The problem of historical objectivity with AI

FBK Aula Piccola
Fondazione Bruno Kessler - Polo delle Scienze Umane e sociali
AI technologies like ChatGPT have problems with “hallucinations” – the unwanted production of errors. Understanding what these errors are and where they come from is crucial to making sense of what AI can and cannot do. This talk will lay out an agenda for historical research about AI, including adjustments to data structure and research in the information sciences that can make AI more effective for a variety of human use cases, including political analysis, medical and legal applications, and the automatic creation of safety guardrails.
JO GULDI | Emory University
Seminar cycle: “AI and history“
Organization:
Andrea Pojer (Università di Trento – FBK-ISIG)
Massimo Rospocher (FBK-ISIG)
Sandra Toffolo (FBK-ISIG)
The event will be held in English.
The presentation will take place in presence in FBK’s Aula Piccola, subject to availability, and online.
Registration is mandatory by 12 June 2025 at noon.
NB: Please note the change in date: this seminar will take place on Friday 13 June (not Tuesday 17 June as announced earlier).
***
Image: AdobeStock_1037404632
Relatori
-
Jo Guldi - SpeakerEmory UniversityProfessor Jo Guldi is a historian and data scientist at Emory University in Atlanta, Georgia. Her prize-winning 2022 book, 'The Long Land War', narrated the story of the global contestation of ownership rights and the assertion of international rights of occupancy from 1881 to 1978, in a story where Ireland and Scotland took the major role. In her 2023 book, 'The Dangerous Art of Text Mining', she engaged modern strategies of research with machine learning and AI, warning against the haphazard deployment of research techniques by information scientists insufficiently schooled in the methods of historical reasoning. Guldi teaches at the Department of Quantitative Theory and Methods at Emory University, with affiliations in History, Law, Environmental Studies, and Computer Science. Her research has been covered in the FT, TLS, Wall Street Journal, Nature, PNAS, and Atlantic Monthly. Her current research focuses on text mining the documentation of climate change.
Registrazione
La registrazione a questo evento è richiesta.
RegistratiContatti
Organizzatori
