Age of AI) Ep. 6 Will a robot take my job?

 

• In today’s rapidly evolving digital landscape, one of the most pressing questions on many people’s minds is, “Will a robot take my job?” It’s a query that’s as popular on Google as it is in casual conversations. This concern isn’t unfounded; it mirrors the apprehensions faced by horse messengers when phones first rang across the world. We’re witnessing a similar transformative era where vinyl records, once a symbol of audio revolution, are now more of a nostalgic relic in the age of digital music streaming. It’s a reflection of how we adapt to technological advancements, trading film cameras for digital ones, not because we resist change, but because we inevitably embrace it.
• As we navigate through these technological shifts, it’s essential to recognize that automation isn’t just a harbinger of job displacement; it’s a catalyst for new opportunities. The advent of automation may signal the end for some professions, but it simultaneously lays the groundwork for emerging industries and job roles that we haven’t even conceived yet. Just like the transition from horse-drawn carriages to automobiles opened up countless new avenues, the rise of AI and automation is poised to do the same. This evolution, while daunting, is an integral part of our journey towards a more technologically integrated future.

Trucking

• E-commerce is driving demand for shipping higher than ever, but the harsh realities of long-haul trucking are sending professional drivers looking for other work, leaving a 50,000-driver shortage in the U.S. alone.
• There are of course some industries that will be impacted by A.I. in terms of labor workforce being diminished, but there are some industries that don’t have the workforce to sustain it, and so that disconnect, that’s where A.I. can fit in very nicely.
• Automation is coming, but her job won’t be lost.
• It will evolve. I’ve been in this career a long time, I’ve seen how everything has been changing.
• The monotony of coast to coast, and the pay being low, and, you know, you’re sleeping in a sleeper, you’re sleeping at a truck stop. It’s hard. It’s a hard life. I saw an ad for a test driver, and I said, “That has to be the perfect job.”
• We’ve all heard about self-driving cars, but self-driving trucks, why are they setting the pace of the autonomous driving industry?
• We already have robotic airplanes. Most airliners fly themselves, but self-driving cars are a harder problem
• because the roads have a lot more things going on than the air does, and driving on the freeway is much easier than driving in the city. So we will see fleets of automated trucks, I think, long before we see self-driving cars in the city.
• The truck was originally not made to be automated. We add our intelligence, the servers, and all the sensors to make this work.
• Driving the car is what is called an “A.I. complete problem.” If you solve it, you solve every other problem in A.I. However, to solve it, you need to solve every problem in A.I. So it requires vision, it requires robotic control, so it requires motion and navigation, but it also requires social interaction with the other drivers, so at the end of the day, in order to drive a car well in the city, you need everything. It requires an enormous amount of common sense.
• Common sense. That’s what Maureen is doing here… overseeing the A.I. truck as it learns to navigate real-world conditions, ready to take over if anything goes wrong.

Testing Automation

• I’m a test engineer with TuSimple. My job is to let her know what’s happening, let her be absolutely comfortable the truck is doing what it’s supposed to do.
• Never is one drive or one day ever the same. I feel like I have a relationship with the truck. Easy, easy… I talk to the truck. I tell it when it does good. I’ve never been a test driver before, and it’s never been a truck driving by itself before. We’re both actually learning together.
• Long-haul trucking can be dangerous, it can be boring, and therefore, it’s ripe for automation.
• The A.I. uses the images that are coming from the cameras with other sensors, LiDAR, and radar. LiDAR is a laser range finder that is measuring the distance to objects 360 degrees around. It gives us a three-dimensional picture of the world.
• In the virtual world, there is a direct correspondence of the map and the location where the vehicle is, and the behaviors of other vehicles, their locations, their speed, their future intentions, and once all these things are re-created, it’s basically asking the computer to play a game.
• When I’m looking at the data that the truck is feeding me, not only am I seeing the current environment, but I’m seeing ahead into the future.
• This truck’s tricked out with sophisticated cameras and software, but it still doesn’t have one critical element… the thing that A.I. may never have. The central problem in A.I. is that human beings have common sense and computers do not. We take common sense for granted. We know how the world works, and everything that we do in our daily life involves common sense people have been trying to imbue A.I. with common sense. Since the beginning, and what is extraordinary is, at the end of the day, how little progress we’ve made.
• The endgame is that, one day, the AI truck will not need Maureen, David, or any person to ride with it. Each one of these test runs is a building block toward that goal, designed to help it learn how humans operate and make decisions, even when you get thrown a curveball.
• The truck wants to pass the camera crew, but the constantly changing speeds and unusual behavior confuses the AI so it refuses to take the risk.
• There are behaviors that the truck has to learn to become human, in some sense, because not everybody else is operating at the same level.
• Turns out that getting the first basic systems was a big milestone, but then there’s this long tail of difficult situations that are much harder to solve. When you’re merging, do you have to give way, or does it make sense to go forward? That requires a level of understanding that we humans barely have, but machines aren’t there yet.
• Eventually, we hope that the trucks react to a change in environment as quickly as we can, regardless of what we throw at it
• Watching something develop over time, you never thought it would learn like this, and now it is learning, and you feel like a proud parent. It was making some awesome decisions that I’ve never seen it make before.
• I get asked all the time by people, you know, “Oh, you’re taking truck drivers’ jobs away by having autonomous trucks,” and I don’t believe that’s true, because it’s not taking away jobs, it’s taking the place of a job that a driver doesn’t want to do.
• What Maureen thought was the twilight of her working life actually became the dawn of something new, and it’s an age-old phenomenon.
• Tractors didn’t make farming obsolete, and video didn’t kill the radio star. It just provided new opportunities.

The Port

• What do we do when A.I. changes what the job functions are, and what the workforce looks like, which it’s going to do. So this is a fear, because change, people don’t like, and change that you don’t understand… is terrifying for a lot of people.
• But the vessel sizes just continued to grow, seven times larger than the ships that we ever had in the beginning, so there was a need for a new model, which is what LBCT is.
• When a ship pulls alongside, we’re using different algorithms to figure out traffic, scheduling, dispatching, and planning. We’re doing this with 10 different cranes at the same time, 50 different vehicles at the same time, and we have 30,000 different places to put that container in the yard.
• Ship-to-shore cranes will set the containers on the platform. On that platform, there are over 20 cameras that use Optical Character Recognition, or OCR.
• Another crane will then pick it up, put it onto a fully-automated vehicle. That vehicle will drive through the yard, and it’ll get to the right spot, planned by the system. What looks like a little wad of bubble gum,
• these are transponders buried in the ground. They’re in a grid pattern, there’s more than 10,000 out on the production field. There’s an antenna on the front, and the rear, and they read them as they drive, and then transmit that to the system to let it know where it is. Once it reaches the yard, an automated stacking crane will pick it up and deposit it into a yard block. From the block, it’ll eventually get deposited onto a truck or onto a bomb cart, where it can make its way onto a train.
• They say that we can eventually learn to adapt to an environment that mixes robotic and human elements,but it’s more difficult for machines to adapt to us, so here, they need to be separated, because of safety
• Because humans are unpredictable, automation can be physically dangerous to people, because if I’m a 200-ton machine and I don’t see you? I might just run you over.
• This facility, it is fully automated, but it’s not as if someone comes in, pushes a button, and puts their feet up on their desk. It is still employing hundreds of individuals on any given day. This is an asset-intensive facility. The number of mechanics necessary to maintain this facility has certainly grown.
• About 80% of the time, most of the faults are resettable from here. Automation can make the workplace safer, and this is a good example… taking someone out of the yard, and putting them into the control room.
• A.I., like any technology that comes into our society, it changes the workforce. There’s this misconception that there will be fewer of these new jobs. That’s actually not the case.
• I was a crane operator. Back in the day, we’d have to climb up five stories into the cab. It was a little hard, especially on the back, and then when automation started, they just shifted me over, and now it’s a lot easier.
• If you don’t embrace it, you’re just gonna fight it, and it’s not gonna… You gotta use it as a tool. our obligation was to re-train those same individuals into the new jobs that that technology has created. We are the progress that was needed for the goods movement industry. For the foreseeable future, there will continue to be a lot of things that only humans can really do. And so the future of work is not humans being replaced by machines. It’s humans figuring out how to do their job better with the help of machines.

Robots With Vision

• ALOS is one of the most advanced humanoids on the planet. It can walk, it can talk, it can see you in 3D
• Structured environments are good for automation. Smooth surfaces, right angles, less chance for chaos. But what about more complex, human environments? Can we learn to work with robots hand-in-hand?
• One thing they’re doing is developing A.I. and robotics to use in environments that are unstructured, more human. Like at home. But it can’t do most of those things out of the box, so you really have to take it as a tool and teach it how to do a lot of these things.
• On TALOS, we want to explore two different aspects of A.I. First, to perceive objects in the world.
• Computer vision is a way to mimic how we see the world. I can say, “Well, that’s a TV,” or “That’s a cat,” or “That’s a dog,” so what is it that I’m looking at when I’m looking at an image? Computer vision is trying to figure out what are the objects that are in this image that represents what we would understand.
• Also we want to explore application of A.I. in understanding locomotion in terms of balancing.
• The biggest misconception about robots is that they are more capable than they are, that they’re more generalized than they are and they’re not quite there yet.
• So you may have seen humanoid robots that can do parkour, do backflips. The only ones that are really comparable are the Valkyrie that was designed by NASA, of which there’s only two or three in the world…
• and the Atlas from Boston Dynamics, but most of them have very few sensors, and a lot of the time, they will be remote-controlled, or given a very clean, pre-scripted path, and in a situation like that, what they’re doing is pushing the limits of the mechanical systems. That’s very similar to what we’re doing here, except on our case, we’re pushing the software side.
• The next step going forward is to start replacing all of our dumb, blind, dangerous machines with machines that have sensors and vision built into them, that can work side by side with humans to be more productive and safer at the same time.
• So for a task where the robot might be carrying something with a human, it’s going to need to be able to feel how hard the human is pushing, pulling, much like if you’re being guided in a dance.
• So as humans, we have a sense of touch, and we have a sense of how much force we’re applying, and is being applied to us. With TALOS, it doesn’t have a skin, it’s hard plastic everywhere, so it can only really sense what’s in its motors.
• TALOS is using compliant control in the upper body, which basically makes him soft. The robot automatically senses the forces being applied to the table. The human is still fully in control. This is very important to the safety of humans surrounding the robot.
• Within robotics, the areas where you’re really gonna see the important advances are those environments that are relatively controlled and predictable, so a good example is an Amazon warehouse. In those environments, you already see lots of people, and also lots of robots… and they’re working together.
• Once you have a process, and you’ve reduced it to an algorithm, you can replicate it so that if a robot can learn a whole new skill, you can copy that knowledge through the cloud to all the other robots, and now they all have that same skill. It’s a whole new world, a whole new kind of economics, and we’re just beginning to understand its implications
• A whole new automated world still hard to imagine? Maybe it’s more about what we’re gaining than what we’re losing. Will automation make things faster? More efficient? More productive?

Cooking With Robots

• It’s also about big waste. Last year in the U.S., about $200 billion worth of food was wasted. $200 billion.
• Perhaps half of all food that’s produced in the world is wasted. Artificial intelligence can help us change the supply chains so that they are much more efficient.
• One of the classic tensions in every food business is producing too much, and having waste, or not producing enough food, and having stock outs. It takes years and years of practice for a human to forecast that, but that’s actually a data problem that machines, and A.I. in particular, are really, really good at getting right. So at its core, Zume is a fundamentally new logistics model.
• We predict what we’re going to sell before you even order it, and that whole predictive layer of the business is all driven by A.I.
• “Why on Earth does a pizza company want a Chief Technology Officer? But then when I heard the vision, the fact that we were going to be able to change an industry by focusing on this amazing end-to-end platform, which just happens to also produce amazing pizza, then it made sense.
• To disrupt Domino’s, Zume uses machine learning to try and forecast how much pizza people want, what kind, and when. It crunches dozens of different variables… location, day of week, weather, what’s on TV, past ordering trends, and then predicts how many Sgt. Pepperonis San Francisco will want on a Tuesday, for example.
• Our supply chains are incredibly inefficient. Our food processing systems waste a lot of food. If we could improve predictions, we could eliminate most of that waste. We’d know what the demand was going to be, where the products were going to be, and ultimately, that would make us all better off. Absolute perfection would be no pizza gets wasted, and anytime a customer looks, there’s always one of the type of pizzas they would like available. So it’s an optimization problem. How close can you get?
• Based on its algorithm, the A.I. predicts how many pizzas, and what kind, to load onto the mobile kitchens. When orders come in, it also decides which mobile kitchen will bake it, and exactly when to put each one in the oven. Every single pizza has its own cooking profile and recipe. So each one of these ovens actually is a robot. They’re connected to cloud services that’re always monitored, making it possible for one person working in here to cook up to 120 pizzas an hour.
• We use A.I. in simulation around how do we get the right delivery estimated time of arrival? The A.I. can suggest here’s the best place to put a truck so that we get the hottest, freshest product to our customers.
• When we get the orders, our system controls the ovens, which controls the cook time, and how long the pizzas need to cook. [Satchell] The A.I. can change the workflow on one of our vehicles to make sure the pizza’s cooked at the last moment. …and then the system chooses which driver to send it to.
• It’s algorithms plus, it’s A.I. plus humans, not A.I. instead of humans.
• All of that data gets fed back into the learning algorithm, so every week we’re trying to evolve our learning algorithms to do a better job than the week before. Part of what we’re optimizing for is reducing waste. Maybe if you can use prediction to go back into the supply chain and predict more accurately what you need to grow, and where you need to get that product, I think you could fundamentally change things.

Change Is Coming

• We’re in the early stages of a massive change in technology that’s allowing machines to do a lot of tasksthat previously only humans could do. The future of work is evolving. Automation’s making the workplace safer, greener, more efficient, for sure. If we can figure out how to integrate A.I. technologies, not to replace humans, but to augment their abilities, to make life more fun, to make us more productive and more creative… I think that’s where the power of the machine will come.
• Thousands of years ago, we were hunter-gatherers. Eventually, we became farmers, and now that we’re maybe entering this fourth industrial revolution, one that connects the physical, biological, and digital, it’s not just jobs that are being changed by A.I., but us. We are the progress. So why focus on the rear-view when we can look to the road ahead to see what we’re becoming?
• It’s a real challenge in any case to build robots with the kind of mobility and dexterity that begins to approach what human beings have. Kind of the cliche that we all think of is that we’d like to have a robot that can go to the refrigerator and grab a beer for us. I mean, if you think about what’s involved with that, that’s still an enormous challenge. A robot that’s able to do that has to be heavy enough to go to a refrigerator. It has to be able to open the door, and it has to have the visual perception and the dexterity to reach in and grab the beer, and already that implies a fairly heavy machine, otherwise it would just tip over. Building a robot to do that is not just difficult, but it’s gonna be fairly expensive, so I think that it’s gonna be quite a while before we really see the kinds of robots that we imagine from the science-fiction world operating in our daily lives.

The insights and information presented in these articles are based on the YouTube Originals Series “The Age of AI.” All script and content rights belong to the creators and producers of the series. This series served as a primary reference in the development of these articles.