The early weeks of the Trump administration have posed multiple threats to academic researchers. Not only was federal research funding temporarily halted, but also the National Science Foundation and National Institutes of Health were ordered to delay review of applications for new funding.
Universities in the United States spend $97.8 billion on research, of which more than 55 percent comes from the federal government. Primary areas of funding include health ($28.9 billion), biology and biomedicine ($18.2 billion), and engineering ($15.6 billion). The University of Washington spends $1.37 billion of federal research money each year.
Funding freezes have been temporarily lifted, and some federal grant applications are again under review. As the executive and judicial branches of federal government remain locked in battle over federal funding of research, universities scramble to counter ongoing threats.
The University of Washington Office of Research has prepared guidance on responding to orders to remove “gender ideology activities” and “diversity, equity, and inclusion” principles from grant-funded research. UW researchers are urged to respond to those requests in a “timely manner.”
Pressures from the Trump administration threaten an already beleaguered research enterprise that historically has thrived on doing basic research with minimal influence from partisan politics and industry profit-seeking.
To ensure integrity, academic research depends on peer review. Researchers propose projects. Panels of experts review proposals and often require modifications before funding is granted. Prior to publication, research reports are reviewed by peers who are knowledgeable about the subject but who were not involved in any earlier stage of the research and do not have commercial or personal interest in findings.
Research funding supports the gathering and analysis of data but traditionally did not fund the publication process. Academic journals typically do not directly compensate the authors, editors or peer reviewers. Nevertheless, academic publishing has become big business.
The academic publishing market is estimated at $25 billion annually. The major publishers of academic journals generate profits of more than 35 percent annually, primarily by selling journal subscriptions to academic libraries. Authors traditionally assign their copyright to a publisher as a condition of publication but receive no royalties on journal sales.
Historically, publishers’ primary costs have included typesetting, proofreading, printing, marketing and distribution. The Internet reduced costs to publishers who no longer had to typeset or print manuscripts. But cost to libraries did not go down. Between 1996 and 1999 (the years when internet access to academic research exploded), the cost of traditional academic journals increased by 52 percent. Journals now cost 250 percent more than in the 1980s.
In the early 2000s, researchers and libraries began to push back against the traditional publishing model by developing the open access (OA) model in which content is free and open to anyone with Internet access. Some researchers independently archive their work with their libraries or professional associations. Others publish in open access journals in which authors, their institutions, or their funding organizations pay Article Publishing Charges (APC) that range from a few hundred dollars to several thousand.
About 50 percent of academic publications are now published in open access journals. Open access publishing is much more common in science, technology, and engineering disciplines than in the humanities and social sciences. STEM fields receive the largest proportion of federal funding for research. Most federal funding agencies now mandate open access publishing of research findings and include a budget for APC fees in grant awards.
In a written statement, Cora Murata, director of collections analysis and strategy at the University of Washington Library, noted: “Any reduction of federal research funding would result in a reduction in the actual research being done, a reduction in available funds to publish articles in OA journals, and a reduction in deposits into federal open access repositories.”
Reductions in academic publishing could also result in less in the way of “training materials” for artificial intelligence. Generative AI models access huge amounts of data that are ingested, digested, and regurgitated in new form. That data can come from sources ranging from personal blogs to research publications. Content that has been screened by peer reviewers and editors is in high demand.
The New York Times has sued OpenAI and its partner, Microsoft, for copyright infringement because they used published articles for training AI learning models. The tech giants have denied the claims and the case is making its way through the courts.
The copyright in most news articles is held by news organizations because journalists wrote the articles as a “work for hire.” If news could be “sold” for AI training, the publishers, not the authors, would benefit.
Academic publishers have begun to cut multi-million dollar deals to sell their content for AI training. But some within the academic community argue that the academic research used should be free to everyone – even AI bots.
In a statement, Elizabeth Bedford, scholarly publishing outreach librarian at the University of Washington, wrote that some UW researchers have expressed concern “about the specter of OA articles being used in in training data.” These concerns are less common among faculty whose disciplines receive more grant funding that covers the up-front publishing costs.
Cuts to federal funding for research could result in less overall scholarly output and less open access publishing. AI could become more influenced by personal, commercial, and political entities with fewer “reality checks” from the academic community. Humanity could lose access to life-altering knowledge.
Author’s note: I have published in journals using both the open access and traditional journal publishing models. I used four generative AI products for researching this article: ChatGPT by OpenAI, Claude AI by Anthropic, Chron by the Chronicle of Higher Education, and Google Notebook. I did not directly copy any material generated by AI.
