Lezing
SAILS Lunch Time Seminar: Derya Soydaner
- Datum
- maandag 13 april 2026
- Tijd
- Locatie
- Online only
What Humans See, What Machines Learn: Yet Not Like Us
Abstract
We have built machines that see. We just haven't built machines that perceive. Human perception functions differently: you see a face on a wall socket, sense the uncanny in an AI-generated image, or hear a melody and see it unfold. No machine does. Not really.
Meanwhile, AI generates art, describes paintings, and takes on creative tasks. Yet the gap between human and machine perception is deeper than it looks. Deep neural networks, including the latest vision-language models, can be fooled in ways that would never fool us. Drawing on vision science, art, and computer vision, this talk explores where human and machine vision align, where they diverge, and what this gap reveals about the current limits of AI. We take on human perception and art to challenge and inspire AI, as domains rich in ambiguity, creativity, and complexity that expose what standard benchmarks often miss.
Derya Soydaner is an Assistant Professor in the Leiden Institute of Advanced Computer Science at Leiden University. With a background in Statistics and expertise in Machine Learning, combined with experience in experimental psychology, she is a multidisciplinary scientist working at the intersection of human perception and neural networks. Her research also extends to art, aesthetics, and AI.
She focuses on the development and understanding of deep neural networks, particularly in image processing, generative AI, and explainable AI. Her work is highly interdisciplinary, bridging fields such as computer science, physics, psychology, and art history.
Join us!
The SAILS Lunch Time Seminar is an online event, but it is not publicly accessible in real-time. Please click the the link below to register to our mailinglist and receive participation links for our Lunch Time Seminars.
Sign up