00:00I would say not yet, but you can see it. You can see it in light.
00:05In my opinion, it depends what you call the tragedy moment for robotics.
00:08In my opinion, a tragedy moment for robotics would be one where everybody can experience
00:13what it's like to interact with a robot and the robot responds to what you want
00:17and does what you want in a completely new environment or a completely new setting.
00:22So that's what it means to me. I don't think we're there yet, but we're getting there.
00:26What gets us there? What are the technological barriers that need to be overcome to get to that point?
00:32Yeah, I mean, this is precisely what we're invested on.
00:35So our goal is to build the intelligence layer that can power any robot and make it really intelligent
00:41so that it's intuitive for humans to use, so that it can do a broad range of tasks,
00:46so that it can make the robot smart enough to understand an environment, reason about it,
00:52and then take action to complete a completely new task that you might have asked it to do.
00:56That's what we're going after. I think that there's been significant progress over the last three, four years.
01:03I mean, before, robots didn't even understand what an object was and what their environment.
01:08They were just simply points in space without any meaning.
01:12We've actually introduced around 2022 LLMs and BLMs into robots that taught them to understand natural language.
01:21It also taught them how to reason about their environment and then how to take action in a way that
01:26was towards a particular semantic task
01:29or a particular meaningful task for the robot. That didn't exist before.
01:32Humanoids get a lot of headlines. They get a lot of buzz. Do you think that's distracting?
01:37Do you think investors, the media, others, do you think we're still kind of overly obsessed with humanoids?
01:43That's a great question. I mean, we believe in a world where there will be many different robot types.
01:48I actually think I'm very practical about this.
01:50I think that there's going to be a lot of tasks where a different form factor that is not a
01:54humanoid is the right thing.
01:55However, I do think humanoids are going to become useful and are going to be an important form factor for
02:00a few reasons.
02:01First, I think the world is made for humans.
02:04So if you want to immediately deploy a robot in an environment that is human-centric, then a humanoid is
02:10probably going to be an easy one to deploy.
02:13Actually, there's an interesting research aspect to this, which is if we want robots to learn from humans and the
02:20robot has a human form factor, it's actually easier.
02:23Because it can now translate what the human is doing when it's, say, cooking a recipe into what the robot
02:29should do in order to do the same task.
02:31So that's a learning perspective that I find interesting.
02:34And then the other piece is that my goal is to really show that we can get to AGI in
02:39the physical world.
02:40What does that mean?
02:41It really means that the robot can do anything that human can.
02:44So I feel like humanoids are going to enable us to show and really believe that we have gotten to
02:49that level.
02:49So that is another reason why I like it.
02:52But we care about all kinds of robots, and we're interested in enabling, bringing intelligence to all robots.
Comments