Thank you. I have no slides because I like to keep my talk flexible so that I can rewrite it after the conference breakfast or reflect questions people ask this morning. All right. The ongoing information revolution, or don't panic, the world ended in 1551.
Despair is endemic today, especially in younger people. College and high school age students face the triple fears of the climate crisis, authoritarianism, and GenAI rhetoric claiming that no jobs will exist for them by 2030. Many young people say they are not planning for a career since there will not be a future to live in. Crisis rhetoric inundates us with talk of the end of labor, the end of civil society, the end of human self-governance, and the end of the world.
Two things I think are feeding this above all. The first is that we lack good models between replacement and rejection. So, many characterize current attitudes toward GenAI as polarized, with proponents saying GenAI will saturate every facet of our lives and opponents rejecting all GenAI presence like a contaminant. Yet both extremes are symptoms of our lack of alternate paradigms for thinking about technological change.
As a historian of the print revolution, I am often asked how long it took for print to replace manuscripts and for manuscripts to die out. The answer is, how many blank books were on sale the last time you went into a Barnes & Noble? We in 2026 are producing two orders of magnitude more manuscripts per year than the Middle Ages ever did.
Print did not replace manuscript. It joined manuscript production in a world where we now use each technology for the things it is best for: print for news and formal circulating literature, manuscript for personal writing, rough drafts, many official documents that need to exist in unique instances, note-taking, and aids to memory. There was no replacement. We use both.
The same misunderstanding applies to the ebook revolution. When ebooks first took off commercially, everyone talked about the imminent death of the paper book. But here we are in 2026, and the paper book continues to be produced in greater quantities every year than before, with the exception of 2021.
What died was the disposable glue-bound mass-market paperback produced for magazine racks, since ebooks proved more practical for disposable reading. This partnered with an increase in the production of beautiful collectible hardcover and high-quality trade paperbacks, whose elegant designs bought to grace our bookshelves are selling in greater and greater quantities every year.
New technologies join old ones. They rarely replace them. And when they do, that replacement is gradual. Yet the rhetoric around GenAI, both pro and con, centers on a binary of replacement or rejection. More realistic models of coexisting technologies can help even GenAI opponents envision less stark absolutes.
One hostile yet fruitful model for imagining a possible future for GenAI is radioactivity. When we first discovered radioactivity, we put radium and plutonium in everything: in wristwatches, in hair tonic, in whiskey glasses, in face cream, and over 95% of what we put it in turned out to kill us and destroy our environment.
We then removed it from the vast majority of the applications we tried, but continued to use it where fruitful: in power plants, in cancer treatment, in chemistry, in tagging, in space exploration, and in industrial processes, which mean that radioactivity now touches all of our lives daily, but as a second-order presence, powering the systems that created the phones in our pockets, not the phones themselves.
We can imagine a similar model for machine learning as studies confirm the dangers of things like AI psychosis, hallucinated misinformation, and LLM-linked skill degradation, yet confirm the utility of machine learning in protein folding, pharmaceutical research, geographic mapping, and other arenas more removed from firsthand use. Such a radioactivity-adjacent trajectory is plausible.
And having such a model to simply imagine a possibility between replacement and rejection helps us think in a less polarized and binary way about the use and misuse of a new technology. All right. I promised to tell you a story about how the world ended in 1551.
It feels like the current information tech revolution is turning our world upside down, and many leaders within the GenAI world discuss the so-called singularity: a moment that can be hyperbolically described as the moment when we create an AI god which takes over the Earth, but more concretely as the moment when our tools make other tools, our creations uncover new knowledge, our computers do the research for us, and the knowledge in our data banks outstrips the knowledge in our heads.
Singularity theorists argue that it will be an absolute line drawn across human history when we enter the era when the solutions that advance our studies and the texts that change technologies that change our lives are understood by no human, spawned from a science that no human mind can encompass. Knowledge will leave humanity behind, making humanity no longer master of our destiny. In fact, this happened in 1551.
The Gutenberg Press caused an exponential acceleration of book production after 1450. So, the number of books that had ever existed on this planet quadrupled between 1450 and 1510, doubling again each decade to the 1550 crisis point. Before 1550, Europe's definition of an educated person was someone familiar with all the books.
An educated gentleman or lady had read the majority of books that existed. A vast and universal common core anchored in Plato, Boethius, Augustine, and Erasmus. Well, there were also some specialist books like medical or law textbooks or Bible commentaries, which only a few people read thoroughly, but all learned people knew about them, could name their titles and their authors, could describe them. Each learned mind was a living and complete card catalog of all there was to know. That's what education meant.
But by 1540, there were too many books. One human being could not read or know all the books. You just couldn't. There was too much knowledge for anyone to know. Everyone in the learned world freaked out.
It was the end of unity, the end of shared knowledge. No one would ever be truly wise or learned ever again. Human knowledge would forever fragment into innumerable segregated and mutually unintelligible factions. And learned people would no longer have a common ground. Each citizen formed by reading a unique sub-slice of human culture and thus in effect living in a different world.
There would soon be as many realities as there were readers, as many interpretations of the law as there were citizens, as many Christianities as there were Christians, and no argument could ever be won, lost, or even valid since the two parties would not have the same data shaping their arguments. Humanity could no longer understand its world, its tools, or itself. Knowledge had outstripped us. This should sound familiar.
Tools to navigate this crisis became popular immediately. In the early 1500s, as this is happening, we see books of quotations being published, books of summaries, books that offered one-page digests of other books. Does this sound familiar from what people are using LLMs to do, to generate one-page digests of books? It's exactly the same problem.
And in 1550, exactly one century after Gutenberg first printed, Conrad Gessner, the most prolific author and scholar of the half century of this crisis, published his great solution to the problem: the Bibliotheca Universalis, or universal library, a compendium which was to list all books so that an educated person could invest in it as a library within a library. And thus, if you did not know all the authors and books, you could look them up and find the books that contained all knowledge.
The dream was that the volume would solve the crisis of there being too much to know, and put all knowledge back in an educated person's fingertips. The massive tome would be updated with a little slim supplement each year that one might subscribe to like a magazine with new books if there were any, and all educated people would have this and be united by it.
This great solution to the universal crisis became obsolete within a year as the 1551 supplement with titles of new books gathered for the first time threatened to rival the original in length. Gessner had planned for linear growth, but exponential was the brave new world of printing and mass production. And by 1551, the ideal of a universal library died.
And once again, learned Europe freaked out. They really and absolutely, the letters are just incredible, freaked out. Panicking in exactly the same way many of us are panicking now and facing the same problem that the singularity crowd has been discussing for 30 years: knowledge outgrowing individual capacity to know. In 1551, scholars declared this the end of the world.
Nor were they wrong. They were not wrong. The branches of knowledge did fragment. Education as Gutenberg and Gessner knew it did become impossible. Every educated man and woman was shaped by a unique library and education with an audience of one. The world did turn upside down.
Now, none of our textbooks tell us the world ended in 1551. Nor will they in future say the world ended in 2026. Nor was 1551 an averted apocalypse like the successful interventions in ozone depletion and Y2K which prevented crises. This crisis of print outpacing reading came. Libraries outgrew learning. This was not averted, it was adapted to, as we changed the definition of what it meant to be learned and reorganized our tools of knowledge seeking.
In response to this crisis, we formalized the scientific method within a few decades. Not by coincidence, but as a cause. The scientific journal was born at this moment, presenting digests of findings and distilling out facts and results without the elaborate interpretive apparatus which made past scientific treatises into fat tomes.
The general word scholar declined in usage as people identified themselves more by specialty: a philosopher, not a philologist; a chemist, not an astronomer. The subject subdivisions multiplied. History was no longer considered by universities a subsection of ethics, nor music a subsection of mathematics.
Taxonomies of knowledge proliferated, defining specialties and subspecialties, and the growing genre of the encyclopedia transformed from fashion to the necessity that we all, depending upon Wikipedia as we do, taste every day. We made tools. We redefined learning. We adapted and we kept going.
In another parallel to today, our fears of algorithmic news making each person consume as an audience of one curated news. The printing press also gave us too much to know in terms of news. It accelerated and fragmented news as pamphlets spread facts and false facts on an unprecedented scale in the first half of the 1500s.
When Gutenberg began printing, it took 14 days for news to move by post rider from Florence to Milan. A mere 65 years later it took 17 days for Luther's 95 theses released in Wittenberg to be in print in London. Everything reached everywhere all at once.
And pamphlets carried not only explosive ideas like Luther's, but misinformation, false announcements of plagues or invasions, dangerous medical nonsense. By Shakespeare's day, closing on 1600, one generation after Gessner, fake news was such a huge object of crisis that he brings it up in his plays for his anxious audience over and over.
Showing in his Henryad how flawed leaders are deceived by rumor, while the perfect King Henry V only believes official reports. Well, in his Winter's Tale, Shakespeare satirizes the folly of uneducated peasants who believe nonsense about flying mermaids, saying, quote, 'I love a ballad in print, for then we know that they're true.'
As printing presses multiplied, so news, both fake and real, did, too, and people felt at the time that there were now as many realities as there were news readers. This too feels to us very familiar. The solution? In one sense, fake news was never solved. It has existed since the primordial invention of the town crier, and it always will.
But the main solution to the print revolution's misinformation crisis was a pair of tools now so ubiquitous we forget what they were invented for: the newspaper and the magazine. News readers, it turns out, are smart, aware of misinformation, and want reliability. They know how to curate reading.
The regular repeating newspaper let print revolution era audiences choose to get their news from a specific known editorial team, chosen based on its record of reliability. A subscription model meant steady income for printers and journalists, while the need to maintain reputation held newspapers to account.
The world's first magazine, The Gentleman's Magazine, founded in 1731, was created as a fact-checking service, reporting each week's news with notes comparing the account given by each major newspaper, presenting newspapers' disagreements side by side for the judgment of the savvy news consumer.
Newspapers, and especially pamphlets, continued to sometimes spread misinformation, but the public cared, compared, judged, had an apparatus for judgment, and navigated the news ecosystem with skill. Not always right, but enough to reach a kind of consensus, so that learning and news still had an audience of one.
The Venn diagrams of knowledge nonetheless overlapped enough with tools like The Gentleman's Magazine forming that Venn diagram center for stability and democratic debate to thrive. The magazine, the first magazine, also included op-eds from famous op-ed writers like Voltaire,
discovery reports from science and archaeology, scores of popular songs to sing in pubs, and created a new voluntary shared culture serving much the function that the universal literary core had served two centuries before.
This solution to both the crisis of news and the crisis of there being too much to know was not technological. It was organizational and reputational sorting systems invented and chosen by savvy news consumers. It took a while, 120 years in fact, from Gessner's failed Bibliotheca Universalis to the founding of The Gentleman's Magazine, but it worked.
And over those 120 years, many other tools for finding consensus were developed in different regions and used by different communities. Think now of how studies of the last couple years show the young people anxious about the end of the world are increasingly rejecting algorithm-sorted media, recognizing that the machine which curates their flow of information for an audience of one is manipulating, isolating, and harming them.
Social media use is declining, especially among the young. Analog communication, in-person, face-to-face community formation, these are growing as our participation in chosen communities like Discord communities.
We can cultivate that savvy distrust of algorithms as the encyclopedia, newspaper, and magazine did in the past, if we are willing to accept the reality that we have all long been audiences of one. But we can nonetheless create shared knowledge communities through structures of curation, fact-checking, and reputation that incentivize reliability and community.
Our Venn diagrams will remain one circle per human being. But organizational tools can make a growing area of overlap. Healthy journalism is one of the most powerful tools to achieve this.
Our journalistic ecosystem has recently been ravaged by monopoly capitalism and before that by the often overlooked economic disaster as the transition from print media to digital meant that government subsidies which had long sustained paper news, which in America was never economically self-sustaining. It always depended on government subsidies, lost those subsidies because those subsidies were tied to paper, print, and media mail, which funded newspapers and did not switch over to provide a new subsidy for digital and website hosting.
Reconstituting journalism can transform our attitudes, creating new common ground. And my favorite thing about this solution to the crisis of truth is how democratic it is. Since the newspaper and the magazine were created bottom-up, neither by government nor by oligarchic profit chasers, but from the innovation of journalists, truth-tellers, and the demands of a smart public. A smart public is something we've always had and always will.
To get concrete for a moment, one project I've been part of lately is called Fix the News. It's a journalism team founded in Australia focused on the fact that our click-based news ecosystem promotes bad news 100 times as much as good news, amplifying crises and losses while drowning out successes and victories.
This, not reality, births the doom scroll. We see all the extinctions and disease outbreaks, but rarely hear about the successful species restorations and disease eradications that are happening at the same time. The malaria vaccine has saved more lives since 2021 than any other undertaking in all of human history.
A recent UN study confirmed that two-thirds of our efforts to save endangered species have succeeded, and those species are now fine. The solar rollout has progressed so much in the past two years that we are past peak carbon, and global emissions are in fact going down.
We never hear this news. Young people, especially, never hear this news. I often talk to students who say that they would love to work for some eco project, but there's no point because they never succeed. They do succeed the majority of the time, but kids never read about it.
Psychology, our habits as what we click on, what we share, mean good news doesn't get the clicks. So, our experiment, we simply made a newsletter that rebalances this by offering a weekly roundup in your inbox of the globe's underreported good news: river cleanups, civil rights victories, life-saving projects you would never know were happening getting your news in normal channels.
Fix the News subscribers have described the newsletter as life-changing. Our measures show it improves mental health and physical health, stabilizing blood pressure, even reducing the frequency of injuries and how often people get a cold because it reduces the stress of the doom scroll.
This is an example of an organizational information solution, like The Gentleman's Magazine, and well within our power to create through grassroots human bottom-up effort. Something with a larger algorithmic or GenAI tool behind it could do so on a much larger scale.
Turning to conclude with labor rather than information, this is the facet of GenAI which matches the Industrial Revolution more than the Print Revolution. And here historians are advancing the same takeaways in every past labor revolution from the industrial back to the reordering of labor after the Black Death. Policy and empathy are everything.
Will post-pandemic, post-GenAI labor be empowered or disempowered? Will wealth gaps grow or diminish? It would be easy and kind of relaxing if the answer was that the trends are always the same. But historical reality demonstrates that policy determines all.
After the Black Death, workers gained power in some places, and in others policy stripped workers of power and rights, even of land and liberty, binding them to their jobs and creating the first debtors' prisons to enrich elites. Well, across the sea at the same exact moment of change, a different policy enriched the poor majority. The future of labor is a policy question: corporate policy, local policy, and government policy.
A society that cares can declare it unacceptable for change to leave people behind in desperation caused by loss of old employment. A society that cares can choose to pass policy to make great moments of change inclusive, or it can leave people to starve. Helping people helps the future.
For those concerned by the question of maximizing Earth's computing power, often pointed to these days as a great metric of progress, all the data centers we are building in the world aren't changing the fact that the vast majority of computing power on this planet is human brains, human neurons, the majority of which are focused on the problems of daily bread, and dimmed by poverty's desperation and its cognitive toll.
Ending poverty would free that computing power to instead focus on building the future, freeing four orders of magnitude more computing power than is under construction in the entire AI boom.
Returning to the crisis of truth, we can live with every human being being an audience of one. In fact, we've been living in ongoing stages of this information revolution for almost 500 years as technology succeeded technology: radio, telegraph, television. We are now a species native to information revolutions.
Each new stage has made us invent new tools to sort truth and build consensus. Many stages of this ongoing information revolution were filled with fear and tumult as we feel now, but they weren't the end of the world. What can be the end of the world for a community or family or city is to be left behind by policy when whole lives are judged expendable.
The last half century of international interventions and studies has gifted us a staggering portfolio of policies that can help transitions leave no one behind. UBI being merely the most famous of hundreds of ideas, tested and proven ideas.
It is a scary and a wearying feeling to be told that policy is everything. That our decisions right now really matter. It would be much more relaxing if I could give you a graph with a nice wiggly line or hockey stick curve to say that history grinds on like a great gear and the outcomes of labor and information revolutions inevitably go like this.
Real history simply shows us that outcomes are plural. But if the arc of history is not yet bent as much toward justice as we wish it would, it has been toward wisdom. Experience, more policy options than we can bring together with the message.
Yes, it's a crisis. Yes, our Venn diagrams are drifting apart. Yes, a world is ending. But our lives, our fortunes, and our duties are not. We are much better prepared to endure this stage of the ongoing revolution than any generation before us. Nor is major political reform a pipe dream.
Despite authoritarianism surging around the world, because studies clearly show us that in the decades since the development of the cell phone, that great empowering organizational tool, when countries have faced an authoritarian turn, by 10 years later 75% of them ended up more liberal than they were before due to the strength of the backlash and resistance.
If we look at only the last 15 years instead of the last 50, it's 85%, in which authoritarianism is weaker than the backlash that sweeps it away. The momentum toward change is palpable and powerful. We just need to use it.
And to face the exhausting, burdensome, but liberating truth that policy is everything. Our policy on the corporate, local, and political levels. We must use the great wisdom we have gained in the past to put empathy in power, to make new common ground for our Venn diagrams, and to leave no one behind.
The good news is that no generation has ever been so ready to achieve this. Thank you.
