🔊 Link to Spotify (Click)

Before large-scale models, before foundation models, before AI became mainstream, there was a more fundamental problem:

there was no data.

In the second episode of Inside ALLSIDES, we sat down with Prof. Pietro Perona (Caltech, Amazon Fellow) to explore how datasets shaped the evolution of computer vision and why data remains the key bottleneck today.

Watch the Full Conversation

In this episode of Inside ALLSIDES, Pietro Perona shares insights on:

How Caltech 101, created with Fei-Fei Li, helped define visual recognition
Why early computer vision lacked not just solutions—but even the right questions
How datasets like ImageNet and COCO reshaped entire research directions
Why AI progress today is increasingly limited by data, not models
The gap between real-world data and current AI capabilities

‍