We came to synthetic data with a concern for disappearing edge cases. We have gotten muddled in Intersectional Hallucinations. Using sensitivities informed by intersectionality theory and machine learning techniques, we are developing a vocabulary and toolkit do identify and measure intersectional fidelities and hallucinations. This work triggers questions about the uses of synthetic structured data (when are hallucinations good? When are they a problem?) and guardrails to ensure its reliability.
Speakers As a Deputy Head of Department and Professor at Linköping University, Ericka Johnson's research explores how the world becomes data. With a background in Science & Technology Studies and medical humanities, she's looking at the nexus of ontologies, epistemologies and AI. What happens when the data that represents the world meets AI? As a Principal Research Engineer at Linköping University, Saghi Hajisharif's research interests are in visual machine learning, HDR imaging, photo-realistic rendering, compressive imaging, and light fields.
Please note. The subtitle was primarily generated by AI in collaboration with human review and may contain some errors.