Where do ideas come from?
We’re used to the idea that ideas come from individuals—and in the age of AI, the insistence on human individuality, creativity, and originality has become particularly urgent.
Here’s the problem: I don’t necessarily buy it. I don’t necessarily buy that ideas come from individuals, and I don’t necessarily buy that originality is even a real thing.
Part of this skepticism has to do with my training as a historian. Of course, we as humans are deeply influenced by the culture in which we live, as is, inevitably, our art. Our thoughts, too, draw from what’s already there. And it’s a good thing they do, for this is what makes us legible to one another.
As Helen Lewis wrote recently in The Genius Myth: A Curious History of a Dangerous Idea, “We find it intuitively easy to understand human-sized stories, where someone does something, whereas vague wafts of social change driven by multiple factors might get academics excited . . . but tend to leave everyone else bored to tears.”
Well, I’m one of those academics who gets super excited about vague wafts of social change.
People hate to be told this. They want to believe that their ideas are theirs, and theirs along. But there’s a lot of good research to support the argument that ideas emerge in a specific cultural context.
The concept of the adjacent possible, popularized by Steven Johnson in his book Where Good Ideas Come From (2010), holds that once certain social, economic, and technological conditions have been met, new discoveries/inventions are largely inevitable. If they can be discovered/invented, someone is going to discover/invent them. The prevalence of simultaneous discoveries in the history of ideas and innovations suggests as much. For example, during the Cold War, Soviet and Western mathematicians separately reached the same breakthroughs without any access to each other’s work.
Another part of my resistance to the idea that ideas come from specific individuals has to do with righteous feminist anger. As Scottish writer Thomas Carlyle insisted in his book On Heroes, Hero-Worship, and the Heroic in History (1840), “The history of the world is but the biography of Great Men.” This isn’t just a case of era-specific language. Rather, both materially and philosophically, the idea of originality is bound up with masculinity.
But if this is true—if the concept of human originality is not all it’s cracked up to be—why do we need humans at all? Is there another argument for the importance of human contribution that doesn’t hinge on originality?
Here’s where AI comes in: So much of the debate over this new technology hinges on whether it can produce anything original—and so many of the arguments against it center on originality as something fundamentally human. But if we revert to “originality” as the defining human characteristic, are we in effect rejecting the cultural/environmental in favor of the individual—and in the process reifying Great Man theory?
Ultimately I have a hard time accepting either possibility: that ideas come from individuals, or that they do not.
About a year ago, I realized that I could tell immediately when something had been written with AI.
It’s not the em dashes that give it away. We writers love our em dashes.
Rather, it has something to do with the cadence, the line breaks. There are just too many of them. I think they’re supposed to make each line feel important and profound, but really they just make everything sound the same.
Herein lies the problem.
I’m not entirely opposed to AI. If I need a new backpack or ideas for five-ingredient meals using only items from Trader Joe’s, I’d rather ask ChatGPT than wade through eight thousand listicles. In these cases, I want the consensus, the collective wisdom of all humanity. I want thoughts that have already been formed (though the more I learn about the environmental degradation it causes, the less justified even these frivolous uses seem).
AI, as we know, is (usually) really good at this kind of synthesis—taking pieces from everything already out there and assembling them into usable knowledge.
This strength, however, is also its weakness. The predictive nature of large language models (LLMs) means that AI can only draw from what’s already out there. LLMs tend toward conformity, to the most common patterns.
But why does the debate over so frequently turn on whether it can create art? I myself spent months trying to write an essay about how writing needs to get weirder, to more fully reflect its individual creator. We would confound the algorithm by making the confounding connections that only humans can make, by drawing on the weird combination of obsessions that fuel us and our writing.
And then I remembered that I’m the same person who spent half of grad school railing against auteur theory, for all the same reasons I rail against Great Man theory.
This debate has existed since humans first envisioned thinking machines. As computer scientist Ada Lovelace wrote of a colleague’s invention in 1842, “the Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform” (whatever it has been programmed to do) and only that.
A century later, in 1949–50, professor of neurosurgery Geoffrey Jefferson and mathematician and computer scientist Alan Turing engaged in a version of this debate.
Jefferson’s point in “The Mind of Mechanical Man” echoes Lovelace’s. A machine, he believed, could only do what it was programmed to do—a limitation that lay most importantly in “the machines’ lack of opinions, or creative thinking in verbal concepts”:
It is not enough, therefore, to build a machine that could use words (if that were possible), it would have to be able to create concepts and find for itself suitable words in which to express additions to knowledge that it brought about. Otherwise it would be no more than a cleverer parrot, an improvement on the typewriting monkeys which would accidentally in the course of centuries write Hamlet. . . . Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain.
(I think here of Data in Star Trek: The Next Generation, who writes a poem to his cat. The consensus aboard the Enterprise is that “Ode to Spot” is not a very good poem. Nor is Data a particularly good Shakespearean actor, painter, or violinist. Or rather, his offerings are deemed technically proficient but emotionally bereft, clearly not the result of individual thought and emotion.)
Turing, for his part, responded not so much by arguing that machines could be creative (though he believed this was possible), but by challenging our assumptions about human thought and creativity.
Could we ever really know if a machine was thinking? Perhaps not. For some, “the only way by which one could be sure that a machine thinks is to be the machine and to feel oneself thinking.” But adherents of this view must concede that, similarly, “the only way to know that a man thinks is to be that particular man.” How do I know you’re not a zombie? I don’t.
Responding to Lady Lovelace’s objection, Turing added: “Who can be certain that ‘original work’ that he has done was not simply the growth of the seed planted in him by teaching, or the effect of following well-known general principles.”
So maybe that’s the fear behind all of handwringing over AI and creativity: not so much a defense of human values and capabilities . . . but a gnawing doubt that these capabilities were never quite what we thought they were.