Link Roundup October 2023

OA Week
International Open Access Week will take place October 23 to 29, 2023. UW-Madison Libraries will be hosting a panel on October 23 1-2 p.m. exploring this theme of “Community over Commercialization.” For more information and other ideas on how to celebrate locally, visit the  Libraries Open Access Week page

Open-source generative AI models
Political and data scientist Arthur Spirling makes the case in Nature for the need for researchers to develop their own transparent open-source large language models to avoid reliance on proprietary models as a way to continue to make progress on research ethics and reproducibility. 

OVCRGE launches initiative to foster new large grant proposals
The new initiative provides principal investigators (PIs) with seed money for preparing large research grant proposals of at least $10 million over 5 years. Up to $150,000 may be requested through the initiative to hire a professional grant writer/consultant/coordinator or to compensate a faculty or staff member to devote a portion of their time to preparing proposals. 

Machine learning analysis of citations highlights importance of federal funding
UW-Madison’s B. Ian Hutchins discusses using a machine learning algorithm to explore citation information from a pool of 38,000 scholarly papers and how that revealed that while biomedical research funded by the NIH makes up 10% of published scientific literature, it accounts for 30% of substantive research cited by others in the field or supporting further discoveries. 

Generative AI @ UW–‍Madison: use & policies
DoIT has put together a resource of all of the policies relevant to the use of generative AI. It reaffirms the need for faculty, staff, and students to abide by legal and ethical obligations to protect institutional data. Data entered into generative AI tools becomes part of their training data which it may then share with others. No internal, sensitive, or restricted data should be entered into generative AI tools and no AI tools meet the university’s security, privacy, and compliance standards for anything other than public data. Visit their page for more information on responsible handling of data and generative AI.

US Court rules that AI generated art cannot receive copyright
A US District judge affirmed the Copyright Office’s rejection of an application filed by computer scientist Stephen Thaler on behalf of his DABU system, ruling that only works with human authors can receive copyright.