The tip and the iceberg: deep learning and embodiment (CVPR 2024 keynote).

meclab.org
How can artificial intelligence create robots from technological and/or biological components?
Why might embodied AI be safer than non-embodied AI?
What will happen when the planet is increasingly populated with AI-designed organisms? What will happen when it becomes easier for malign actors to create them?
I attempt to address these questions for the Computer Vision and Pattern Recognition (CVPR) community at their annual conference in Seattle.