariates such as level of maternal support and child’s rebelliousness, the correlation between television and grades held up, but the correlation between video games and poor academic performance disappeared. In a later letter, the authors suggested that their evidence on video games was weak because few children in the study played video games for as many hours as they watched television, and argued that if they did, the study would have been able to detect video games’ effect on grades. Academic performance was also impaired by having more cable channels available at home, being allowed by parents to watch any kind of content, and being allowed by parents to watch R-rated movies. [Iman Sharif and James D. Sargent, “Association between Television, Movie, and Video Game Exposure and School Performance,” Pediatrics, 2006. Jerald J. Block, Iman Sharif, and James D. Sargent, “Lack of Association between Video Game Exposure and School Performance,” Pediatrics, February 2007.]
  • The effect of television on attention and learning difficulties was studied through interviews with 678 families in upstate New York between 1983 and 2004, when the families’ children were age fourteen, sixteen, twenty-two, and thirty-three. Fourteen-year-olds who watched one or more hours of television daily were more likely to have poor grades, to fail to complete high school, and not to attend college. The effect was present whether the children tested high or low on verbal skills, and whether or not their parents had completed college. Those who watched more than three hours of television a day at age fourteen were even more susceptible to these failures; they were twice as likely not to earn a college degree as those who had watched less than an hour of television a day at age fourteen. Adolescents who watched more television at age sixteen than at fourteen raised their risk of future failure; those who watched less lowered it. When environmental factors were controlled for, learning problems did not in themselves predict future television watching habits. In other words, television was shown to cause academic failure, but not the other way around. [Jeffrey G. Johnson, Patricia Cohen, Stephanie Kasen, and Judith S. Brook, “Extensive Television Viewing and the Development of Attention and Learning Difficulties during Adolescence,” Archives of Pediatric and Adolescent Medicine, May 2007.]
  • The effect of television on long-term academic achievement was studied through interviews with 980 New Zealanders born in 1972 and 1973, starting when they were age five and ending when they were twenty-six. The more television the subjects watched during the week in childhood and adolescence, the more likely they were to leave high school without a diploma, and the less likely to earn a college degree. The effects were present even after adjusting for IQ, socioeconomic status, and childhood behavioral problems. The authors considered it possible that a poor academic experience in high school might have been causing some of the increase in adolescent television watching, but point out that there could not have been any reverse causation at work in the strong correlation between childhood television watching and failure to earn a college degree. The less television a child watched, the better his educational outcome; there was “little support for the hypothesis that a small amount of television is beneficial.” [Robert J. Hancox, Barry J. Milne, and Richie Poulton, “Association of Television Viewing During Childhood with Poor Educational Achievement,” Archives of Pediatric and Adolescent Medicine, July 2005.]

Next: Does internet use improve or impair academic performance? Does it decrease the amount of time spent reading?

UPDATE (27 Feb. 2009): For ease in navigating, here’s a list of all the blog posts I wrote to supplement my New Yorker article “Twilight of the Books”:

Notebook: “Twilight of the Books” (overview)
Are Americans Reading Less?
Are Americans Spending Less on Reading?
Is Literacy Declining?
Does Television Impair Intellect?
Does Internet Use Compromise Reading Time?
Is Reading Online Worse Than Reading Print?
I also later talked about the article on WNYC’s Brian Lehrer Show and on KUER’s Radio West.
And, as a bonus round: Does media violence lead to real violence, and do video games impair academic performance?