Podcast Recommendation: Spatial Intelligence

Date: September 25th, 2024

I recently listened to an a16z podcast episode featuring Fei-Fei Li and Justin Johnson on Machine and Deep Learning in spatial intelligence. Fei-Fei is a computer science professor at Stanford University and Co-Director of Stanford’s Human-Centered AI Institute (Stanford University, 2024). She established one of the most significant datasets in Computer Vision called ImageNet that was used to train models such as AlexNet. Justin Johnson is an assistant professor at the University of Michigan and a Research Scientist at Facebook AI Research (FAIR) (University of Michigan, 2024). Justin specializes in Computer Vision and Machine Learning. In 2016 he co-authored a paper with Fei-Fei and another researcher on a novel image transformation method. A use case of the method is taking a blurry image and improving the resolution.

The episode starts with Fei-Fei and Justin going over the history of machine learning and deep learning. They talk about how researchers eventually realized that not only the algorithms used to create models are important, but also the data and computing power. The latter half of the episode focuses on spatial intelligence, which Fei-Fei Li and Justin Johnson describe as an AI or machine learning system's ability to understand and interpret three-dimensional space and the temporal dimension in visual data. Spatial Intelligence can be used for problem sets in the real world and in the artificial world. In the real world, Spatial Intelligence can be used with Augmented Reality to assist users with handling tasks such as fixing a car. An example in the artificial world that Justin gave is utilizing Spatial Intelligence to create video game environments with less labor than is currently needed.

The podcast episode on YouTube is linked below.

References

Stanford University. (2024). Fei-Fei Li Profile. Retrieved from stanford.edu

University of Michigan. (2024). Justin Johnson. Retrieved from web.eecs.umich.edu