00:00My name is Raquel Urtasun. I am the founder and CEO of Wabi. We are providing a solution
00:09that is much more capital efficient, as well as a much faster path to market, starting
00:14with self-driving. But you're going to see us doing much more than just self-driving
00:17trucks, which is our first product.
00:22I started working in AI 25 years ago as an undergrad, where I got really fascinated by
00:26this concept of an artificial brain that can really do the same task as humans. However,
00:33this was just an idea at the time, so we needed to really push the technology to the next
00:37level. So I spent most of my career doing research as a professor. Once this technology
00:41was really ready to get it to the next level, then I identified that there was a big hole
00:48in self-driving, and in particular in trucking, where the incumbents were going with AV 1.0,
00:57which was an approach that was very hand-engineered, that was not scalable, that was not going
01:03to get us there to a product that will really solve the issues that the potential customers
01:08will need.
01:09Generative AI is at the core of all the technology that we do at Wabi, but in particular there
01:17is two big pillars of very differentiated GNI technology. On one side we have the virtual
01:24driver, this is the brain that drives our self-driving trucks, and it's key in order
01:30to be interpretable, something that you can verify, something that technology, where you
01:36can really, in the event of something going wrong, you can trace the decisions of why
01:41the system decided to do a particular maneuver. So that's one big pillar of technology.
01:46The second pillar of technology solves the other core piece, which is the data problem.
01:52In order to be able to train systems and in order to be able to make the safety case and
01:59test under all possible conditions the system, it's not enough to drive on the real world,
02:05you will need to drive so many billions of miles that you're exposed to too much risk
02:11and nobody has the fleets necessary to do that.
02:16Instead, what we did with the second generation of generative AI technology is to build a
02:21simulator which we call Wabi World, which is the same as the real world. So we actually
02:25solved the data problem. So we can really expose the system to safety critical accidents
02:30or unavoidable accidents, et cetera, so that our system is safer, but also probably safer
02:37than anybody else.
02:41Where we are now, it's very clear to everybody that actually this technology is going to
02:45revolutionize the physical world and that has become kind of like the aha moment for
02:51everybody like, you know, this is the next big revolution. This is really where the big
02:58disruptor is. However, when we started the company three and a half years ago, you know,
03:03this was, you know, we could see this ahead of time and it was very contrarian.
03:08AI is a tremendously powerful technology that can really help business leaders to increase
03:16the productivity of their workforce, to make better decisions about, you know, business
03:22decisions. But at the same time, there is also a lot of hype about, you know, AI in
03:28the sense that, you know, everybody's doing AI these days and there is this thought that,
03:33you know, a standard tool will solve everybody's problems. So it's very important to make
03:38informed decisions, very careful decisions about where is this technology the right solution
03:45or the right tool, the way that we interact with machines, the way that we interact with
03:48the physical world. The, you know, the world is for this technology to conquer and I think
03:53it's, you know, I'm very excited about that future and, you know, how we can provide,
03:59you know, a better world for the generations to come.
04:03Microsoft Mechanics
04:05www.microsoft.com
04:07www.microsoft.com
04:09www.microsoft.com
Comments