Optimus Subprime |
“Right now, people think of Tesla as an electric car company or as an energy company. I think long-term people will think of Tesla as much as an AI Robotics company as we are a car company or energy company.” – Elon Musk
While the idea of a humanoid robot is nothing new, Elon Musk recently announced that Tesla will develop a humanoid robot, with the first prototype to be unveiled in 2022. Ever since the Tin Man stepped out of the Land of Oz and into the hearts and minds of children of all ages in 1900, we have allowed robots into our lives, to the point that robots now freely roam the streets in our collective imagination.
100 Years of Robots
The word “robot” was first used by Czech playwright Karel Čapek in his 1920 play R.U.R. (Rossum’s Universal Robots). Now, 100 years later, some of our most endearing Hollywood characters are robots. Think of C-3PO, the eloquent and often hilarious humanoid robot character from Star Wars, or the friendly Iron Giant from the movie of the same name, or the intelligent child-like robot character, David, played by a young Haley Joel Osment in A.I. Artificial Intelligence. And then there is the dark Terminator robot character, personified by Arnold Schwarzenegger in the 1984 James Cameron blockbuster The Terminator, along with five sequels, each with an ever more capable and terrifying Terminator robot, sent back from a dystopian future to kill the fictional character, Sarah Conner.
Three Laws for Robots
While The Terminator highlighted the possible dangers of robots in general, and Artificial Intelligence (AI) in particular, the ethical and moral dilemmas of robots were first addressed by Isaac Asimov as early as 1942 when he penned the “Three Laws of Robotics” in his short story Runaround. This theme is explored with poignant dramatic effect in the 2004 movie I, Robot, based on Asimov’s 1950 short-story collection Hardwired, which also references the three laws:
- A robot may not injure a human being or, through inaction, allow a human being to come to harm.
- A robot must obey the orders given by human beings except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
Elon and the Robot
Elon Musk has often credited Isaac Asimov as being instrumental in stimulating his creativity and future-oriented thinking, going so far as to say that Asimov’s **Foundation** series “was fundamental to the creation of SpaceX”. Perhaps Asimov’s fictional robots are equally fundamental to Musk’s ambition to develop a humanoid robot, designed to perform “dangerous, repetitive and boring tasks”. Perhaps Musk’s humanoid robot is the next pillar of future human development, built upon the same “foundation” first laid by Asimov.
Real-World AI
Of course, there is more to creating a fully autonomous humanoid robot than reading a sci-fi series. Indeed, in light of Tesla’s recent advances in the development of what Musk calls “real-world AI”, as demonstrated by its Autopilot and Full Self Driving systems, using passive vision and artificial neural networks to solve autonomous driving, a humanoid robot seems like a logical next step to solve the far more daunting real-world AI problem. So, while others like Google and OpenAI (another Musk startup) are trying to solve the AI problem in a virtual environment (like Google’s DeepMind computer and OpenAi’s GPT-3), Musk and his team of engineers at Tesla are approaching the problem from the other end, from a real-world environment – on city streets, roads and highways teeming with moving and stationary objects – which is several orders of magnitude more difficult to solve, even if the ultimate goal is the same for all AI developers.
A Robot on Autopilot
Musk confirmed as much when he stated that “Tesla is arguably the world’s biggest robotic company. Our cars are basically semi-sentient robots on wheels.” Andrej Karpathy, Tesla Senior Director of AI, went into more detail as he explained: “We are effectively building a synthetic animal from the ground up. So, the car can be thought of as a synthetic animal. It moves around, senses the environment, and acts autonomously and intelligently. We are building all the components from scratch in-house. So, we are building all the mechanical components of the body, the nervous system, which are all the electrical components, and for our purposes, the brain, which is the Autopilot, and specifically the synthetic visual cortex, much like the biological visual cortex in the human brain”.
Tesla’s visual and neural net architecture has complexified from a simple 2D image-based network four years ago to a 3D vector space and a spatial recurrent neural network video module that now adds memory (time) to transform this architecture into a 4D visualisation, similar to human vision (something that has not been done before).
“It makes sense to put that onto a humanoid form,” concluded Musk.
Tesla’s Got Talent
So that explains why Tesla is developing a humanoid robot. Or does it? Perhaps the development of a Tesla robot, which Musk jokingly referred to as **Optimus Subprime**, is simply a more optimal allocation of human resources.
Think of it like this: If Tesla is recruiting the world’s best AI engineering talent to solve just one finite problem – autonomous driving – will this exceptionally talented pool of engineers become redundant once they solve this problem? The solution to the redundancy problem is simple. To retain these exceptional engineers long term, Tesla needs to challenge them with a far more lofty goal. Instead of solving autonomous driving only on a 2D plane of operation, these engineers are now challenged to solve the autonomous operation of humanoid robots on a 3D plane of operation. This is as big a challenge as you can get, and the only challenge beyond this would be to solve Artificial General Intelligence (AGI). But that’s a challenge for another decade.
If you think you have what it takes to develop real-world Artificial General Intelligence, here’s a job posting on Tesla’s website:
“Develop the next generation of automation, including a general-purpose, bi-pedal, humanoid robot capable of performing tasks that are unsafe, repetitive or boring. We’re seeking mechanical, electrical, controls and software engineers to help us leverage our AI expertise beyond our vehicle fleet.”
This may be your last chance before the robots take over!
Robots to Mars
Clearly, the world has no real need for humanoid helper bots, as simple tasks like vacuuming the floor and washing the dishes are already addressed more efficiently with autonomous vacuum cleaners and dishwashers. SpaceX, however, will need humanoid robots if it is to deliver on its mission to make humanity a spacefaring civilisation. These robots will be better suited for long term space travel, and once on Mars, they will be better suited to work in the harsh environment without the need for oxygen, food, and water and without the fear of radiation. These humanoid robots would be ideally suited to build a sustainable human habitat long before SpaceX ever sends humans to Mars.
So why use Tesla, a company solely dedicated to accelerating the world’s transition to sustainable energy, and not SpaceX, to develop humanoid robots? To answer the “mission creep” question, there is a twofold answer. Tesla already has the building blocks (the synthetic visual cortex) in place, but more importantly, Tesla is free to hire engineering talent at its discretion. SpaceX, on the other hand, is constrained by the US Government’s national security concerns. As part of the military-industrial complex, all SpaceX employees and contractors need some form of security clearance from the US Federal Government, requiring all SpaceX employees to be US citizens. On the other hand, Tesla is a public company free from all this government scrutiny, and therefore more flexible to hire talent as it sees fit.
Last Word
While there may be significant demand for humanoid robots that can perform dangerous, repetitive, and tedious tasks here on Earth, the optimal application of such robots seems to lie further into the future and farther away from Earth. Developing a humanoid robot will help Tesla retain its phenomenal AI engineering talent well into the future while advancing Musk’s vision to make humanity a spacefaring civilisation over the next millennium.
Of course, Optimus Subprime may be nothing more than the fulfilment of the childhood sci-fi fantasy of a kid from South Africa who developed and sold a computer game called Blastar at the age of ten.
Ad astra!
Report by Wilhelm Loots | Images © Tesla
