The Chief Technologist at Amazon Robotics, Tye Brady, joined InnoLead on stage at the Impact conference last October. In an interview with Scott Kirsner, CEO and Co-founder of InnoLead, Brady shared his take on the intersection of artificial intelligence, robotics, and automation.
He doesn’t see humans being engineered out of the equation. “Collaborative robotics means that people and machines are working together,” Brady said. “It’s not people versus machines, but people and machines working together.”
This session was recorded at Impact 2023 on October 27, 2023. To watch highlights, click “play” above. Below is a transcript of the complete session.
• • •
Scott Kirsner:
There are many different lenses you can look at AI through, and many different definitions, but how do you see it intersecting with the world of robotics and physical things right now?
Tye Brady:
A couple of things. One is machine learning and AI have completely transformed robotics. It has taken us a long time to actually get to, I’d say, an applied real-world robotic state, which we are doing well inside of Amazon, and machine learning, particularly in the last 15 years for computer vision, for perception, and the planning of robotics has really changed the game.
At Amazon, we’re really proud that we’ve been using machine learning really since the start, right? So you can imagine if you needed to order and predict the supply for people to buy since the mid-90s, machine learning has been part of that system. That has continued to evolve with the birth of AWS. We have now, I’d say, a pretty healthy cluster of companies — over 100,000 companies that are using our general AI toolset and some of the foundational models that we have evolved. We use them as well inside of our everyday applied robotics. I think the one big thing that we’re going to get across, hopefully here in this talk, is we’ve focused very much on the practical, very much on the applied sense. We make people at the center of our robotics universe. We aim to extend human capability. We aim to eliminate the menial, the mundane, and the repetitive, and I can share with you a lot of examples of how we’re doing that today.
Scott Kirsner:
I want to show the video and just the latest generation of robotics, but just in terms of the capsule history — so there was a robotic startup company called Kiva Systems here in Boston that Amazon acquired probably in 2012. It was a significant acquisition for Amazon to really decide, “Hey, we want to own this capability.” You’ve been continuing to build on the warehouse robotics capability. One thing that I just find interesting, which I’d love for you to talk to for a second is, some of the robots in the world that we’re familiar with, like the Roomba vacuum cleaner robot, they’re made in Malaysia, they’re made in Taiwan. You guys actually make robots here in the Boston area by the thousands. Why?
Tye Brady:
Yeah. It’s another thing that we’re really proud of, actually, here in Massachusetts. We have the world’s largest fleet of industrial mobile robots. We have more than 750,000 of these drive units. We have a lot of other robotic systems, but our mobile drive units, which I think we’ll see a video of, we manufacture all those right here in Massachusetts. We try to source locally as much as we can. The labor pool is fantastic. Our robots are built to last for 10 years in the field, which is a really big deal. We have the evidence now, because that acquisition happened more than ten years ago.
Scott Kirsner:
So 750,000 is the number you’ve produced that are out in Amazon’s centers?
Tye Brady:
Yep. They are all over the world, and they all ship out of a dock door here in Massachusetts. It’s remarkable when you think about the system of manufacturing and assembly. It can be done, especially with the help of technology and robotics, right here in the United States, it can be done right here in Massachusetts. It creates lots of jobs in robotics, from manufacturing to the care and the assembly of robots, all the way up to the design and frontline implementation. We actually have two facilities where we manufacture here in Massachusetts, and we can we can crank it up if we need.
Scott Kirsner:
Alright, cool. Let’s look at the video, which I think doesn’t have sound so you can feel to talk over it. I’ve noticed you guys now have Amazon blue robots —
Tye Brady:
That’s our Amazon Prime color. So this is a system that we just announced called Sequoia and what Sequoia represents is the first containerized storage solution. That acquisition that you mentioned, Scott, 10 years ago, primarily was an acquisition to bring in what we call goods-to-person-type fulfillment system. So instead of a person going down to fulfill your orders with a shopping cart and walking many miles a day, picking the item out and putting it into a cart. Instead of that, you have the goods come to the person, where they can be picked out and sourced just at the right time.
The system that you’re looking at here, as compared to our Kiva pods, those yellow pods that are somewhat famous in the robotics world at least, we change dthe game here with our containerization. Containerization allows us to inbound our goods about 75 percent faster. Under the roof, which means it goes from our storage, Amazon.com, where we sourced the item, to the dock door about 25 percent faster as well. It’s kind of a big deal to us that, given the scale, — we shipped eight billion packages last year, eight billion. Which is sizable for sure. Our robotics systems are involved in about 75 percent of those to date. That continues to grow. But we could not do the eight billion without two key elements. One is the robotics and the automation that we’ve added inside our systems. In a very practical sense, I’m really proud of the women and men who have done that job. They are pioneers, truly pioneers. Then the second essential element is amazing frontline employees that are using the systems that we build their systems for, so that it’s a symphony of people and machines working together to do the task.
Scott Kirsner:
I’m curious like in this audience, how many people are at companies where robotics or industrial automation is part of your business? Alright, so an interesting chunk…
Tye Brady:
Interesting. I thought every hand would be up.
Scott Kirsner:
Well, I mean, I think a lot of people are at financial services, maybe healthcare, what’s an example of another industry where you didn’t raise your hand — like Scott Burns? Did you not raise your hand?
Tye Brady:
Energy, yeah, oil and gas seems really, really ripe for robotics and automation. I’m not criticizing by any stretch. I will say that we’re very much in the early days of robotics. I believe it’s taken us 50 years to get here, like 50 years to get here, right? The birth of the computer, actuators, connected computers, low-cost sensors, and machine learning. We have finally gotten here. Now we have an application. This is why I say that we’re pioneers. We’re not really doing crazy things.
We’re doing basic things. We’re moving things. We’re manipulating things to identify things. We’re storing things. We’re sorting things. We’re packing things. These are what I call the core fundamentals… [Robotics is] going to impact many sectors….
Tye Brady, Amazon Robotics
We’re doing basic things. We’re moving things. We’re manipulating things to identify things. We’re storing things. We’re sorting things. We’re packing things. These are what I call the core fundamentals, that in an applied world, in a very practical world, which is why robotics is working in e-commerce, it allows us to attain a world-class capability in robotics. I believe that the skills that we are pioneering, we can see it impacting other sectors, even inside Amazon. Amazon is a rather large company, but I also believe that it’s going to impact many sectors, and we’re seeing this in the medical field and agriculture.
Scott Kirsner:
Talk about where you think it goes next — medicine, agriculture? At Mass Robotics yesterday, we saw some interesting stuff like robotic tackling dummies, and there were some video conferencing robots.
Tye Brady:
Yeah, well, first of all, I’m super proud. I like a proud papa when it comes to Mass Robotics. It’s the world’s largest incubator and accelerator for robotics out there. Really proud that we co-founded, that I co-founded that in 2015, I think it was something like that, with the belief that robotics and automation, with a people-centered approach will change the world to a very good place, right? I’m super optimistic about that. The more people become, I’d say, users, the more people become acquainted, and the more people see the power of robotics, not as displacement, not as replacement, but as augmentation. When you reframe your relationship with machines, it changes your fundamental world, right?
Collaborative robotics means that people and machines are working together. It’s not people versus machines, but people and machines working together. That allows us to gain more productivity…
Tye Brady, Amazon Robotics
For us, it’s very foundational. We believe that the future of automation is in collaborative robotics. Collaborative robotics means that people and machines are working together. It’s not people versus machines, but people and machines working together. That allows us to gain more productivity, like the example that you just looked at. We have efficiency productivity, but we also have safety, a lot of safety enhancements as well. As compared to our manual buildings, the recordable injury rate is lower by 15 percent, because we can put picks in kind of the right places.
Scott Kirsner:
There’s a question that kind of ties into that, which jobs have the robotic platforms created versus removed? How close is the margin of creation versus replacement? Is the story you would tell, “Hey, we just keep adding these jobs in distribution centers, and we just keep making them more efficient?”
Tye Brady:
We like to think of these virtual cycles and these flywheels, right? So more productivity gains you more customers, and when you have more customers that is sacred. We aim to delight our customers with the world’s largest selection of goods at a low cost with ultimate customer convenience. Then, that gives us more revenue to invest more in both jobs, and also in better robotics. It’s been 10 years since Amazon has really invested heavily in robotics, right with the acquisition of Kiva Systems. Which was mind-blowing for, I think it was $775 million at the time. Which was like, oh, my gosh, in robotics, that was a big deal, because it hadn’t really come to age yet. But the point is that since that investment, we have created hundreds of thousands of new jobs. In robotics alone, we’ve created more than 700 job types — just job types…. It sounds sensational, but it’s not. I mean, if you just think about 20 years ago, we didn’t have social media coordinators, right? I think we have a job opening right now for a blockchain architect for our Prime Video streaming platform. Like what’s a blockchain architect? That wasn’t really a big deal. Streaming platform? What is that? And Prime Video? I don’t understand that.
Jobs will change for sure over time. The question is, are you doing what’s right for your customers? And are you doing it in a way that makes people more empowered, that makes people safer, and that gives people the ability to think at a higher level?
Scott Kirsner:
You mentioned the founder, Jeff Bezos, and I mean, there was this interesting dynamic, right? Where his venture capital firm was investing in robotics startups, like Rethink Robotics. He was kind of hanging around as an informal adviser to iRobot for a lot of years, and so it was an interesting example to me of, when he was CEO, really learning about the space and absorbing kind of what was possible, trying some things through startup investing.
Tying back to the first session today, we were talking about the CEO’s role and the board’s role.
Would you talk about him, or Jeff Wilke, or others who have seen the future, and supported this investment in the future? It just feels qualitatively different than some CEOs that are a little more numbers-driven, just Wall Street analyst relations driven?
Tye Brady:
Yeah, I find…the best visionaries are the kindest people, are the people that you can relate to, are the people who are not in a conversation to tell you how smart they are. They’re in a conversation to be genuinely curious about what it is that you do and open a platform for curiosity, not just for yourself, but also for them, but a curiosity about what is possible in tech.
I think Jeff has an amazing way of doing that, of creating a safe space for people, and allowing you to think big. It’s one of our leadership principles. And it’s very inspiring. I think it’s a thing that all of us should take away, in that, it’s all about connection, genuine connection. It’s about passion. I’ve been programming since 1976. I love computers, and I love robotics and space. I happen to have that same overlap Venn diagram with Jeff, which is great. I try to do that with my own team, I try to create a culture of innovation.
Scott Kirsner:
When you say culture of innovation, can we just drill down?
A culture of innovation to me means that it is in your DNA — you can’t help but be curious. A culture of innovation means inclusion, it means a propensity to look around corners when things are difficult.
Tye Brady, Amazon Robotics
Tye Brady:
Sure, a culture of innovation to me means that it is in your DNA — you can’t help but be curious. A culture of innovation means inclusion, it means a propensity to look around corners when things are difficult. It means wanting to understand different points of view. It’s wanting to grow a team that is rooted in kindness to each other. Kindness is really important, because kindness can be a great parachute for failure, right? When things go wrong, and you have a foundation of kindness, then you can get through it just fine. When things go great, and you have a foundation of kindness, it’s a huge force multiplier. I can tell you in my 30-plus years, and you know when you get to be my age, it’s 30-plus and it just kind of stops there, right? I can tell you, I think more about the people and the amazing people when I was at MIT. I worked at MIT for about 10 years, I worked at Draper Laboratories, and Amazon, I’m coming up just over eight years.
I think about the people really, first and foremost, and their ability and desire to tackle hard problems. No doubt it’s definitely a natural function of who you want to be around for sure. But when you get high-tech-focused people that are kind, that are gentle with each other, that have a propensity for curiosity, like just a propensity of why does this work? Why doesn’t this work? What can we try? How can we experiment better? Let’s give it a shot. Some really amazing things can happen. I think, back to your original point with Jeff, Jeff has an incredible curiosity. He tested out Amazon in the mid-90s, just with books, just to see coudl this idea work, based on his own curiosity and it worked out.
Scott Kirsner:
There’s this question [from the audience] that’s really interesting about, you know, while automation does create new jobs, for those who are tech-savvy, what happens to low-skilled workers who don’t fit these categories of like building the robots, assembling them, designing them?
If you got plucked out of Amazon, and you’re sent to DC to think about what do we do with folks who may have graduated high school and not have gone to a two-year college or four-year college, what sorts of stuff are you telling them? Or what sorts of stuff would you be doing for the labor force — separate from warehouse workers, truck drivers, and people washing dishes? I’ve seen robots that assemble salads — we have robotic salad assembly startups here in Boston. What would you be doing or thinking about that workforce?
Tye Brady:
So first of all, my first job was as a dishwasher as well. When we require people to have exceptional thinking or higher education to use machines, it’s a fail. It’s actually on us, the designer of robotics, the designer of our machine learning systems, to make it in a way that’s intuitive and natural for people to use. When people use it in a way in a way that benefits them, then those products tend to rise. Even inside what we call our fulfillment processes in robotics, we keep very much in mind — if it’s hard to use, it’s not going to work very well. If there’s friction there, it’s not going to work. So what are the —
Scott Kirsner:
I think the question is coming from maybe the perspective of like, what happens? How do we think about these low-skilled workers?
Tye Brady:
Here’s the problem that I have. It’s the term low-skilled workers. The human brain is incredible. It’s an incredible thing to think about the capabilities that we have, and the ability to understand, to problem-solve, to think with reasoning, to just use common sense. I don’t need higher education to do that. Now, don’t get me wrong, because I can just see being quoted, I do believe in higher education, for sure.
I think education is what actually is going to save the planet. But as a designer of a machine, I can’t think about specialized widgets. What I would say is, many people have cell phones in their pockets and many, many people use computers, and those allow you to be more productive in your job. The designers of those machines…do a really good job of making it intuitive and natural for you to use. So when it gets to the workforce itself, this is where I get very passionate about how we need people to be more… We need people to just be people, right? It’s up to us to build the machines to adapt to the human environment, and we avoid all costs for people to adapt to the machine environment.
When I hear friction points like that, it’s really on us as designers of these incredible machines to make it more natural, more easier to use. When you do that well, then what you do is you create entry points into all levels of job creation. We have jobs opening on a manufacturing line to create the robots. We have jobs open for deployment to deploy these robots all over the world. We have jobs in support of robotics. Of course, we have jobs for the Ph.D.s working on the design of robotics, but also undergraduates working on the design of them. So, it creates many entry points.
Scott Kirsner:
Okay, I think we’re short on time here. There’s a good question about humanoid robotics that I want to get to really quickly: what’s preventing the humanoid type robots from being broadly used i warehouses? Obviously, Elon Musk seems very excited about humanoid robots. We haven’t seen Amazon, I think, talking about that. What’s the barrier? Why are you guys not wasting cycles on humanoid robots?
…My opinion is when you see a humanoid form, your expectations are set really high.
Tye Brady, Amazon Robotics
Tye Brady:
We actually have an investment with Agility Robotics right now, that’s a bipedal robot. I’m really interested in the motion and the movement of bipedal robots. I think this question is more about the humanoid form. The humanoid form is very interesting. Again, if it’s people and machines working together, my opinion is when you see a humanoid form, your expectations are set really high. Usually, my expectation is, the reality is going to be very low. Okay, and that turns into a frustrating experience. So, you definitely bring in game if you want to bring in the humanoid form, but you will have to really rise to a great challenge.
I’m more of a person of form and function in design. What actually makes sense. Going back to your dishwasher example, if you gave the challenge to a new roboticist to say, “Hey, I need you to build a dishwasher.” They would start with “Okay, well, how do we do it today?” You reach over, grab a dirty plate from the bottom of the sink, turn on the sink, grab the sponge, rinse it and you put it there. So we’re going to need a human form. I’m gonna need human hands. I’m going to need perception systems to detect dirt and soil and whatever…and that’s really the wrong way to go.
Instead, we have this amazing robot already called the dishwasher in our kitchen, just put it in there. It’s a fantastic robot. It’s such a good robot that you don’t even think of it as a robot. Roboticss have done well in becoming ubiquitous, and they just blend into the background. That’s what we want to do inside of Amazon as well.
Scott Kirsner:
I guess just a very last question to touch on, and maybe to tie this back to Cassie’s discussion, too, is like we are at this like, really interesting moment with what is becoming possible with robots other than dishwashers, with the convergence of data and machine learning, and generative AI is doing. I think it does create a lot of anxiety. It almost feels like when you talk to some people, they feel like we’re at this industrial revolution moment where they want to be the Luddites and say like, “Hold on, let’s smash the machines before they disrupt jobs or disrupt my company, or even worse, my job.” You sound like you’re more of an optimist that like, “Hey, there is this powerful technology that’s kind of flowing into the economy,” but you’re seeing it as a force for good.
Tye Brady:
I’m very much the optimist here. It’s just my own hands-on experience with technology. Like why would you ever say no to a digital assistant that would allow a doctor to detect breast cancer sooner? Why would you say no to that? That would be an amazing machine learning system that does that, that augments a doctor who treats the patient. Why would you say no to a machine learning system?
My daughter is studying for medical school, and she takes the MCATs, and she does these online tests that assess where she’s weak, and where she’s strong, and she gets more questions. There’s a lot of machine learning behind that. If you came here, maybe you popped in your favorite navigation app that got you here sooner? Why would you say no to that? Those are helping the human condition; those are helping us be more human and more intelligent.
For us inside Amazon, where our aim is to drive a low cost for our goods through the efficiencies that we gain inside our fulfillment processes by using robotics, and employing our frontline people. We want to have the world’s largest selection of goods. We want to have the ultimate in customer convenience when it comes to delivery of those goods. If we can do that smartly and safely with robotics, why would you say no to that?
So again, it’s just a reframing of the relationship with machines where it’s not you versus the machine. If you go down that path, you’re down a path for failure — 100 percent. There is no such thing as a perfectly automated system. It has never existed, and nor will it ever. It will never exist. You could play this back 20 years from now. I am 100 percent sure that there’s no such thing. That brings in the element of people, and how can you build your machines in a way that extends people’s creativity, ingenuity, innovation, intelligence, and even connection to one another.
