ai.bythebay.io: Danny Shapiro Interview
Recording: ai.bythebay.io: Danny Shapiro Interview
you [Music] you artificial intelligence is the ability to take system and be able to perform different types of tasks and in this case we're able to use deep learning a form of artificial intelligence running on GPUs the graphics processing unit to be trained and then actually perform an activity much better than a human possibly could we're seeing a huge application of artificial intelligence in the automotive industry and specifically we're working with over 80 different automakers tier 1 auto suppliers different automotive startups and research institutions that are all developing self-driving technologies they're using our dr DX to the AI brain of the self-driving car and our drive work software which is a number of different AIS or recognizing lanes pedestrians free space as well as end-to-end deep learning to understand the whole environment and drive a car rapid advances now in deep learning are accelerating the race to the self-driving car we're using massive amounts of data collected from driving as well as simulation to train these neural nets once we've trained and validated and on that we can then deploy it in vehicles and in fact I Drive an AI self-driving car every day Tesla is using our technology developed their own software on top of our software that is now deploying that in production for traffic jam type situations as well as highway automated driving and soon residential when we're talking about self-driving cars though it's a very serious matter all right we have many tons of steel moving at high rates of speed with people inside those cars as well as people outside those cars the AI running those cars has to be perfect or as near perfect as possible of anything in the world and so it's not just a recommendation engine on a movie website but this is something that's taking data in analyzing it in making decisions so we need to ensure that the systems we're building our as robust as reliable as they possibly can be so that means having a whole variety of different sensors feeding our AI brain as well as then being able to test it and validate it before it goes out into the field so we make a system hardware and software that others build applications on we're working with so many different automakers tier 1 suppliers and startups and they're all developing different applications taking the deep learning process training and deploying and evolving I think the key thing here is that we need to not rush it and bring something to market before it's ready but be able to fully test it validate it and make sure it adheres to all regulation before we do so you [Music] you