WEBVTT - Chief Technologist for Robotics at Amazon Tye Brady Talks Smart Glasses to Guide Drivers

0:00:02.600 --> 0:00:08.639
<v Speaker 1>Bloomberg Audio Studios, podcasts, Radio News Emily. I don't know

0:00:08.680 --> 0:00:11.920
<v Speaker 1>if you saw this earlier, but Amazon is reportedly developing

0:00:11.960 --> 0:00:15.920
<v Speaker 1>smart eyeglasses to guide its delivery drivers around and within buildings.

0:00:15.920 --> 0:00:19.159
<v Speaker 1>This according to Reuters, who cited five people familiar with

0:00:19.200 --> 0:00:23.159
<v Speaker 1>the matter. If these work, glasses would navigate drivers on

0:00:23.200 --> 0:00:25.880
<v Speaker 1>a small embedded screen along routes and at each stop.

0:00:25.920 --> 0:00:28.280
<v Speaker 1>The ideas that it aims to reduce delivery time.

0:00:28.280 --> 0:00:31.080
<v Speaker 2>What do you think so because you see it in

0:00:31.120 --> 0:00:33.400
<v Speaker 2>your vision field instead of having a look onto the

0:00:33.440 --> 0:00:35.239
<v Speaker 2>dashboard of the truck runder or.

0:00:35.320 --> 0:00:37.320
<v Speaker 1>No, or when you're walking around holding a package, you're like, Okay,

0:00:37.400 --> 0:00:39.120
<v Speaker 1>this is exactly the house it needs to be delivered to.

0:00:39.720 --> 0:00:39.960
<v Speaker 2>Yeah.

0:00:40.040 --> 0:00:42.360
<v Speaker 1>Kind of a cool little USh for I should note.

0:00:42.360 --> 0:00:45.400
<v Speaker 1>An Amazon spokesperson told Reuters it is continuously innovating to

0:00:45.440 --> 0:00:48.199
<v Speaker 1>create an even safer and better delivery experience for its drivers,

0:00:48.600 --> 0:00:51.800
<v Speaker 1>but it would not compliment comment on its product roadmap.

0:00:52.159 --> 0:00:54.280
<v Speaker 2>I want to see, like a prototype, what they would

0:00:54.320 --> 0:00:54.640
<v Speaker 2>look like.

0:00:54.880 --> 0:00:57.800
<v Speaker 1>I think this is a real use case for smart

0:00:57.840 --> 0:00:59.520
<v Speaker 1>you know, for like the Google glass type thing. They've

0:00:59.520 --> 0:01:01.440
<v Speaker 1>been used for quite a while. They never caught on

0:01:01.480 --> 0:01:03.520
<v Speaker 1>with consumers. They're trying to make that happen at Snap

0:01:03.520 --> 0:01:05.800
<v Speaker 1>and at Meta. Still we'll see if it happens, but

0:01:05.840 --> 0:01:08.679
<v Speaker 1>the industrial use case is certainly interesting. We got with

0:01:08.760 --> 0:01:11.520
<v Speaker 1>us Ty Brady's chief technologist for robotics at Amazon.

0:01:11.600 --> 0:01:11.839
<v Speaker 2>Tie.

0:01:11.880 --> 0:01:14.200
<v Speaker 1>I'm not gonna put you on the spot and ask

0:01:14.240 --> 0:01:17.400
<v Speaker 1>you to comment about this report because we already got

0:01:17.400 --> 0:01:20.679
<v Speaker 1>a comment from an Amazon spokesperson, but it does talk

0:01:20.720 --> 0:01:24.080
<v Speaker 1>a little bit about innovation happening at Amazon. I know

0:01:24.120 --> 0:01:27.920
<v Speaker 1>you guys did this interesting survey with MIT on what

0:01:28.000 --> 0:01:31.120
<v Speaker 1>AI could mean for workers. Before we get to that,

0:01:31.319 --> 0:01:35.200
<v Speaker 1>just give our viewers our listeners an idea of the

0:01:35.319 --> 0:01:38.040
<v Speaker 1>role that robotics play right now at Amazon.

0:01:39.360 --> 0:01:40.720
<v Speaker 3>Well, first of all, it's a pleasure to be here,

0:01:40.720 --> 0:01:43.400
<v Speaker 3>and thank you for having me onre I really appreciate it. Boy.

0:01:43.520 --> 0:01:47.080
<v Speaker 3>I'll tell you the age of physical AI is here

0:01:47.200 --> 0:01:50.960
<v Speaker 3>and it is really proving useful for our customers. We

0:01:51.040 --> 0:01:56.280
<v Speaker 3>have our robotic systems inside of our fulfillment centers. We

0:01:56.480 --> 0:01:59.800
<v Speaker 3>actually just roll our next generation fulfillment center down in

0:02:00.000 --> 0:02:02.760
<v Speaker 3>Louisiana and we're seeing a lot of returns and gains

0:02:02.800 --> 0:02:03.040
<v Speaker 3>on that.

0:02:03.360 --> 0:02:06.360
<v Speaker 1>Is this the Kiva system that Amazon bought over a

0:02:06.400 --> 0:02:07.520
<v Speaker 1>decade ago.

0:02:07.880 --> 0:02:10.480
<v Speaker 3>Wait, it's based on that. That is part of it.

0:02:10.720 --> 0:02:13.280
<v Speaker 3>What we do is that we have robotics that can

0:02:13.320 --> 0:02:15.680
<v Speaker 3>move goods, that's part of the Kiva systems, but also

0:02:15.840 --> 0:02:19.480
<v Speaker 3>robotics to store goods and sword goods and identify all

0:02:19.560 --> 0:02:22.520
<v Speaker 3>the various goods. And we have this in this next

0:02:22.520 --> 0:02:25.400
<v Speaker 3>generation fulfillment center where we have ten times the amount

0:02:25.400 --> 0:02:27.600
<v Speaker 3>of robotics that we've ever had under one roof, and

0:02:27.639 --> 0:02:30.160
<v Speaker 3>we're already seeing that we can process those orders twenty

0:02:30.160 --> 0:02:34.040
<v Speaker 3>five percent faster and also pass along to a lower

0:02:34.040 --> 0:02:35.320
<v Speaker 3>cost to our customers.

0:02:35.400 --> 0:02:38.600
<v Speaker 1>Just proving just so we understand when when we order

0:02:38.639 --> 0:02:40.720
<v Speaker 1>something on Amazon using the app or using the browser,

0:02:42.120 --> 0:02:47.560
<v Speaker 1>does a robot now pick that item at fulfillment center?

0:02:47.680 --> 0:02:48.680
<v Speaker 1>Is it still a human being?

0:02:49.320 --> 0:02:54.320
<v Speaker 3>Yeah, it's It's an amazing, amazing series of events that

0:02:54.360 --> 0:02:56.959
<v Speaker 3>actually happens. When you go on Amazon dot com. You're

0:02:57.000 --> 0:03:00.520
<v Speaker 3>getting your holiday order redder ready. Of course, want to

0:03:00.520 --> 0:03:02.440
<v Speaker 3>have the world's the largest selection of good for our

0:03:02.520 --> 0:03:05.760
<v Speaker 3>goods for our customers. We want to process that at

0:03:05.760 --> 0:03:08.600
<v Speaker 3>a low cost and then have the ultimate customer convenience,

0:03:08.600 --> 0:03:12.160
<v Speaker 3>and robotics is helping all along that way. Right, people

0:03:12.200 --> 0:03:15.720
<v Speaker 3>and machines working together, we have AI systems that source

0:03:16.200 --> 0:03:18.160
<v Speaker 3>that put the right products in the right area so

0:03:18.200 --> 0:03:20.120
<v Speaker 3>that can be closer to our customers. That helps on

0:03:20.160 --> 0:03:23.720
<v Speaker 3>our delivery times. We have robots that move safely around people,

0:03:23.760 --> 0:03:27.239
<v Speaker 3>that we inbound those goods into our buildings where it

0:03:27.240 --> 0:03:30.160
<v Speaker 3>can take on more goods, put more goods in the

0:03:30.600 --> 0:03:34.040
<v Speaker 3>same footprint as we've had as compared to our manual buildings.

0:03:34.080 --> 0:03:37.040
<v Speaker 3>We can store actually forty percent more goods, and then

0:03:37.080 --> 0:03:39.760
<v Speaker 3>we have robotic systems that help sort and even package

0:03:39.960 --> 0:03:43.320
<v Speaker 3>those items for our customers. All this is so that

0:03:43.360 --> 0:03:46.240
<v Speaker 3>we can just have the ultimate in customer convenience and

0:03:46.280 --> 0:03:48.119
<v Speaker 3>get the right good to right to the customer's door.

0:03:48.640 --> 0:03:53.400
<v Speaker 2>Ty, what is a twenty twenty four robotics system and

0:03:53.480 --> 0:03:56.320
<v Speaker 2>how is that different than the systems that we saw

0:03:57.000 --> 0:03:59.880
<v Speaker 2>ten fifteen years ago. I'm just thinking about how you

0:04:00.120 --> 0:04:02.760
<v Speaker 2>to watch that show how it's made, and it would

0:04:02.800 --> 0:04:05.720
<v Speaker 2>show factories and there were always robots working in the

0:04:05.720 --> 0:04:09.040
<v Speaker 2>factories and moving things along conveyor about So what is

0:04:09.080 --> 0:04:13.080
<v Speaker 2>so different about, you know, the twenty twenty four robotics.

0:04:13.800 --> 0:04:17.600
<v Speaker 3>That's such a great question. We have really seen just

0:04:17.600 --> 0:04:20.080
<v Speaker 3>just huge advancements over the last few years when it

0:04:20.120 --> 0:04:23.679
<v Speaker 3>comes to robotics. This is why we call it physical AI.

0:04:23.800 --> 0:04:28.400
<v Speaker 3>It's really the embodiment of adaptive behavior in our robotic systems.

0:04:28.640 --> 0:04:32.680
<v Speaker 3>But our robotics systems shouldn't be viewed as singular and

0:04:32.720 --> 0:04:35.560
<v Speaker 3>as alone as just machines. Instead, what we do is

0:04:35.560 --> 0:04:38.679
<v Speaker 3>we build our machines to extend human capability. We build

0:04:38.680 --> 0:04:43.760
<v Speaker 3>our machines to augment folks to do their jobs better

0:04:43.839 --> 0:04:46.479
<v Speaker 3>and create a safer environment for them. We've seen those

0:04:46.520 --> 0:04:49.240
<v Speaker 3>benefits in the last four years with regards to safety

0:04:49.440 --> 0:04:52.800
<v Speaker 3>and also in regards to the efficiencies and the productivities.

0:04:53.120 --> 0:04:57.160
<v Speaker 3>But the big you know, I could talk about more

0:04:57.200 --> 0:04:59.640
<v Speaker 3>and more of the robotics that we've seen, but we

0:04:59.680 --> 0:05:03.480
<v Speaker 3>also have to I think the big mindset is when

0:05:03.520 --> 0:05:08.440
<v Speaker 3>you reframe your relationship with machines. Yeah, people first attitude

0:05:08.440 --> 0:05:11.760
<v Speaker 3>towards of how we build those machines that allows people

0:05:11.760 --> 0:05:13.760
<v Speaker 3>to do more. And now we are upscaling. We put

0:05:13.760 --> 0:05:16.520
<v Speaker 3>one point two billion dollars into an upscaling pledge for

0:05:16.600 --> 0:05:21.200
<v Speaker 3>our employees. And then we, as Tim said, we funded

0:05:21.200 --> 0:05:24.760
<v Speaker 3>this independent study with MIT in order to understand the

0:05:24.800 --> 0:05:29.120
<v Speaker 3>perception of technology and how people adopt machines and AI

0:05:29.720 --> 0:05:30.760
<v Speaker 3>into their work environment.

0:05:30.839 --> 0:05:33.039
<v Speaker 1>So Let's talk a little bit about that, because I

0:05:33.080 --> 0:05:35.599
<v Speaker 1>think you hit on a really important point here, and

0:05:35.640 --> 0:05:37.520
<v Speaker 1>it's the idea that and I was kind of joking

0:05:37.520 --> 0:05:39.760
<v Speaker 1>about this a little while ago. I was like, can

0:05:40.320 --> 0:05:44.000
<v Speaker 1>these robots be programmed to love? But the fact is

0:05:44.000 --> 0:05:49.120
<v Speaker 1>is you do have to be comfortable being around automated

0:05:49.279 --> 0:05:53.640
<v Speaker 1>things that could look like humanoids. They might not look

0:05:53.680 --> 0:05:58.839
<v Speaker 1>like humanoid robots. What did you find through this collaboration

0:05:58.920 --> 0:06:02.200
<v Speaker 1>with MIT about how people want to interact or need

0:06:02.240 --> 0:06:04.160
<v Speaker 1>to learn how to interact with machines.

0:06:05.040 --> 0:06:08.680
<v Speaker 3>It was a fascinating study. It was actually groundbreaking. MIT

0:06:10.120 --> 0:06:14.080
<v Speaker 3>enlisted to IPSOS to survey more than nine thousand people

0:06:14.160 --> 0:06:19.200
<v Speaker 3>across nine different countries to understand how they perceive technology,

0:06:19.240 --> 0:06:23.000
<v Speaker 3>because perception is really important to adoption, even to innovation. Right,

0:06:23.360 --> 0:06:24.960
<v Speaker 3>as I said, we put people at the center of

0:06:24.960 --> 0:06:27.320
<v Speaker 3>the robotics universe. We really want to understand how people

0:06:27.880 --> 0:06:32.120
<v Speaker 3>and if they will adopt technologies. And broadly we found

0:06:33.040 --> 0:06:36.920
<v Speaker 3>the study found that a majority of people see robots

0:06:36.960 --> 0:06:40.279
<v Speaker 3>having a positive impact on their pay and on their career.

0:06:41.320 --> 0:06:44.320
<v Speaker 3>And there's three key three key findings that came along

0:06:44.360 --> 0:06:46.240
<v Speaker 3>with that. First of all, if people are asked to

0:06:46.279 --> 0:06:48.320
<v Speaker 3>work at a higher level and can focus on higher

0:06:48.440 --> 0:06:54.400
<v Speaker 3>order tasking. Then they're more keen to adopt technology, and

0:06:54.440 --> 0:06:57.000
<v Speaker 3>they're more optimistic when it comes to the use of

0:06:57.000 --> 0:06:59.839
<v Speaker 3>the technology for their own career goals in their own pay.

0:07:00.279 --> 0:07:03.360
<v Speaker 3>The second is if they felt valued by their employer. Right,

0:07:03.400 --> 0:07:06.640
<v Speaker 3>So being valued is both are you working in a

0:07:06.680 --> 0:07:08.719
<v Speaker 3>safe environment and do you have a great benefits in

0:07:08.760 --> 0:07:10.800
<v Speaker 3>pay And Amazon we're really proud of the benefits and

0:07:10.840 --> 0:07:14.280
<v Speaker 3>pay that we offer our employees, and also through automation,

0:07:14.400 --> 0:07:19.160
<v Speaker 3>we're actually reducing the number of recordable injuries significantly over

0:07:19.160 --> 0:07:21.800
<v Speaker 3>the past four years. And then the last part is

0:07:21.920 --> 0:07:25.040
<v Speaker 3>really those that want to learn more and take control

0:07:25.080 --> 0:07:26.800
<v Speaker 3>of their career. So if they want to learn and

0:07:26.840 --> 0:07:30.920
<v Speaker 3>grow in their career, they're more optimistic for technologies. And

0:07:30.960 --> 0:07:32.600
<v Speaker 3>that was really good good news for us as we

0:07:32.680 --> 0:07:36.360
<v Speaker 3>put in one point two billion dollars into upskilling our employees.

0:07:36.560 --> 0:07:39.200
<v Speaker 3>So they're good signals. But we're not done yet because

0:07:39.240 --> 0:07:43.200
<v Speaker 3>this is across many many workers, across many industries, and

0:07:43.240 --> 0:07:46.440
<v Speaker 3>we're going to follow on the study actually a survey

0:07:46.520 --> 0:07:49.720
<v Speaker 3>of our Amazon employees directly because we're always interested in

0:07:49.760 --> 0:07:51.640
<v Speaker 3>the voice of our customer, the voice of our employees,

0:07:51.680 --> 0:07:53.520
<v Speaker 3>and we always want to make this safe for more

0:07:53.560 --> 0:07:54.440
<v Speaker 3>productive environment.

0:07:54.640 --> 0:07:57.160
<v Speaker 1>What's a takeaway that you had from the survey about

0:07:57.240 --> 0:08:00.560
<v Speaker 1>how you're going to either design or implement robotics across Amazon, Like,

0:08:00.600 --> 0:08:02.520
<v Speaker 1>what's one actionable thing you took away from this?

0:08:03.920 --> 0:08:05.880
<v Speaker 3>Well, the first action that we had is that we

0:08:05.920 --> 0:08:08.520
<v Speaker 3>actually want to fund more studies. We want to fund

0:08:08.520 --> 0:08:11.400
<v Speaker 3>the study, particularly for Amazon employees, so we can hear

0:08:11.480 --> 0:08:15.160
<v Speaker 3>directly for them of their perception and how robotics is

0:08:15.200 --> 0:08:19.000
<v Speaker 3>really augmenting and extending the capability. So we are the

0:08:19.120 --> 0:08:22.960
<v Speaker 3>voice of the customers. Our first actual takeaway that we

0:08:23.040 --> 0:08:26.320
<v Speaker 3>have from the study. The second is really validation of

0:08:26.320 --> 0:08:29.040
<v Speaker 3>our philosophy that of putting people the center of the

0:08:29.080 --> 0:08:33.240
<v Speaker 3>robotics universe. Right, It's this validation of augmentation and extension

0:08:34.000 --> 0:08:38.320
<v Speaker 3>and allowing people to do what they do well, thinking

0:08:38.360 --> 0:08:41.200
<v Speaker 3>with common sense and reasoning and really understanding the problem

0:08:41.760 --> 0:08:45.079
<v Speaker 3>at hand. People do this amazingly well. So our job

0:08:45.160 --> 0:08:49.040
<v Speaker 3>as roboticist is to complement that amazing skill that our

0:08:49.080 --> 0:08:52.280
<v Speaker 3>employees have that people have with machines that can be

0:08:52.280 --> 0:08:55.640
<v Speaker 3>better designed. So very concretely, for example, we have a

0:08:55.800 --> 0:09:01.040
<v Speaker 3>robot called Proteus that is free roaming inside of our

0:09:01.040 --> 0:09:05.080
<v Speaker 3>fulfillment centers that moves goods on demand, and it understands

0:09:05.120 --> 0:09:07.960
<v Speaker 3>where people want to be, and people can look at

0:09:07.960 --> 0:09:10.480
<v Speaker 3>the robot and understand what the intent of that robot is.

0:09:10.600 --> 0:09:14.920
<v Speaker 3>So its job is to move these vessels of goods

0:09:15.360 --> 0:09:20.040
<v Speaker 3>to our dock doors safely, and it's fully around people. Right.

0:09:20.080 --> 0:09:22.960
<v Speaker 3>So the concrete actual thing is here is that we

0:09:23.000 --> 0:09:27.600
<v Speaker 3>are getting this really strong validation that by extending and

0:09:27.720 --> 0:09:31.040
<v Speaker 3>augmenting human capability, not only can we and more productive,

0:09:31.120 --> 0:09:32.920
<v Speaker 3>but we can also create a safer environment.

0:09:33.360 --> 0:09:35.400
<v Speaker 2>We don't have a ton of time left, but I'm

0:09:35.440 --> 0:09:40.120
<v Speaker 2>curious when you talk to Amazon employees, what are some

0:09:40.240 --> 0:09:43.640
<v Speaker 2>of the biggest concerns that they have when it comes

0:09:43.679 --> 0:09:45.240
<v Speaker 2>to robotics in action.

0:09:46.920 --> 0:09:49.800
<v Speaker 3>Yeah, well, we always think about the voice of the

0:09:49.800 --> 0:09:52.240
<v Speaker 3>customer first and foremost. It is one thing in order

0:09:52.320 --> 0:09:54.920
<v Speaker 3>to do like a robotic system in your lab and

0:09:54.960 --> 0:09:57.200
<v Speaker 3>convince yourself that it's going to work in the lab.

0:09:57.200 --> 0:09:59.400
<v Speaker 3>But when we roll out what we call our alpha

0:09:59.400 --> 0:10:02.720
<v Speaker 3>and beta employments with our customers, we get first hand

0:10:02.760 --> 0:10:05.440
<v Speaker 3>their quotes of what is working and what is not working.

0:10:05.800 --> 0:10:10.240
<v Speaker 3>We should build our machines too that you can reasonably

0:10:10.360 --> 0:10:13.600
<v Speaker 3>and tangibly use the machine in very intuitive ways. Right,

0:10:13.600 --> 0:10:15.840
<v Speaker 3>You shouldn't have to have eighteen degrees in order to

0:10:15.880 --> 0:10:17.720
<v Speaker 3>figure out how to use the machines, And we get

0:10:17.720 --> 0:10:20.760
<v Speaker 3>that feedback quite often to say, it needs to be simpler,

0:10:20.960 --> 0:10:23.120
<v Speaker 3>it needs to be less complex, and it needs to

0:10:23.160 --> 0:10:27.839
<v Speaker 3>have the utility in order to extend and complement human creativity.

0:10:28.640 --> 0:10:30.440
<v Speaker 1>We're going to have to We're going to have to

0:10:30.520 --> 0:10:35.040
<v Speaker 1>leave it there. Ty, do appreciate you joining us. Ty Brady,

0:10:35.160 --> 0:10:39.240
<v Speaker 1>chief technologist for robotics over at Amazon, joining us this

0:10:39.320 --> 0:10:41.120
<v Speaker 1>afternoon from Boston