Six Gutenberg Revolutions in a Row

Ada Palmer is a science fiction writer and historian. She teaches early modern European history at the University of Chicago, and her Terra Ignota quartet, beginning with Too Like the Lightning, imagines the world of the 25th century. Ken Liu is an award-winning science fiction and fantasy writer, the author of the Dandelion Dynasty series and The Paper Menagerie and Other Stories, and most recently the author of All That We See or Seem, the first novel in his new series. 

I direct the Existential Hope program at Foresight Institute, where we work on circulating more constructive visions of the future. I wanted to know what two people who build worlds for a living think human flourishing would actually look like, and why our imaginations keep sliding toward empire and catastrophe instead. They ended up talking about work far more than I had expected, and about the assumptions underneath it, some of them 2000 years old. I also left with a new term I plan to spread: the Etsy economy.

This interview has been edited for clarity and includes follow-up questions from Asterisk editor in chief Clara Collier.

Beatrice Erkers: There’s a trend right now in Silicon Valley of reaching for historical analogies, especially the Renaissance and the age of empires. Why those periods, and why now?

Ada Palmer: Right now, everybody feels like the world is in a state of crisis, and when the world is in a state of crisis, we tend to look at other time periods, both past ones and imagined future ones. You see enormous interest in both past and future during and after the French Revolution, when everyone is suddenly wearing clothing that’s a reproduction of what they imagine ancient Rome looked like. You see the same in the Renaissance, which is also a period of crisis, in the aftermath of the Black Death. History is a wide palette of alternatives that people reach for as a tool to help them understand our world in times of crisis.

The narrower answer is that the periods tech people keep invoking are the ones we associate with disruptive technology. When we talk about the Renaissance, we usually attribute its growth to technology. Tech people love to compare what’s happening now to a Gutenberg revolution. They love to think that a new technology they're creating is going to accelerate historical change. Just as the Middle Ages were supposedly stagnant and then we suddenly had a period of acceleration toward modernity, they imagine we're in a period of acceleration toward an exciting science fictional future that’s just around the corner. And the 19th century, another age of empire, is the other age we strongly associate with technology, via the Industrial Revolution.

There's also the fiction problem. We have an awful lot of science fiction and fantasy about empire, because empire is easy to write about. The power is in the hands of a few characters. It's very big, but it's also very dramatic. It's much harder to write fiction about a complicated plural republic with 52 interlaced political offices and everybody coming from complicated, slightly middle-class backgrounds. Just try writing fiction about the European Union and you will figure this out. So when people look back to history dominated by technology, and forward to futures dominated by technology, they're seeing a lot of models of empire. 

Ken Liu: I agree with all that, and I'd add that there's an element of justification going on. When history is invoked, it’s often a way to explain and legitimate exercises of power. When Venice was trying to become a great city, it went to Alexandria to steal the body of St. Mark, so that it could prove itself a great Christian city. When people think they’re making great transformations, it can seem a little scary, and one way to legitimate it is to say: Historically, things happen because they’re inevitable. You can complain about what I'm doing all you want, but I’m just an agent of history.

Ada: We’re all aware of the type of history we call Great Man theory of history, which centers all agency in a couple of heroized figures, and which we've been trying to move away from. But one of the biggest types of history people turn to instead is Great Forces history, in which everything is giant gears grinding, and as soon as there's a certain amount of torque in the system, the machine tips. That’s an important analytical tool. But just as much as Great Man history, Great Forces history erases agency for everyone, except perhaps economists, who get to believe they’re wielding the great forces.

Ken: That’s my favorite version: Economics as storytelling with numbers.

Ada: And if you don't have a choice in the thing you're doing, if it’s an inevitability, that’s one of many ways to accept it and justify it, both to others and to yourself.

Beatrice: How much of this is really about AI, though? That’s the revolution Silicon Valley believes it’s living through.

Ada: I think we’re wrong to separate generative AI and call it a completely new revolution. I often talk to Silicon Valley people who want to ask, are we so special that we’ve had six Gutenberg revolutions in a row, 10 years apart? And I say no, we're having one Gutenberg revolution: the computer. 

The personal computer, the internet, the cell phone, social media, and generative AI are all iterations of the same one fundamental technological change, the same way Gutenberg’s press kept having new impacts for decades through news networks, pamphlets, and eventually the newspaper. A little too much of our analysis of Silicon Valley right now is of the past five years instead of the past 20.

Beatrice: Ada, when I interviewed you before, you mentioned that Petrarch could imagine political change but not the end of disease. And what happened instead is that we ended a lot of diseases. What’s our version of that blind spot?

Ada: He thought it would be easier to defeat the Dire Horseman War than the Dire Horseman Plague. And we've found that in fact we’ve made more progress against the Dire Horseman Plague than against the Dire Horseman War. Some of the STEM problems have proved easier than some of the social science problems.

We’re not entirely missing this, but you constantly hear people encouraging STEM, STEM, STEM while diminishing the humanities and the social sciences as partners, whereas people who are really deep in it, who are scientists, understand there’s a symbiosis: We don’t make progress by funding just one of the three. There’s a perception that wealth flows toward technological innovation, rather than recognizing that wealth flows toward lots of kinds of innovation, only some of which is technological.

Ken: I don’t want to focus so much on Silicon Valley, though. The problems we have today don't stem from a few people or a few inventions. They have their roots much further back, in what I’d call the gradual mechanization of humans. 

Take AI displacing jobs. This is not something that happened in the last five years. We’ve been creating the conditions for it for a century or more. If you worked in customer service 40 years ago, you were a respected professional. Over time we turned those into scripted jobs. Once you’ve gotten things to that point, it's not very difficult to throw the human out and put in a language model. It’s not the model that caused this. It’s that we've been making humans behave more and more like machines, up and down the chain: devaluing work, taking the meaning out of it, turning it into a mechanistic process where you're a cog performing what your role demands. The pushback Silicon Valley is getting now is the tip of the iceberg. What’s really being expressed is a much deeper loss of meaning, dignity, and participation.

Ada: I’ve been using a simile lately for when people say generative AI is causing problem X, when we've been talking about problem X since before the 1980s. 

You’ve filled your refrigerator for the first time with something weird, synthetic meat, say, and the refrigerator is straining. Meanwhile there’s a guy behind the refrigerator hitting it with a sledgehammer, and he's been hitting it for weeks. But somehow everyone’s blaming the new substance rather than the fact that there is a guy actively dismantling the refrigerator from behind. There have been active dismantlings of worker strength, solidarity, unions, skill, and employee rights, systematically, for decades longer than generative AI has existed.

Beatrice: So what should we be moving toward? What does a flourishing direction actually look like?

Ken: We should not be thinking about this in terms of preserving the status quo against disruption from AI, because the status quo is terrible. We should be thinking much harder about how to restore meaning to people’s lives.

In the last few generations, most of us have become used to tying our identity to our employment. When we fight to preserve jobs, we’re treating a job as two things at once: an economic means of support and a source of meaning. We need to separate them, and think about how people find meaning outside gainful employment: from connection with other people, from a sense of control, from a sense that we make a difference. 

Why do so many people repeat, as though it were proven, that not all of us can be artists, singers, poets? Says who? Surely there are ways for all of us to live a life of meaning without being employed or paid for it.

Ada: This is something I think about a lot. Classic science fiction that engaged with the future of automation kept producing the same few visions. In the ‘60s and ‘70s, quite a few writers advanced the thesis that if mankind didn't have to work, everyone would commit mass suicide: Work was the only thing keeping humanity from self-destruction. There’s the vision of the machines rising up, the Frankensteinian robot, The Matrix. And there's the vision of humans living in paradise domes making art and philosophy all day while robots do everything else. Notice that last one assumes that, liberated from work, humans will only want to do very elite, high-status activities. Art and philosophy, not gardening and taking care of injured bats.

But look at our present: There are so many meaningful vocations people would love to do and don’t, because meaningful vocations are economically exploited. We don't pay teachers. We don’t pay people who do social work, or care for the environment, or rehabilitate injured seagulls. We don't pay our artists and actors much, except a very few. A lot of people, given a steady paycheck to live on, would much rather work at the bat sanctuary. 

Another future I enjoy playing with is: What if the future is the Etsy economy, where we all make beautiful handcrafted bespoke things and buy them from each other? Maybe you enjoy drawing pterodactyls, maybe crochet. Everybody enjoys making and doing things. A future where we all make and do things and purchase them from each other is a beautiful future.

Ken: That’s my low-art future. Not everybody has to become a bestseller. You can make things for a few people, and that can be very satisfying. Or the future in which staying home with your kids as a dad is a perfectly respectable, meaningful thing to do. If you want to stay home and teach your kids and play games, why the heck not, as long as you find meaning in it.

Ada: And if anyone says that doesn't contribute to humanity: You’ve raised kids, and they're going to go contribute to humanity. We know good, supportive parenting produces happier, more creative, more generative kids. But it’s another kind of labor we devalue, partly because it’s historically coded female, like childcare, elder care, and teaching generally. If we simply make sure everyone can have a living no matter what they do, a lot of people have things they would love to go do.

Ken: Decouple remuneration from the status of the work you do, and things will look very different.

Ada: And we have study after study confirming that when we run universal basic income experiments, this is what happens. Everyone’s afraid people will be lazy, and instead people are energized. They start businesses they’d long dreamed of, open the café, start producing art, start taking care of the town square. 

There are surprisingly many win-win investments a society can make that we don’t make, because we've trained people toward zero-sum thinking about both business and politics. When you show them something where the investment makes people happy and boosts the economy and pays for itself, they just feel suspicious. Too good to be true. 

Ken: There’s a reason so much ’60s science fiction has as its central theme the idea that humans must be made to struggle for humanity to move forward. There are tons of Star Trek episodes dedicated to the idea that being taken care of by your machines is actually a prison, and you must be put back into a position of struggle. That attitude came of age during the Cold War, and there are ideological reasons it was being pushed. We don’t have to accept it as the only way to imagine the future.

Clara Collier:  A world where humans are free to make art, study, care for injured seagulls, and so on, assumes a very high degree of automation of all the less pleasant tasks that exist in the economy, including ones that are quite difficult and involved. I’d love to tie that back to the conversation about AI and the devaluation of human labor. How do both of you think of the challenge of creating a world where humans aren't needed to do all the grunt work of civilization, but still have real power?

Ken: Our fear is not giving up power to AI (you can no more give up power to AI than you can give up power to your refrigerator), but ceding power to those who build and control AI. It’s the same with every technological transformation: The real problem is almost never the machine itself, but those behind the machine. This is not a new problem. We’ve been dealing with the concentration of power and the corrupting effects of the few deciding the fate of all for practically the entirety of our history. The answer is always the same: Devolve power to the individual, constrain the power of states to monopolize knowledge, build social technologies for collective decision-making that will prevent the few from controlling everything.

Ada: Another way to think about it is that this wouldn’t be a big change, just a continuation of a millennia-long trend. Since the earliest days of tool use, human civilization has been a long process of humans finding ways to do less grunt work, to have more time for leisure, comfortable living, and more pleasurable work like making art. Computers, printers, washing machines, bicycles, motors, horses, the wheel —  they all let us do less grunt work, and did not reduce our power. The only things that reduce our power are institutional changes, government, disempowering labor. It’s another instance of “Policy is everything.”

Beatrice: So how do we make the win-win argument actually stick?

Ada: One interesting approach is to notice that the fear is really two separate assumptions. A: There are lots of people who, if they didn't have to work, wouldn't work. B: If there were a bunch of people around who weren't working, that would somehow be harmful. 

And B goes way, way back. I was just rereading the beginning of Machiavelli's Discourses, where he agrees with Livy that if you want a thriving country, you need to found it somewhere with constant difficulty, to keep the popular classes busy and striving. Because otherwise there will be indolence, and indolence will make the populace foolish and manipulable. We're fighting a 2,000-year-old assumption that people comfortably living without hard work will somehow be harmful.

Ken: I do want to point out one possible pushback against this position. There will be a percentage of the population who derive their happiness primarily from having higher status than others. The question is how numerous they really are, and whether their unhappiness would make the system not work. I don't think that's clear. But let's not presumptively say the vision is impossible because some people won't like it.

Ada: It’s worth zooming in on people's emotional investment in inequality. The person who fears that if people don't have to work, lots of people won’t work, and that this would be terrible, is generally speaking perfectly happy reading 19th-century novels about the aristocratic class, all of whom have inherited wealth, do nothing, and have Regency balls. 

What that semi-hypocrisy reveals is an assumption that there are two categories of people: one that can be trusted with leisure, and another that becomes harmful and destructive given leisure. And that goes back to a much deeper assumption running through huge swathes of history: that there is a category of person that knows how to rule, and a much larger portion of the population that cannot understand power. 

Humanity keeps reinventing aristocracy with new justifications. A lot of what Silicon Valley calls meritocracy right now is exactly this, and it is fundamentally undemocratic.

Beatrice: We’ve been circling AI this whole conversation. What does the discourse around it get wrong, and what is it actually good for?

Ken: I’m frustrated with the conversation around AI largely because the things we focus on, such as the loss of jobs (which have already been drained of humanity and meaning) and the use of copyrighted materials (as if copyright were an adequate proxy for our dignity or non-economic rights), are so uninteresting. Many of the negative effects we attribute to the AI revolution are really the result of how we've chosen to dehumanize labor and reward capital, and AI is nothing more than a symptom of much deeper root problems. The questions I find more interesting: If this technology is supposedly so powerful, why should it be controlled by a few companies, and the few governments in charge of those companies? Concentration of control is the much bigger problem. Labels like “pro-AI” or “anti-AI” are meaningless — we should all be pro-human, and we should all be leery of humans who would try to use technology to control other humans. 

And if we're talking about flourishing, look at what these technologies can actually do. So much of the supposed productivity from AI is about writing emails and generating PowerPoint slides, workers using tokens to decompose PowerPoint slides into Markdown text. Absolutely useless stuff. 

The real benefit, and I believe there is one, is in scientific research. These machines are capable of cognitive tasks that used to be extremely costly: folding proteins, proving new theorems, exploring large state spaces of chemical possibilities. Go talk to working scientists. They will tell you AI has genuinely transformed the way they work. I wish more of the conversation focused on that. AI will not transform our lives in a positive way by writing our office emails.

Clara: I’m curious to discuss this more, because of Asterisk’s audience. We have many readers who are software engineers or work in other technical fields, and for them, AI has absolutely transformed their work. Conversely, I hear a lot of scientists these days complaining about how AI labs are overselling the benefits, especially in biomedical research. 

Ken: Like any technology, AI’s ultimate impact has to be worked out by each individual scientist, artist, student, doctor, lawyer, etc. In the same way that the internet, after the bursting of the bubble, transformed our lives in ways none of the pioneers had envisioned, AI is going to do the same — the bubble will burst, and then we’ll see, in 10 years, that the true impact will be nothing like what we are predicting now. The cognitive boost from AI will not be useful for some tasks, but it will be incredibly useful for many other tasks. It takes time, creativity, and experimentation before individual humans can figure out how this technology can (or cannot) transform their work in meaningful ways. And once again, in order for this sort of creativity and experimentation to be possible, AI must evolve in ways that allow individual data sovereignty and leave control in the hands of users, not in the hands of a few large companies and their state patrons.

Ada: Here I really like Cory Doctorow’s terms centaur and reverse centaur, and I think they do a great job describing why some people are finding generative AI very helpful and others are finding it loathsome. Here centaur means when the human is on top supported by the machine, helped to do more, and that’s what a lot of scientists using machine learning or programmers using coding tools are finding, the tool helps them do more as the horse part lets a centaur run faster. A reverse centaur is when the human is in the loop to support the machine, like when people are tasked to correct LLM-generated material, or run errands to the machine’s direction; the horse is on top with the human running at the bottom. What we keep finding is that people using generative AI in a centaur mode often like it, while those who have miserable experiences tend to be those put in the reverse centaur position. Cory has a great book analyzing this more.

I’ll also add another arena where generative AI is useful: small-scale bespoke tasks where there's no budget. I saw a really good use of an LLM where the city of Bowling Green, Kentucky, gave a huge survey to every citizen, got about 8,000 responses back, and had the LLM sort them. Their office with its tiny staff could never have gone through those themselves. It let them discover that the top two issues in town were funding for the English-as-a-second-language program and one intersection with bad traffic, neither of which was on any politician's radar. Masses of information sorted on a small scale to solve a social science problem. 

And here’s a middle-ground model I’ve been enjoying, from a comparison someone made on Bluesky. We keep talking as if the two possible futures are that generative AI transforms everything, or the bubble pops and it vanishes. But consider the trajectory of radioactive material. We discovered radium and uranium, and we were so excited. We put them in power plants, in weapons, in face cream, in cough syrup, in wristwatches. Well, 90% of what we put them in turned out to kill us, so we dialed it back by 90%. But the 10% where it was really useful, we still use, and all of us benefit from it every day: power, medicine, space systems.

I can imagine that future for generative AI. It’s really bad as a therapist, really bad as a teacher substitute, and PowerPoint is bad for all humans. But it's really good at sorting survey responses and folding proteins, so we keep those. That’s not a middle ground I hear people talk about much. Like any technology: We try it in a bunch of ways, we find what it's great at, and then we use it for the things it's great at.

Beatrice: Does it matter how we imagine futures? Dystopian visions are certainly the more widespread and available ones right now.

Ada: What’s harmful is when our visions of the future are homogeneous and unquestioned, whichever direction they point. That’s true of Golden Age science fiction unquestioningly assuming an inexorable historical arc toward a Western-dominated, imperial future for humanity. And it’s true of the glut of dystopian and apocalyptic fiction that started just after the year 2000, a decade and a half before things started to feel bad in real life to most people. 

When I first came to the University of Chicago, I had lunch with a colleague in creative writing and asked what fraction of the students were writing science fiction or fantasy. He said: We only have two students in the major who aren't writing dystopia. That’s how saturated the consensus was. And when you get consensus about the future people stop thinking about the fact that we have agency. A lot of people say: We know the future is going to suck, so why bother doing anything? But despair is how we lose.

Ken: Beyond homogeneity of vision being bad, homogeneity of interpretation is bad too. Interesting portrayals of the future function like mythologies, or dreams. They’re open to interpretation. Writing science fiction as a philosophical blueprint for how to construct society is a valid mode, but it's not the only one. And a lot of the most powerful science fiction is mythological, a thought experiment that opens up possibilities rather than a literal prediction. If I were to portray the Etsy arts-and-crafts future Ada described, I would not write it as everyone happy, like lotus-eaters. I’d try to imagine why the vision is compelling for some people and why some would find it dystopic, and find the human story in it. By portraying that emotional dream, readers can find their own path through it and figure out whether the vision speaks to them. 

There’s a reason so many wonderful works of imaginative literature can be read multiple ways. Take Susanna Clarke’s Piranesi: Some readers finish it finding that world incredibly beautiful, and some don’t, but you will have an emotional reaction you then have to explain to yourself. That’s where the power of this kind of imagination comes from.

Ada: That’s one of the most valuable things about not just imagining a future but having a storyteller actually sit down and tell a story in it: You explore the premise in its messiness, working and not working, for the people it works for and the people it doesn’t. I hang out with a lot of people involved in fixing democracy, and they'll have a great idea, say sortition, choosing officials by lottery instead of voting, and sortition will work perfectly. 

The reason they think so is that their sortition exists in a perfect abstract world, and they’re comparing it to American democracy, which has existed for 250 years, during which everyone has had an incentive to hack it. Of course perfect sortition looks better than imperfect voting. What a science fiction writer, or a historian, can do is show you sortition in action: working, being corrupted, working for these people and not those. Then you’re comparing the imaginary-and-flawed to the real-and-flawed, which puts them on par. That’s why science fiction is so good at helping us spot the weaknesses, the people left behind, and the "we could use this badly, guys" problems of innovations before we get to them. We don't just depict the world, we tell a story in it, with conflicts and failures.

Beatrice: Do you each have a favorite element from your own worldbuilding that you'd like to see in our toolbox in terms of creating a more flourishing world?

Ken: Imagine a future in which we run our AI on our own hardware, our own little boxes, open source, open weights, with total control over what data and what access to our lives we give it. In that future, your personal AI becomes part of your extended cognition. In my novel, the main character, Julia, has an AI like this and uses it essentially as a mirror. She uses it to examine her own actions, figure out where she went wrong, what she’s doing well, and how to amplify her abilities. In the same way that writing allowed humans to reflect on our own cognition and think deeper thoughts, a personal AI would be cognitively transformative.

Ada: Besides better transit, so we can all see our friends more often, in Terra Ignota I imagine work divided into two categories. There are "voker" jobs, from vocateur, vocation: a job you do because you feel called to it, because you love it, a passion project. And there are the jobs somebody has to do. Somebody has to repair the robots that repair the other robots when they get into a state that requires human ingenuity. In our society we don't differentiate these, even though they operate completely differently and we reward them completely differently. In this imagined future, your identity is not your job unless you love your job. Notably, this requires no technological change at all. It's just a change in how we think about work.

Beatrice: Last question. Thinking big for what is possible in the future: What's your existential hope vision, the thing we should be aiming toward?

Ada: Poverty and educational access. We have all these studies confirming that when people are desperate, worried every day about where tomorrow's meal is coming from, it takes a huge cognitive load. The vast majority of computing power on Earth is human brains, and a huge amount of it is dedicated to worrying about where lunch is coming from tomorrow. If you want to free up computing power, solve poverty. 

Ken: I like that answer a lot, but I’ll go in a slightly different direction. In a lot of far-future science fiction, humans or aliens build some amazing computing machine that harnesses a black hole, and somehow all that computing power is devoted to militaristic ends: how to kill, how to defend, how to conquer the universe so no one else can take it over. I find that incredibly sad, and not realistic. Look at the computing power on the planet today, human brains: I don't think the vast majority of it is devoted to destroying our enemies. A huge amount of our energy goes to questions like: How do I become a better person? How do I love my friend more? How do I get closer to ultimate reality, and why do we exist? We devote enormous computing power to contemplative questions, to matters beyond materialism. I’d love to see us engage more with that side of ourselves: exploring consciousness, engaging with what Hinduism has to say about what consciousness is, with Daoism, with esoteric philosophies, with meditation. Things that have nothing to do with the day-to-day struggle to survive, and everything to do with ultimate reality. I don’t want people to forget that’s part of human flourishing too.