Wolfgang Ernst, University of Zurich: Protecting Research Integrity -- A Practitioner’s View
For two decades, natural sciences have been plagued by a failure of replication crisis. In the humanities, plagiarism cases paint an equally dire picture. A vast amount of research, often dating back many decades, calls for a renewed critical review. While the advent of AI technologies facilitates such reviews, one is afraid that it will also bring new waves of research . Alongside scientists' associations, journals, and academies, universities investigate alleged breaches of standards of research conduct. The presentation will explain the regulation and procedure that control investigations carried out for the University of Zurich. Problems, failures, and successes of university-based investigations will be discussed. Alternative mechanisms and measures of prevention will be looked at. The speaker, who has contributed to the University’s efforts and will continue to do so, expresses his personal views.
Matt Spick, University of Surrey: When every hypothesis is published: AI, the exploitation of Open Science Assets, and false discoveries (online)
Open science has given us a shared public good, in the form of publications and public datasets, free to access and reuse by anyone. But AI can now generate, test and write up hypotheses against those assets faster than our community can evaluate them. This talk looks at what happens to open repositories when they become inputs to industrial-scale hypothesis generation, and whether this is likely to become a systemic problem. We will also provide case studies of outputs that can be generated in under an hour, and examples of the impact on the literature (including the de facto banning of some open science resources altogether). It also asks what data stewards, funders and publishers can realistically do about it.
Part of the UZH-i-Ringvorlesung: Fake Science – Isolated Cases or a Systemic Problem?
Veranstaltungsort
Zusätzliche Informationen
Kosten
