Return of the solo author: The changing division of labor in science in the age of generative A
This study analyzes over 300 million scientific works to reveal that the decades-long decline in solo authorship halted and partially reversed following the release of ChatGPT, particularly in fields where AI can substitute for human coauthors, indicating a reconfiguration of cognitive labor rather than a simple change in team size.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine science as a massive, bustling construction site. For decades, the trend was clear: nobody built anything alone anymore. Just like a pin factory where one person can't make a whole pin by themselves, scientists realized they needed huge teams to handle the complex, specialized work of modern research. The average paper became a group project, with more and more names on the cover.
But then, in late 2022, a new tool arrived: Generative AI (specifically, the public release of ChatGPT). This didn't just make the construction site bigger; it changed the blueprint entirely.
The Great Solo Turnaround
The paper's main finding is a bit of a plot twist. For years, the number of scientists working alone (solo authors) was slowly disappearing, like a rare animal heading toward extinction. But right around the time AI tools became available to everyone, that decline hit the brakes and actually started to reverse.
Think of it like a long, slow slide down a hill that suddenly flattens out and starts to curve back up. This isn't just a tiny bump; it happened across 26 different scientific fields, involving over 300 million works. The authors measured this by looking at the "left tail" of the data—the small group of papers with only one name. While the average team size is still huge, the number of people working alone has started to grow again.
Why the Change? It's Not a New Crowd
You might guess that this is because a bunch of new, young scientists just started writing alone. The paper explicitly rules that out. They checked the "roster" of researchers and found that the mix of people didn't change. It's not that new solo artists entered the stage; it's that the old artists, who had been working in teams for years, decided to try working alone again.
Even scientists who had never published a solo paper before, and those who had been collaborating for decades, started going solo. The paper suggests this happens because AI can now do the "grunt work" that used to require a human partner. If you need someone to write a draft, clean up data, or run a statistical test, you used to need a coauthor. Now, you can ask an AI to do it.
The "Substitution" vs. "Acceleration" Debate
Before this study, there were two guesses about what AI would do to science:
- The Acceleration View: AI makes research so easy that teams will get even bigger, with more people joining in.
- The Substitution View: AI replaces the need for human helpers, letting one person do the work of a whole team.
The paper doesn't say one view is totally wrong, but it shows they are happening in different places. The "Acceleration" idea might still be true for the biggest, most complex projects (the upper tail of the data). But for the solo authors, the "Substitution" view is the one playing out. The AI is stepping in to take the place of the human coauthor.
Where It Happens (and Where It Doesn't)
This solo comeback isn't happening everywhere equally. It's strongest in fields where the work is mostly mental or digital, like Computer Science, Math, and Economics. In these fields, the tasks AI can do (coding, writing, analyzing) are exactly what coauthors used to do.
However, in fields that rely on physical labs and giant machines, like Chemistry and Physics, the trend didn't change. You can't ask an AI to hold a test tube or operate a massive particle accelerator. The paper notes that in these fields, the decline in solo authorship just kept going flat or continued to drop. This proves that the change is about what the AI can replace, not just a general mood swing in science.
What Does the Solo Work Look Like?
When these scientists go solo, they aren't suddenly writing about totally new topics. They are sticking to the same general area they always worked in, but the content of their solo papers shifts. The paper found that these solo papers are moving toward "computational work." They are narrower in scope, focusing on tasks that an AI can help with, rather than broad, exploratory research.
It's like a chef who used to have a sous-chef to chop vegetables and prep ingredients. Now, with a robot that does the chopping, the chef is back in the kitchen alone, making the same dishes but focusing more on the cooking and less on the prep. The chef isn't suddenly deciding to bake cakes instead of cooking dinner; they are just doing the prep work themselves (or with a robot) and calling the dish their own.
The Bottom Line
The paper suggests that we are seeing a reconfiguration of how science gets done, not necessarily a revolution in team size. The "solo author" is no longer just a lone genius working in a vacuum; they are a researcher using AI to fill the gap left by a missing human partner.
The authors are careful to say this is a measured observation, not a guaranteed future. They note that we don't know if this trend will last forever or if the quality of these solo papers is the same as the old team papers. But the data is clear: the decades-long slide away from solo science has paused, and in many fields, it has started to climb back up. The "solo tail" of the distribution is getting thicker, proving that for some parts of science, the AI is taking over the role of the coauthor.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.