Machines with Brains
Throughout this series, we’ve explored the changing nature of humanity as an increasing number of automated and robotic systems enter our lives. But what exactly is a robot? One might think of the robots depicted in pop culture, or perhaps the hulking great objects being developed by companies like Boston Dynamics. The answer isn’t particularly clear—which is why Quartz asked a group of very smart people what they believe a robot to be.
“To me, a robot is a tool that helps us with physical tasks. It has to have two things: a body capable to do the work required by the task, and a brain that controls the body to execute that task. My dream is to develop reliable autonomous robots for a broad range of tasks large and small—and I think that recent developments in the fields of 3D-printing and soft robotics have the potential to make that happen!”
“In the future, a robot will simply be a physical entity controlled by artificial intelligence interacting with the world. In a few years, we’ll have robots that look like us that we can control from afar. Called avatars, this will enable us to be there without traveling, for example, give a speech, mingle with the audience, cook a meal, visit with a loved one, do rescues, perform remote surgery, and many other tasks. Robots are already carrying out missions in dangerous environments (as the French did recently in a terrorist incident).
To me, the most exciting future mission of robots will be to carry out medical interventions inside our body. I recently swallowed such a robot, which went through my digestive tract sending signals and movies to look for problems (I was fine!). In the 2030s, these medical robots will be blood cell sized. They’ll start out by delivering medications to precise places where they are needed. Ultimately, they will extend our immune system to kill pathogens (such as cancer cells), carry out micro surgery (such as removing plaque from our arteries), augment the function of failing organs, and a myriad of other health interventions.”
“I think the classical definition of a robot is something that senses its environment and takes action accordingly. And I think that’s actually a very wide net: If you think of older technology that we’re familiar with, you can think of a dishwasher as a type of robot—you put your dishes in its mouth and it licks them clean. I think smartphones could be viewed as robots; some smartphones now turn off when you’re in a movie theater, they know where they are and they don’t bother you or ring because they’re location-aware. A Nest thermostat is a classic robot: it monitors the temperature in your home, and it adjusts it up and down the way you like it; it trains and learns from experience to know how you like your home, and it takes action to make that happen for you. Those are really simple robots—they don’t have a humanoid shape or arms, but they actually do act in the world and adjust it for you.”
“I’m going to give you an answer and an example.
I think to properly understand robots, in the general sense, is to let go of the bipedal mechanical man that is the image in our minds when people say the word “robot.”
A robot is something that has three basic characteristics: It is something that can sense the world in one or more ways; it has computation (so it can think about the sensing that it got from the world), and it has actuation; it can have some impact back on the world.
So an example of a robot in that sense is the seatbelt reminder in your car. It can sense whether the seatbelt is buckled, and whether there is weight pressure on the seat—there is a tiny bit of computation—and it can affect the world by beeping. That is a robot.
My favorite example of where I believe a lot of technology is going is another example from a car: It’s anti-lock brakes. When you slam your foot against the brakes, you’re not actually, mechanically, connected to the brake anymore, in any modern car. You’re talking to a robot when you have that interface, which is that brake pedal—it used to be mechanically connected, it’s just not anymore—you’re now telling the robot how fast you’d like to stop, and the robot says, “OK, good to know how fast you’d like to stop, and I will stop that fast for you, and no faster, but I might stop slower than you’d like, because I’m solving this other problem for you, which is I don’t want you to lose rolling friction, so I’ll pump the brakes if I think it’s necessary, and as much as I think it’s necessary, because I don’t trust you to pump the brakes yourself.”
That, the sensing, the pressure of the brakes, and also sensing in the wheels, that tell you whether or not you’re losing rolling friction, the computation from experience—and there’s learning that’s involved in that—and then the actuation of how hard to press on the brake pads, that is a robot taking your desire and doing something much more sophisticated in a way you don’t have to think about. Isn’t that what we want robots to be like in our lives? Where they take very simple inputs we don’t have to think about and do very sophisticated things that make our lives better? I think there’s going to be a lot of that in our future.
“I’ll tell you an anecdote that happens to me multiple times a day, especially when I’m out talking to people like I am now: There was a round of acquisitions of robot companies at Alphabet and there was a period of time after Andy Rubin [the founder of Android, then in charge of Google’s robotics division] left where they were somewhat in limbo. In the end, most of those (all but one of those) moved over to Alphabet. And so I get asked, “What amount of what you’re working on is ‘the robots’?” And I say, “Well, that sort of depends. Almost everything we do is a robot.” And they say, “What about Waymo?” And I say, “Waymo is a robot.” And they say, “What about Loon?” And I say, “Loon is a robot.” And they say, “Well what about your airborne wind turbines?” “Those are robots” “What about Wing?” “That’s a robot too!” So I think once you really think about it, there are robots without stretching the term past any kind of painful point—most of what we do is robots.”
“A robot is a device that does something mechanical based on sensor input. A thermostat is a simple robot; a self-driving car is a highly advanced robot.”
“I have a very expansive view. Basically, almost anything is a robot—and here’s why: Those of us studying walking, you know, you originally come in there and you think, “Wow, you’ve got to have a brain, you’ve got to have muscles, you’ve got have sensors.” And as we learn more and more about the mechanics of the task, we figured out, and there was a guy named Tad McGeer, there was another set of people led by Tom McMann, who figured out you can actually get a mechanical device to walk without a brain, without muscles, and without sensors. Okay, just mechanics. So now we had this problem that, wow, a mechanical device can in some way compute using mechanics. Is that a robot? I’ve got to say yes, because it’s walking. So now, we have the question, when they open a door, is the door sort of mechanically computing what path it can take on its hinges? And I’ve got to say that’s a robot, too. Once you’ve said those things, you say pretty much anything is a robot, because we could build robots without any sensors. We could build robots without any brains. And we could build robots without any muscles. So there’s no key ingredient that a robot has to have.”
“Robots are essentially autonomous hardware controlled and empowered by artificial intelligence software.
With 37 years of researching, building, and investing in AI, I see the AI and robotic revolution is on the scale of the Industrial Revolution, but probably larger and definitely faster. Though robots may take over jobs, however, believe me when I tell you there is no danger that they will replace human beings. Based on my engineering extrapolation, these AI and robots run “narrow” applications that master a single domain each time, but remain strictly under human control. Robots by themselves are unable to learn common-sense reasoning, creativity, and planning, nor self-awareness, feelings, and desires as humans do. The “general AI” isn’t here. There are simply no known engineering algorithms for it. And I don’t expect to see them any time soon.
I think that the real danger is not that such a scenario will happen, but that we won’t embrace the option to double down on humanity while also using AI and robots to improve our lives. This decision is ultimately up to us: Whatever we choose may become a self-fulfilling prophecy. If we choose a world in which we are fully replaceable by machines, whether it happens or not, we are surrendering our humanity and our pursuit for meaning. On the other hand, if we choose to pursue our humanity, and even if the improbable happen and machines truly replace us, we can then capitulate knowing that we did the responsible thing, and that we had fun doing it. We will have no regrets over how we lived.
I do not think the day will ever come—unless we foolishly make it happen ourselves. Let us choose to let machines be machines, and let humans be humans. Let us choose to use our machines, and make human lives better.”
“In my mind, the word “robot” is not from this academic field. Most technology terms are from some scientific field, but “robot” is from a play. I think that the meaning of “robot” should stay as the definition in that play, which is humanoids. We should call other things some other words—currently the meaning of “robots” is expanding to vehicles and drones and other things, and I actually don’t like that.”