François Chollet works on deep learning at Google in Mountain View, CA. He is the creator of the Keras deep-learning library, a contributor to the TensorFlow machine-learning platform, and the author of the popular textbook Deep Learning with Python. Besides leading Keras development, he does deep-learning research, with a focus on the application of machine learning to abstract reasoning. His papers have been published at major conferences in the field, including the Conference on Computer Vision and Pattern Recognition (CVPR), the Conference and Workshop on Neural Information Processing Systems (NIPS), the International Conference on Learning Representations (ICLR), and others.
AI has been getting more intelligent in recent years, at a fast pace. Or has it? Because the field has been guided by measures of task-specific skill, disregarding the amount of prior knowledge & training data required to achieve those skills, it turns out that AI has been very successful in developing systems that perform particular tasks without involving intelligence. As a result, modern machine learning is data-hungry, lacks flexibility, and struggles to handle the uncertainty and variability of real-world situations. To make progress towards more robust and flexible AI, we need a better understanding and measure of what it means to be intelligent.