Monica Storss is the creator of The Oracle, a language model built on GPT-2 architecture and trained on Storss’ own poetry corpus. The poem cycle co-created with her algorithmic twin is titled The Oracles. Excerpts, process notes from The Oracles, and an interview about the process were published in Ensemble Park.
Monica’s Machine Learning/Artificial Intelligence work is with large language models, small language models, OCR, and closed models in creative praxis, research, and archival contexts.
Her primary sites of AI/ML work are in creative praxis and the Digital Humanities.
Her recent Machine Learning projects include Organizational Learning and AI Policy, Ethical AI, AI for creatives, AI pedagogy, and creating Hewlett Packard’s Machine Learning/AI learning programs for all 55,000 employees globally.
She also helped train the OCR for The Boston Globe Photo Archive Project at Northeastern University in 2024.
About the project:
MACHINE LEARNING FOR LARGE-SCALE JOURNALISM COLLECTIONS: The development of a computationally-assisted cataloging workflow for curators and archives to provide enhanced access to large-scale newspaper photograph collections. The Boston Globe’s historic photo morgue, under the stewardship of the Northeastern University Libraries, will serve as the test case for this effort.
Research Assistant and affiliate Creative Technologist, trained OCR
AI Computer Vision and LLM project on the complete photo archive and morgue of The Boston Globe.
HAA-296412-24 Digital Humanities Advancement Grants
Digital Humanities
Grant period: 2/1/2024 – 1/31/2026
MACHINE LEARNING FOR LARGE-SCALE JOURNALISM COLLECTIONS