Seeking Alpha has an article on Optimus, Tesla’s humanoid robot under development. Only one commenter touched on the Macro repercussions of Optimus taking over the labor markets. In a nutshell,
@User 28897925 “The world’s multi-billion population of uneducated laborers (most peple) will never be educated enough to advance their financial lot in life. Entire families, villages and countries and billions of people worldwide depend on tough labor-based income.”
Link to comment:
My reply:
You bring up a key issue that most people miss, when robots take over jobs how do people earn money to live on? Elon Musk has said on several occasions that he favors Universal Basic Income (UBI) which is a government handout which would be paid for by taxing robots. The problem I see with UBI is that the drive to get ahead in life would be stifled, I see more homeless, more couch potatoes, and more sick people.
Some see a Luddite type revolution but the Luddites lost out and will lose again.
Some see robots as frightening but humans have been making labor saving devices at least since they started using stone tools. A robot is just a superior, universal labor saving device.
The real challenge is to find a way to fund humanity so we don’t need to labor to make ends meet but to do so in a way that does not eliminate the need to strive, some level of stress is required for good health. Socialism is a system to defend the working man. Since no one is going to work there won’t be a use case for socialism. Capitalism is a system to defend capital which will be the only economic activity left once Optimus takes over. The logical solution points to making everyone a capitalist! Instead of doling out money by dribs and drabs with UBI, the socialist solution, give every new born a startup capital to fund his lifelong needs, not as an annuity (UBI?) but as an investment portfolio. Of course, some people might choose to buy an annuity or some other ‘financial product’ instead of a managing an investment fund on their own.
That “drive” is often motivated by a fear of being homeless, starving in a ditch. Since retiring, I have had a great time, going places, doing things, and, in the winter, learning things.
I submit that doing what you want to, instead of doing something you hate, so you don’t starve in a ditch, is a better way to live.
I lived at home most of the time I was in college, because it was expedient. I could spend my money on a flashy car and girlfriends instead, which was a lot more amusing than living in a run down old house in the student ghetto, eating spaghetti five days a week.
I did the same. But when people say “grownups” nowadays, they don’t mean “in college” or even “over 18”. They mean, reached young adult status, completed school, and works at some sort of job for a period of time. Call it ages 23-25, and in some cases (like graduate school), ages 26-30.
@User 28897925:
“The world’s multi-billion population of uneducated laborers (most peple) will never be educated enough to advance their financial lot in life.”
Um, this statement was made by someone not thinking very clearly.
Sure, at first the bots will take over manual labor done by “uneducated laborers” (a term I hate. I try to use non-degree, but I’m not sure that’s much better.) I’ve met “uneducated laborers” that have more common sense and smarts than several PhDs I know. But, being the elitist I am, I want a Board Certified doctor to examine my health, not a non-degree cherry picker.
However, rather quickly, the bots will take over the jobs of accountants, lawyers, engineers, artists, teachers, doctors, programmers, etc, etc, etc.
There may be some exceptions. After all, what self-respecting bot would want a job that Herschel Walker is qualified for?
But I digress.
Eventually, we will all reside in endless columns and rows of individual pods dreaming a reality that is not reality so we can supply power to the bots while we await some guy named Neo to show up and free us so we can continue killing each other with endless columns and rows of military grade weapons and cool jiu jitsu.
From an economic perspective, this seems really foolish. The specifics may vary, but a rough estimate of the “return” on your capital investment in building your humanoid robot would be the avoided compensation of the person whose labor you replaced with the robot.
Uneducated laborers have among the lowest compensation levels. Their wages are low, their benefits are low to non-existent. If you’ve got an AI that’s advanced enough to be effectively deployed in a humanoid robot, it seems like an enormous waste to deploy it flipping burgers or doing manicures or picking tomatoes. And certainly to use them in developing markets and countries, where annual compensation for unskilled work is a bare fraction of what we (suffering from Baumol’s cost disease) see in the U.S. We would expect that those robots would be utilized as high up the ‘value’ chain as possible - entirely in the richest possible countries, and in the highest-compensated jobs that the robot can do.
If we saw humanoid robots and they ended up doing manual labor, expect them mostly in unionized positions that come with really good health care and retirement benefits. Western nation fire fighters and longshoremen and the like - not textile workers in third world countries.
But honestly, an AI that’s smart enough to safely and economically drive around a humanoid body is probably too smart to waste putting into a humanoid body. Don’t bother with manual labor - stick them in an “office” job where they don’t need to inhabit a humanoid body.
Agreed but it wasn’t The Captain, it was “@User 28897925” at Seeking Alpha.
Sure, at first the bots will take over manual labor done by “uneducated laborers” (a term I hate. I try to use non-degree, but I’m not sure that’s much better.) I’ve met “uneducated laborers” that have more common sense and smarts than several PhDs I know. But, being the elitist I am, I want a Board Certified doctor to examine my health, not a non-degree cherry picker.
The whole problem goes away if you look at Optimus as a labor saving device that happens to be built as a humanoid robot.
Saying that labor saving devices only take over the jobs of “uneducated laborers” is nonsense. In a household of Ph.D.s you are likely to find a dishwasher taking away the job of whoever does the dishes.
The right way to look at it is how Elon Musk presented it, Optimus will do the jobs that are dangerous, repetitive, and boring
Two comments. One - once the AI is built, each copy will be sold at low marginal cost, much like software is sold. The AI will be installed in what ever hardware the buyer dictates.
Two - Musk says Tesla is targeting “humaniod” because the world is designed for humans. The bot must be able to navigate the real world environment that exists.
All those historical markers, world heritage sites, and such, mandate against tearing the human world down, rebuilding a world that better fits robots, and shoe-horning humans into that design.
But the AI in this case doesn’t do anything until it is encased in a hardware shell, which looks to be pretty expensive. Motors, servos, batteries, cables, pulleys, sensors, video cameras, lidar sensors, and oh yea, an OS. So the OS may be at low marginal cost, but everything else is like building, oh, I don’t know, wild example: a car.
You still need to have a factory producing the hardware. It almost seems that producing the hardware will be the more difficult part of the equation.
That’s what Musk seems to be saying, too?
Paraphrase: “We’ve already got the AGI due to Tesla FSD, Dojo, etc. Leveraging that AGI for a humanoid bot is simple.”
Musk says the TeslaBot will cost $20000 or less.
Youtubers say Tesla is designing all the robot parts in-house. No “off the shelf frankenstein stuff”.
Lots of things can navigate the “real world environment” that don’t have a human form factor. If I need a robot that can restock the shelves at my local grocery store, it doesn’t need to have legs or a torso - or be limited to only two hands, any more than robots that work in factories today have that human form factor.
So if we need to dig a ditch, we use a trencher or an excavator. We’re not going to use a bunch of robots with shovels. And if we have an AI that can be the “operator” for digging that ditch, it doesn’t make a whole lot of sense to keep building a dumb excavator and have a robot drive it - we will put the AI into the excavator. I’m not going to put an AI into a humanoid body to drive a lawnmower around my yard - I’m going to build lawnmowers that have AI in them. Etc.
And obviously, I suspect that a humanoid robot that’s capable of functioning in the “real world environment” is going to actually end up being vastly too expensive to be used digging ditches or mowing lawns…which is why I don’t think that this is really a threat to any but the most expensive manual labor.
Actually, it makes more sense to have a (sort-of?) humanoid shape able to utilize low-cost non-AI machines. It makes no financial sense to put multiple expensive AIs into a number of single-purpose, low-value tools.
Well, I wonder. I was just told that the marginal cost for the AI will be slight, so once you build it - they will come?
I see iRobot running TV ads right now - not selling Roombas, but selling “iRobotOS”. (I wonder who it is they’re selling to.) But, in the spirit of the thread, a Roomba is a single-purpose low value tool, with a rudimentary OS (which they keep improving, the new one won’t run over dog poo! No, seriously, it can finally sniff out dog poo!)
I tend to agree with the post up thread (I’d quote it, but my iPad goes nuts if I try to scroll to a prior post while answering a later one in the thread) that I see limited utility for a humanoid form robot. Not that there’s none, just that so many things can be done so much better by special purpose machines. Warehouse stacking, auto line assembling, ditch digging, dish washing, etc. don’t need all the special doo-dads that full humanoid features would provide - and cost.
Im getting the feeling this is an “either - or” argument?
Either only humanoid or only other form?
I don’t see it being so binary. I see “specialized form factors” in the appropriate use, with “generalized/humanoid” in situations requiring such.
Does Musk envision such a binary choice?
Musk has the “car/truck/semi form factor”, the humanoid TeslaBot form, and … SpaceX “reusable rockets” (the rocket flies itself?).
I really don’t see a teslabot pushing a vacuum. Or riding a lawn mower.
But, I can see the TeslaBot rescuing a stuck roomba or mower?
The primordial ooze does not make money. Human beings using intelligence create economies in everything.
The point is taking inanimate objects and building them is all cost. Running a car is a black hole for money. Building an android is a much bigger black hole for money.
I get our factories have robotic arms. I get OS saves on thousands of clerks on a management floor and puts decision making information quickly on a manager’s desk.
What will an android do? It does not feed my ego. If that is all this dream was supposed to be.
I think so. Part of the pitch for a humanoid robot was that making a single design - rather than bespoke chassis for different use cases - reduced the cost of manufacturing. I think he’s taken that lesson away from all his companies: Tesla, SpaceX, Boring. Get the costs down in order to have a more viable product.
As far as situations requiring a generalized/humanoid form factor - I’m skeptical that there’s really all that many out there, especially if you have a very low marginal cost AI. For most jobs that we might expect a robot to do, the human form is generally okay for doing all of them (with access to tools, of course) - but rarely ideal for any of them. So if you’re building a robot that has to move between jobs at the construction site to the grocery store to the being a waiter at a restaurant, and always using the same tools that humans use at those locations, then a humanoid form is definitely the way to go. But if we build a robot that’s always doing construction - and perhaps always doing the same specific job at construction - there’s almost certainly a non-humanoid form factor that will be more efficient. My grocery store restocking robot probably doesn’t “want” to be in a humanoid format - a large wheeled platform with storage and multiple arms on either side might be better. Of course, that restocking robot can’t also work the cashier - but with a cheap AI, I don’t want or need that. I’ll have a different cashier robot - one that can scan, bag, and accept payments - and it won’t need legs or any form of locomotion (or even necessarily a torso or humanoid head).
I don’t know - it seems like a very weird product, commercially.
I think it depends on the job. If you need a few holes drilled you may want a smart humanoid robot to be able to use a common hand drill to drill a few holes. But if you want lots of precision holes drilled over and over on many identical parts you may want a smart CNC machine that can drill lots of holes, quickly and to within less than 1000th of an inch precision.
Do you want the holes so precise that you need to keep track of tool wear?
CNC machines already do this to and they don’t really have any AI in them.
So what is a humanoid robot going to add to this, cost effectively?
You want a picture hung? Yeah, maybe a humanoid robot could drill a hole and put a hook on the wall and hang the picture. An expensive robot for a simple task.
Somewhere inbetween there are probably some good applications for a humanoid robot.
