
We are all getting a bit tired of Musk’s nonsense by now. Like an inadequate child claiming his dead-beat dad is actually an astronaut, he has been spouting demonstrably false techno-fantasies for decades now. Sure, he paid overworked engineers to produce EVs and self-landing rockets, but these concepts were already proven to work in the 90s. Everything else which falls out of his mouth, from Mars bound rockets, to vaccum trains, to self driving cars, to even his video game ability is all pure weapons grade bullshit. However, one example of his hogwash stands heads and shoulders above the rest: Tesla’s AI-powered robot, Optimus. Musk claims it will cost $30,000 when it eventually hits the market, and will rake in trillions of dollars as it replaces workforces across the globe. Musk couldn’t be more wrong if he tried, and this is yet more evidence he should stay off the drugs and fire the yes-men.
Let’s start with a recent video Tesla shared of Optimus, seemingly showing significant progress. Past versions have done nothing but shuffle around like my nan waiting for a hip replacement, or be pupeteered by interns functioning on nothing but coffee and desperation. A far cry from the do-it-all robot Musk promised. But this video showed Optimus doing a very jumpy dance while tethered to the ceiling (watch it here), before it stops abruptly, seemingly because it was about to tip over backwards. Such a manoeuvre is incredibly difficult and is a giant leap forward from where Optimus was just a few months ago.
But, this is only impressive if you live your life in the Musk bubble. Optimus isn’t even catching up to the competition in the real world. Over a decade ago, Honda’s ASIMO was doing far more controlled hops and jumps, totally untethered (watch that video here). Boston Dynamics had a far more complex robot doing repeatable backflips on uneven ground while untethered eight years ago (read more here)! Their latest robot can walk, run, jump, breakdance and interact with the complex world around it with such grace, precision and accuracy that it just looks human (watch more here). Now, sure, that robot was exponentially more expensive than Optimus. However, UniTree’s G1 is designed to be a similar commercial humanoid robot, and they recently showcased it performing repeatable sideways flips while untethered (watch the video here). And you can buy the G1 now for half what Musk promises to eventually sell Optimus for.
If Tesla unleashed Optimus in 2009, it would have been mildly impressive. But now, compared to the competition, it’s utterly embarrassing. It demonstrates how little technological understanding or advantage Tesla has.
But that isn’t really the problem here. Honda, Boston Dynamics, and Unitree all know that humanoid robots are actually a bit useless, and more of a technical demonstration than an economy-shattering revolutionary tool. Something Musk definitely didn’t get the memo on. Indeed, anyone with even a passing understanding of robotics will know that a humanoid robot is the worst solution in any scenario. As such, Musk’s obsession with them is not just perverse but also demonstrates his fundamental lack of knowledge. Let me explain.
Humanoid robots are inherently flawed. We already have robots that can do jobs faster, more consistently and more efficiently than the human form allows. For example, even the best humanoid robot couldn’t hand-weave cloth faster than an automated loom. This isn’t a limitation of robotics, but of the human form. We evolved to be general-purpose and adaptable, so our form is not the ideal solution for specific tasks. So, why make a machine with the same limitations as us?
Okay, but what if you want a general-purpose robot? Well, trust me, you don’t.
For most applications, a specialised robot is far cheaper. Why buy a $30,000 Optimus robot to hoover the office when you can buy a $500 Roomba instead?
Using a specialised robot is also always much easier. All you need to do is press a button on the Roomba, and it just works. Whereas, with humanoid robots, you need to program them to do this task. Not only is this arduous, time-consuming, and skill-intensive, but if you are getting someone else to program it, it’s also damn expensive. Oh, and for every task you want it to do, you need to reprogram it all over again.
Now, some people will say that AI will solve this. You can just tell the robot to hoover the house with that hoover, and it will crack on. But let’s not kid ourselves, AI is so far from doing that. Tesla can’t even make an AI that can drive better than a drunk 12-year-old, and doing complex tasks from simplistic voice commands has far more variability than self-driving. So no, programming and using these humanoid robots is an absolute nightmare, and will be for years.
When you stop and think about it for one second, it’s also painfully obvious that humanoid robots are inherently less consistent and reliable than specialised systems.
Take an articulating arm on a car production line, welding chassis together. It is bolted to the floor, making its movements incredibly precise and accurate. As we know what movements it will do, it’s easy to put in hardware and software checks to ensure it is doing the job well enough. As it is doing the same constrained task over and over again, it requires simple programming. This makes improving efficiency, accuracy, and reliability incredibly easy, as identifying areas of improvement, implementing solutions, and testing solutions is very straightforward. This is why these machines, once optimised, can work at superhuman speeds and with near-perfect reliability.
A humanoid robot has the exact opposite. It isn’t bolted down, so its movements have giant variations that need correcting for. As it can’t just run through a basic cyclical program, it must be powered by AI. This means we don’t know what movements it will do, so we can’t put obvious hardware or software checks in place to ensure it is doing things to standard. Moreover, optimising AI can be incredibly hard as AIs have a black box problem. We can’t know how they made a specific decision, so correcting a bad one is damn hard. As such, not only will AI be used to power this humanoid robot, but we will also need to use AI as a check to ensure it is doing the job correctly. But, even the best AIs are only 70% — 80% accurate, and have already hit a performance ceiling (read more here), so this humanoid robot will inherently do a terrible job.
So, specialised robots are often cheaper to buy, always cheaper to set up and operate, and, as long as they are set up correctly, guaranteed to give us far better results than a general-purpose humanoid robot. As such, if you have multiple jobs you want robots to do, it’s easier to use multiple specialised robots rather than a general-purpose one.
So why would you even bother with the general-purpose humanoid robot?!
The only scenario in which humanoid robots make sense is if there is a task in an environment that is specifically designed for humans, where you can’t install a specialised robot and don’t want to just use a human. So, going through a dangerous building that is still structurally sound, that the humanoid robots already know the inside of? But when does that scenario pop up? I have never actually heard a proposed use for a humanoid robot that a specialised robot wouldn’t better serve.
So, when Musk bangs on about his Optimus robot, not only is Optimus so far behind the cruve its like looking at a kids crayon drawring compared to the Mona lise, but the curve itself is utterly moronic.
So, Musk is being doubly stupid. However, that is not all; it also reveals an underlying violence in Musk. Humanoid robots only make sense to those who look at a workforce and see mindless drones, not people, souls, community, humanity or potential. Optimus demonstrates that Musk sees no merit in humankind, in you, in collective effort, in investing in people and community. Instead, he sees people as things to be exploited and consumed. Because if he did value people and humanity, he would make technology that augments and supports us, not do a horrifically terrible job at replacing us. What does that tell you about the future Musk wants to create?
Thanks for reading! Don’t forget to check out my YouTube channel for more from me, or Subscribe to join my monthly film club. Oh, and don’t forget to hit the share button below to get the word out!
Sources: Electrek, The Register, Daily Galaxy, Unitree
Will, I've been reading your Tesla coverage for some months and now their stock is at $350, what gives? They have come back hugely. Why?
That Boston robot is something. Can they teach it to throw a football? Asking for a Patriot.