Imagine the most expansive possible market for a physical product you could possibly sell. Are you considering cellular phones at this moment? Cars? What is it?
According to Geordie Rose, the chief executive of Sanctuary AI, all of these markets are sizable, but in the future decades, a new product will be introduced that will completely eclipse the significance of those giants.
Phoenix is a humanoid robot that is currently being developed by a company based in Vancouver. When it is finished, Phoenix will have the ability to grasp what we want, how the world works, and will have the abilities necessary to carry out our directions.
“The long term total addressable market, which is the labor market, is the largest one that has ever existed in the history of business and technology. “It’s everything that we want to have accomplished,” he says.
That assertion is qualified by him as follows: “There is a long way to go from where we are today.” He says this so we won’t get too far ahead of ourselves.
Mr. Rose is unwilling to establish a time frame on when a robot might be in your house, washing your laundry or cleaning the bathroom. He says this is because the future is unpredictable. But some other people in the industry who I’ve talked to suggest it might happen within the next ten years.
There are dozens of other companies all over the world working on the technology right now.
Dyson is making investments in artificial intelligence and robotics in the United Kingdom with the goal of automating home tasks.
Tesla, the electric automobile firm founded by Elon Musk, is currently one of the most well-known businesses in the industry.
It is currently working on the Optimus humanoid robot, which, according to Mr. Musk, might be available to the general public within the next five years.
We will find out in a moment if that turns out to be accurate. What we are able to say at this point is that the development of humanoid robots is advancing as a direct result of leaps forward in artificial intelligence.
If technological advancement continues at its current rate, ten years from now will feel like an eternity. “You know, every month there are new developments in the world of AI that are like fundamental change,” says Mr. Rose, who has a degree in theoretical physics and previously created a business that specializes in quantum computing. “You know, every month there are new developments in the world of AI that are like fundamental change.”
Late in 2017, a robust version of ChatGPT was made available to the public, which sparked a widespread interest in artificial intelligence. Because of its capacity to produce a wide variety of relevant text and images, it has given rise to competitors and sparked a rush of investment in AI technology.
Nevertheless, the work of developing the AI that would make it possible for a robot to carry out useful activities is a distinct and more challenging endeavor.
In contrast to ChatGPT and its competitors, humanoid robots need to be able to explore the real world and comprehend the connections that exist between the various things that make up that reality.
For humanoid robots, accomplishing tasks that many people would consider simple are big achievements.
In one of Sanctuary’s test projects, for instance, the robot Phoenix has been stuffing articles of clothing into plastic bags while working in the backroom of a store in Canada.
Because bags are floppy, they are transparent, and there is a spot where they open, this is a problem that engages a lot of different complicated issues in an AI driven robotics system. “This is a problem that engages a lot of different complex issues in an AI driven robotics system.
Mr. Rose explains that in most cases, after physically opening a bag, you are required to free one hand before proceeding to place things in the bag.
“The manipulation of bags is actually very, very hard for robots,” he continues, which is a line that makes the humanoid robots of today sound far less frightening than some of their counterparts in Hollywood.
The Sanctuary is equipped with a method for instructing Phoenix in various chores, such as bag packing. It will first video a certain process being carried out in collaboration with a company, and then it will digitize the entire event.
This information is utilized to construct a virtual world that, in addition to including all of the objects, replicates the physics involved, including gravity and resistance.
After that, the AI is able to perfect the task by performing it in the simulated setting. It is granted a million tries, and when the developers believe the AI has mastered the event in the virtual world, they will let it try it in the real world. It can have as many as it wants.
By being prepared in this manner, Phoenix is now able to perform around 20 distinct jobs.
Mr. Rose believes that mastering specific jobs that will be valuable for business is the future of humanoid robots and that this is the way forward. A robot that is capable of doing housework is still quite some way off in the future.
Providing the robot with a sense of touch so that it can determine the amount of force to apply to an object constitutes one of the most difficult problems.
“We have a facility with these types of tasks that comes from an evolutionary heritage, that’s like a billion years long… they’re very hard for machines,” says Mr. Rose. “We have a facility with these types of tasks that comes from an evolutionary heritage, that’s like a billion years long.”
A significant amount of effort remains to be done in order to construct a robot that is capable of handling all of the situations that could arise in a busy workplace or a busy household.
“You cannot place a robot in an environment that is not structured and then ask the robot to move around without essentially causing damage to the environment.” According to Professor Alireza Mohammadi, who was responsible for establishing the Robotic Motion Intelligence Lab at the University of Michigan-Dearborn, “it is too much for technology to ask at this moment in time.”
He makes the point that even if you train an AI using millions of different situations, there is always the possibility that it could encounter something in the real world that it has never seen before, causing it to behave in an unpredictable and possibly harmful manner.
According to him, one of the problems is that humans have an innate comprehension of the relationship between context and outcomes. For instance, we might speculate that a dog that is really eager will jump in front of us and account for this possibility in our plans.
Converting something like that into a robot is a really challenging task.
“Within ten years, we might have robots that are capable of walking around with some guidance, but not in completely unstructured environments,” adds Professor Mohammadi.
But if those obstacles can be solved, then humanoid robots might be able to start taking over professions that are currently performed by humans.
Mr. Rose notes that there is a labor shortage in a number of countries, and he suggests that his robots might one day be able to replace such places.
Heriot-Watt University and the University of Edinburgh have collaborated to create the National Robotarium, an institution that is dedicated to artificial intelligence (AI) and robotics. Stewart Miller serves as the National Robotarium’s chief executive officer.
“At some point in the not-too-distant future, there will be robots performing tasks that were traditionally carried out by people. The question then is, what does that signify?” he explains.
There will be some discomfort associated with our further expansion. But when we really stop to think about it, we have the ability to start putting more of an emphasis on and concentrating on what it is that human people do best; we just need to free up the capacity to do it, and we don’t have to spend any time doing what it is that machines do best.