Most researchers believe the rise of artificial intelligence (AI) is harming academic publishing, says a new study which warns a surge of “poor-quality or AI-generated content” is threatening to overwhelm peer review processes.
Drawing on a survey of more than 3,000 researchers globally the Publishing Futures report, released on 16 October by Cambridge University Press, found 53 per cent of respondents believe AI would negatively impact the research publishing system while 18 per cent feel it would improve academic publishing.
In follow-up workshops interviewees explained they feared AI would lead to more research trained on poor-quality datasets, more papers containing AI “hallucinations” and more frequent research integrity concerns related to credit for intellectual property used by AI.
That last concern was particularly prevalent, with 33 per cent of respondents stating they are worried that AI would make it harder for researchers to gain due credit for their work. That was roughly the same as the proportion (32 per cent) who feel AI could help ensure researchers are fully credited for their research.
The technology’s potential to help academics churn out low-quality post-publication commentary was also flagged by researchers – a theme picked up by the publisher’s managing director Mandy Hill, who, in the foreword to the report, warned more generally about the impact of AI-assisted papers.
“Important work risks being lost or drowned out by a surge of low-quality or AI-generated content,” she said, adding: “Financial models are becoming unsustainable, and the sheer volume of publications threatens to overwhelm the ecosystem.”
Wider concerns were also raised about how the growth of open access publishing over the past decade had led to a “system under strain from publishing volumes”, with the report noting how “approximately 897,000 more indexed articles in 2022 than in 2016 – an average growth rate of around 5.6 per cent”, referencing a??paper from 2024.
Some 81 per cent of respondents agree an increase in the number of research outputs published has put the peer review system under pressure, with 50 per cent stating poor-quality peer review is linked to the rise in submitted papers.
“This high rate of growth, which includes an increase in poor-quality or AI-generated content and the rise of papermills, threatens the very integrity and credibility of research,” states the report.
“We fundamentally believe that publishing less – but better – is essential for the health of the entire research system worldwide,” it continues, recommending that “not all research outputs need to be published as traditional articles and much research now would benefit from alternative options being recognised”.
To help achieve this, universities should “de-link academic success from publishing volume and instead recognise quality over quantity”, suggests the report, which calls for “more holistic approaches to evaluating academic performance and contribution”.
“We encourage publishers to develop new metrics for research outputs and to support new approaches to building a new culture and infrastructure for research assessment,” suggesting institutions should consider “broadening recognition beyond journal articles (including data, software, peer review, teaching, and community leadership), piloting narrative CVs, and reducing reliance on journal-based metrics in hiring and promotion”.
Commenting on the current push to put “everything into article format” and publish as a paper, Hill said publishing in “high-quality journals” should be accompanied by “lower-cost, scalable alternative publishing pathways [that] need to be seen as credible, attractive options for academics too.
“To achieve this will require understanding and addressing what is really incentivising authors’ publishing habits, including the reward and recognition models of universities and funders,” she said.
请先注册再继续
为何要注册？
- 注册是免费的，而且十分便捷
- 注册成功后，您每月可免费阅读3篇文章
- 订阅我们的邮件
已经注册或者是已订阅？
