Forgot your password?
typodupeerror
AI Science

AI Boosts Research Careers But Flattens Scientific Discovery (ieee.org) 64

Ancient Slashdot reader erice shares the findings from a recent study showing that while AI helped researchers publish more often and boosted their careers, the resulting papers were, on average, less useful. "You have this conflict between individual incentives and science as a whole," says James Evans, a sociologist at the University of Chicago who led the study. From a recent IEEE Spectrum article: To quantify the effect, Evans and collaborators from the Beijing National Research Center for Information Science and Technology trained a natural language processing model to identify AI-augmented research across six natural science disciplines. Their dataset included 41.3 million English-language papers published between 1980 and 2025 in biology, chemistry, physics, medicine, materials science, and geology. They excluded fields such as computer science and mathematics that focus on developing AI methods themselves. The researchers traced the careers of individual scientists, examined how their papers accumulated attention, and zoomed out to consider how entire fields clustered or dispersed intellectually over time. They compared roughly 311,000 papers that incorporated AI in some way -- through the use of neural networks or large language models, for example -- with millions of others that did not.

The results revealed a striking trade-off. Scientists who adopt AI gain productivity and visibility: On average, they publish three times as many papers, receive nearly five times as many citations, and become team leaders a year or two earlier than those who do not. But when those papers are mapped in a high-dimensional "knowledge space," AI-heavy research occupies a smaller intellectual footprint, clusters more tightly around popular, data-rich problems, and generates weaker networks of follow-on engagement between studies. The pattern held across decades of AI development, spanning early machine learning, the rise of deep learning, and the current wave of generative AI. "If anything," Evans notes, "it's intensifying." [...] Aside from recent publishing distortions, Evans's analysis suggests that AI is largely automating the most tractable parts of science rather than expanding its frontiers.

This discussion has been archived. No new comments can be posted.

AI Boosts Research Careers But Flattens Scientific Discovery

Comments Filter:
  • by Anonymous Coward
    So, quantity over quality. Who knew?
  • by Mr. Dollar Ton ( 5495648 ) on Friday January 23, 2026 @03:33AM (#65943514)

    Basically, people sift through data trying to find "relationships" without giving too much construction on the actual knowledge behind these.

    So, a lot of "discoveries" of correlations, not a lot of effort to explain the causative links.

    At least in the fields I try to follow.

    Sad, really, but enhanced by the publish-or-perish crap.

    • Not "construction", auto-carrot, "consideration to". Or something similar.

      Especially true for the younger researchers, who are very much stuck on "AI", but not limited to them.

    • by znrt ( 2424692 )

      Sad, really, but enhanced by the publish-or-perish crap.

      probably the other way around, llms enhance publish-or-perish frenzy, and the study is talking particularly about papers, which already was a swamp even before ai. i guess there are many other applications where generators and deep learning shine and may be actually very useful: imaging and diagnosis, discovery of molecular structures, antibiotics, protein folding, genomics, ...

      • It's most likely circular. Publish or perish means more publishing means more AI means more papers published in total means you have to publish still more papers in order to look important means more AI

  • classism and greed are destroying our institutions and our societies, all you selfish irresponsible elites are wrecking everything for everybody

    this is exactly what evil looks like

    • classism and greed are destroying our institutions and our societies, all you selfish irresponsible elites are wrecking everything for everybody

      this is exactly what evil looks like

      Evil and stupidity enhanced with shortsightedness look like twins. I suppose greed plays a role as well, but I feel like the root of what we're seeing sweep over the world right now is stupidity and shortsightedness. Greed is always there. We're just dumb enough as a species at this point in time to let the greedy make all the decisions about the important things, and the rest of us just shrug and go along with it as if there's nothing to be done to stop the greedy from riding roughshod over us, the biosphe

      • by 2TecTom ( 311314 )

        with great power comes great responsibility

        when those with power, and hoarding all our capital, act completely irresponsible, then a collapse is certain

        history has shown us this is inevitable and natural, all systems emerge, grow, decay and then die

      • by 2TecTom ( 311314 )

        classism produces corruption which breeds incompetency, greed is clearly evil, yet we put greedy people on a pedestal

        we are all getting what we deserve for letting evil people rule over us

        people get the governments we deserve and we don't deserve good things because we are an irresponsible and unethical peoples

        look around with honest eyes and tell me it isn't so

  • The Big Expectations (Score:5, Interesting)

    by devslash0 ( 4203435 ) on Friday January 23, 2026 @05:09AM (#65943578)

    The problem with scientists' careers is the expectations to make great discoveries. Science doesn't work like that. You may spend 20 years running experiments and crunching numbers and make a handful of mediocre observations at best. Then, one day, if you're extremely lucky and all the stars align, you may discover something more valuable when sitting on the sofa in the evening doing nothing. Still no guarantees that it's going to be anything big. Maybe 3 out of 10. Perhaps 4.

    Meanwhile, all the young people want to be superstars from day 1 these days. And, of course, AI only enables to spin more bullshit to big themselves up instead of accepting that their career may ultimately amount to nothing at all.

    • That expects those big discoveries. From what I can tell seeing several videos and reading blogs from scientists complaining about how their careers have suddenly become heavily heavily politicized working scientists are just that, working scientists. They slog through every day trying to make something new but they're not doing it because they're expecting some great amazing Discovery but because they are kind of obsessed with a narrow field of study and because it's their job.

      By the time a young pers
    • by 2TecTom ( 311314 )

      Those days are gone and we need to move on. Science without ethics is how we got into this mess and all the unethical overly-affluent 'scientists' are never going to solve our problems. One of which is how corrupt science and academia have become. Sadly, there no room for real science when science is only about making the big bucks and having a prestigious position, clearly what we call science has become corrupted and moribund.

  • I am slightly skeptic of this study, as AI hasn't been with us long enough to actually get the real impact in the academic data. A proper academic research can take 2-3 years to mature, so the reak research that AI might have impacted is not matured and born yet. What we see in the papers currently is just easy fruits of the academic tree that are becoming available due to AI.
    • Let's see if an AI can discover Relativity on its own when trained purely on Shakespeare and Newton. I'm not holding my breath.
  • I'll assume the "current AI" is LLM based. Why would anyone expect it to make new discoveries of intellectual depth. All its doing is rearranging the deck chairs, there's nothing deep about that. It can do it faster than humans, I suppose there's that.

    As someone above noted, most scientists are in trenches mining salt (my paraphrase). To put it in a different light, most science is pushing current theories a bit farther, or figuring out new consequences of current theories; they are not discovering new theo

    • by 2TecTom ( 311314 )

      Honestly, what i see is most science is someone doing make work and publishing bs so they can keep their grant and their title. As long as they are tenured and publish, they can enjoy a prestigious free ride, that is what science is about these days.

  • So what i gather is that a subset of science benefits from AI that does the data analysis. The vast majority doesn't when doing the data analysis. The LLM is not going to perform the enzyme assay or clone the gene, or inject the drug into the mouse. This subset of science clusters more tightly around popular, data-rich problems.
    This is my opinion: once you are some sort of senior researcher who leads a very big consortium that produces lots of data, you use AI to process all that data and that very big s
  • by jd ( 1658 )

    AI cannot discover or innovate, because AI is only capable of looking at pre-existing patterns and the preponderance of pre-existing patterns will always be where people have already done all the real work.

    It is my conjecture that specialists using AI will continue becoming worse and worse at their subjects because that is NOT what AI is actually good at.

    This is not to say I think AI is useless (although it largely is). I would argue that AI can be used by generalists to find interesting patterns between we

  • publish more less useful
  • This really seems to be the recurring pattern with AI and how it is impacting work. It helps your career, it helps income (all that slop brings in the ad dollars), but it isn't producing products or value. We are seeing the flow of money get even more diverged from actual productive economic activity. Granted this has always been a problem at the higher income levels, but the rot seems to be seeping down in a way that, well, we are losing our sheep buffer.
  • This is the problem
    Science, as a concept, is great
    As a business, measured by the number of papers published, it sucks
    Instead of rewarding people based on the number of papers published, they should be rewarded by the quality of papers published, even if there are very few

  • Thing that can only regurgitate what it knows can't actually discover anything. Big shock. But I guess it does work well for another shitty agenda-driven meta-analysis.

The finest eloquence is that which gets things done.

Working...