News: Academic Publishing Weekly

Harvard's public AI training dataset, combatting fake science, and the year's best book covers

Posted on December 20, 2024 in News

Harvard Unveils AI Training Dataset and Safeguarding Copyright in the Age of AI

Harvard University has released a publicly accessible “high-quality dataset of nearly 1 million public-domain books” for the purpose of training AI. The dataset was developed by Harvard’s Institutional Data Initiative with funding from Microsoft and OpenAI, and aims to provide wider “access to the sort of highly-refined and curated content repositories that normally only established tech giants have the resources to assemble.” Wired noted that the Institutional Data Initiative is also pursuing partnerships with entities including the Boston Public Library. Next up, the Creative Rights in AI Coalition, composed of creatives including publishers and authors, released a statement and several principles to guide decisions on generative AI. The principles center on transparency, innovation, and copyright protections, and the coalition’s statement encourages “government to adopt the principles as a framework for developing AI policy.” The coalition also cited a poll from Reset Tech and YouGov, which reveals public support for transparent training data and paying creators royalties.

---

---

Accessibility Concerns in Academic Papers and the Aftermath of eLife Losing Its Impact Factor

Nature’s Dalmeet Singh Chawla reports that academic papers continue to lag behind when it comes to accessibility. In particular, Chawla referenced a study of roughly 20,000 scholarly paper PDFs published from 2014 through 2023, finding that, “Only around 3% of the analysed studies met all six [accessibility] criteria…and just under 75% met none of the criteria at all.” Co-author of the study Anukriti Kumar observed a drop-off in 2019, pointing to the impact of preprints and quick publication turnaround times during the COVID-19 pandemic. Chawla also covered current accessibility efforts from publishers, with a Springer Nature spokesperson noting that, “Improving PDF accessibility requires changes to culture, systems and processes that can be challenging for publishers and authors to achieve.” Turning to the open access journal eLife, Diana Kwon of Nature found that submissions have declined since the journal lost its Impact Factor last month, and, in response, “the editorial team is collecting community feedback and watching submissions carefully.” Kwon also highlighted comments from researchers on eLife’s controversial publication model, the value of papers being indexed in the Web of Science, and whether journals will begin steering away from nontraditional business models.

---

CORE Gains Funding for Open Research and a Call for Tackling Fake Science

The Open University’s not-for-profit open access service, CORE, has received five years of funding from Microsoft to support open data and open research. The funding will cover the costs of new technology, indexing processes, and “research on ethical and responsible ways of using academic content in the age of AI.” Burton Davis, Microsoft’s Vice President and Deputy General Counsel, commented, “We are proud to support CORE’s important work of making more scientific scholarly content openly available, helping foster innovation, including the responsible development of AI.” In other news, the Forensic Scientometrics (FoSci) Paris Declaration launched out of a research integrity event at Institut Universitaire de France (IUF). The Declaration advocates for the use of FoSci to combat fake science. FoSci, or “data-driven forensics,” “combines forensic investigation and scientometrics” and can detect “patterns that uphold or threaten the integrity of science itself.” The Declaration will work to increase communication with policymakers, researchers, and publishers, support training, and promote the “reform of institutions involved in scientific research based on the group’s findings.”

---

---

New African Journals Initiative and Project Muse OA Updates

Pluto Journals and African Books Collective are collaborating on a new pilot program to bolster the visibility of open access social science and humanities journals published in Africa. The African Journals Initiative is backed by ScienceOpen and JSTOR, and the three-year initiative will “provide a range of services for journals, promoting their visibility and discoverability, indexing, user-friendly submission systems, annual detailed impact reports, and much more.” The program will also organize “financial support from academic libraries and consortia.” Next, Project MUSE announced that The Dial, published by the Ralph Waldo Emerson Society, and Nuevos Horizontes, sponsored by the Mellon Foundation and published by University of Puerto Rico-Mayagüez, have been added to the platform. Getty Publications’ Getty Research Journal is also fully open access on Project MUSE after issues 1-18 previously being available only to its subscribers.

---

Ingram Partners with Brepols on Distribution and NYT’s Best Book Covers of the Year

Starting in February, Ingram Publisher Services UK will handle distribution for the academic publisher Brepols Publishers in the United Kingdom. Brepols, founded in 1796, is based in Belgium and its publications center on the humanities. Matthew Dickie-Cross, senior manager at Ingram Publisher Services UK, remarked, “Brepols has a long and outstanding reputation for publishing high-quality scholarship, and we are proud to partner with them to make their publications even more widely available.” As for booklists, Matt Dorfman, art director of the New York Times Book Review, spotlighted several captivating book jackets as part of NYT’s annual best book covers of the year round-up. The covers play with unique color schemes, fonts, and photography, including something for every reader.

---

Related Posts