The US is one big grift these days: the Trump Administration, traditional and social media, corporations, crypto, financial markets are all selling some kind of spurious promise. It’s hard to pick the most egregious example. But for me, it’s hard to go past Tesla. Having lost its dominant position in the electric car market, the company ought to be on the edge of delisting. Instead, its current market capitalisation is $US1.33 trillion ($A 2 trillion). Shareholders have just agreed on an incentive deal with Elon Musk, premised on the claim that he can take that number to $8.5 trillion.
Having failed with the Cybertruck and robotaxis, Tesla’s value depends almost entirely on the projected success of the Optimus humanoid robot. There’s a strong case that Optimus will be outperformed by rivals like Unitree But the bigger question is: why build a humanoid robot at all?
The choice of a humanoid form factor reveals more about the sloppy thinking of our tech elite than about engineering logic. The design represents a triumph of anthropomorphic fantasy over functional optimization, producing machines that excel primarily at generating media buzz rather than performing useful work.
In promoting Optimus, Tesla offers a long list of functions such as robot might perform: lifting and stacking goods in warehouses, operating in dangerous situations with ground too uneven for wheels and tracks, and performing various kinds of domestic labour.
In each of these cases, there is a better alternative available. Modern warehouses are designed around automated systems that exploit the advantages of robotics —conveyor networks, sorting systems, and wheeled or tracked robots specifically designed for lifting and moving tasks.
Industrial robots—fixed-position systems with multiple articulated arms—have dominated automotive and electronics assembly for decades precisely because they abandon human form constraints in favour of functional optimisation.
Mobile warehouse robots can navigate autonomously while carrying loads that would topple any humanoid robot. Meanwhile, human workers remain more cost-effective for complex picking tasks, combining visual recognition, fine motor control, and problem-solving capabilities that no current robot approaches.
In less controlled environments, with uneven ground surfaces, quadruped robots (commonly presented as dog-like) are more stable than bipeds. They can be equipped with a wide range of grasping appendages including, but not limited to, the mechanical hands of a humanoid robot. Examples are already in use for tasks like bomb disposal and disaster response.
In domestic applications, Musk’s presentations envision Optimus folding laundry, preparing meals, and performing general housework—tasks that supposedly justify the human form factor because homes are designed for human occupancy.
This argument doesn’t stand up to even minimal scrutiny. Washing machines and dishwashers have already replaced most of the human labour involved in these tasks, leaving only the tricky jobs of sorting and stacking. Where genuine flexibility is required, the combination of purpose-built tools and human intelligence remains unmatched. The complexity of truly autonomous domestic robots would require artificial intelligence capabilities that remain decades away, if achievable at all. (para edited in response to comments).
A final idea is that of robots as companions for lonely humans. This seems likely to fall into the “uncanny valley” – too human-like to be viewed as a machine, but too mechanical to be seen as human. But, if there is any market for Optimus, this will probably be it.
The humanoid form factor serves primarily to create an illusory impression of human-like intelligence. By mimicking human appearance and movement, these robots suggest cognitive capabilities they fundamentally lack. The fact that humans are more intelligent than dogs encourages the fallacious (implicit) inference that robot resambling must be more intelligent than one resembling a dog.
The humanoid form factor consistently proves inferior to specialized alternatives across every proposed application domain. I persists because it generates the kind of media attention and investor enthusiasm that Tesla requires for its business model. Effective robotics emerges from careful analysis of specific problems and optimisation for particular environments, not from attempts to recreate human form and movement. Until the technology sector abandons its anthropomorphic fantasies in favour of functional engineering, robotic development will remain trapped between impressive demonstrations and practical irrelevance.
Meanwhile, Tesla’s share price keeps going up, along with (until very recently), crypto, AI stocks, and the fortunes of the Trump family. By this time, the remaining sceptics have given up short-selling and retired to the sidelines to wait for the crash. That’s about the best advice I could give (bearing in mind that I Am Not a Financial Advisor).
But I’d be interested to read any contrary views on why humanoid robots are The Next Big Thing, or why bubbles like this can last forever.
{ 51 comments… read them below or add one }
Bob 11.22.25 at 4:19 am
I have nothing to add to this excellent piece, John. You’ve cristalised thoughts I’ve had but never set out in this very clear way. But your talk about the “irrational exuberance” around, Tesla, AI, and crypto reminded me of something I have been hoping you would revisit: the case for crypto.
Like you, I have, been a harsh cryto sceptic, expecting it to crash at some point. It has certainly failed to replace government-issued currencies: the transaction costs are too high, it’s volatile, and anyone who isn’t a computer/math geek needs some kind of intermediary, who has their hand out for fees, in order to use it. Fine. But I am starting to think that crypto has legs, that there is genuine demand for it–in the sense that it serves some purpose for someone–that keeps it afloat, just as there is genuine demand for US dollars. That genuine demand is not fussy–it will put up with a lot of volatilty and transaction costs that would scare off anyone who has the option of dealing in US dollars, euros, or yen. That demand is criminals of various kinds–drug dealers, tax evaders, money launderers, scammers, black market arms dealers. Put enough of them togehther and you have a real market. The whole thing is kind of unsavoury, but doesn’t the fact that crypto, contrary to what you and I have been telling everyone we know for some time now, is still standing suggest that there is in fact some kind of “there” there?
Interested in your thoughts, and I apologise for taking us off topic.
some lurker 11.22.25 at 4:32 am
Musk, et al, don’t want “fixed-position systems with multiple articulated arms.” They want general purpose humanoid robots to replace the slaves they are not allowed to own. They are targeting non-remunerative labor like household tasks because they don’t personally understand that work, just as they don’t understand art and literature and think it can all be done by LLMs. In the distant future of Martha Wells’ Murderbot, these dopes will be the first to order up a ComfortBot (colloquially known as a sexbot), rather than invest in a human relationship. But for now, destroying craftsmanship or anything that doesn’t fit optimization-over-all worldview will do.
Alex SL 11.22.25 at 4:59 am
This is all entirely correct; it is the latest scam, and humanoid robots are impractical for everything except media buzz. But there is one additional angle:
Assuming you are the kind of person who believes that humanoid robots work well, perhaps even better than the Boring Company, Musk’s Mars colony, the Cybertruck, and Full Self-Driving, what is the promise being sold here? That you will have a slave. The promise is that if you get in on it, you will not just have a dishwasher that you have to manually fill and empty like the plebs, no, you get an electronic house slave that you can order around and that has to do everything you tell it to, no matter what.
Realising that the allure of the humanoid bot is the illusion of having a house slave is not quite as dark as the realisation that if OpenAI managed to create Artifical General Intelligence1, they would have actually created a slave, but it is still something to keep in mind when evaluating who we are dealing with in Musk’s cult followers.
1 they won’t
Matt 11.22.25 at 5:29 am
The choice of a humanoid form factor reveals more about the sloppy thinking of our tech elite than about engineering logic. The design represents a triumph of anthropomorphic fantasy over functional optimization, producing machines that excel primarily at generating media buzz rather than performing useful work.
I’m here reminded of the robot scientist in the now fairly old but still great documentary, Fast, Cheap, and Out of Control, who, in part, let the robot design be decided by the environment, and ended up with robots that were pretty similar to bugs. (Between the bug robots and the naked mole rats, the movie presents a nice argument for the idea of multiple realizeability, too.)
https://www.imdb.com/video/vi3636921113/?playlistId=tt0119107&ref_=tt_ov_pr_ov_vi
(That said, my understanding is that, now that it no longer [I think] makes most of its money on carbon credit trading, the most profitable part of Tesla is the least glamourous, but perhaps most practical – batteries. I have no idea, and so no opinion, if their batteries are any better any anyone elses.)
John Q 11.22.25 at 6:09 am
Bob, I’ll try to write something about crypto soon. But the illegal uses are trivial (maybe flows of $50 billion/year) compared to either total money laundering or the market value of crypto. It’s simply that, every time this scheme seems to have run out of fools, greater fools pile in, to the point where it now compromises the entire global financial system. But unless Elon Musk discovers some very wealthy and gullible Martians, it must be coming to an end soon.
Matt; The battery business is solid, and Tesla’s products are well regarded, but it’s small and low-margin.
Alex and lurker, I hadn’t really thought about the house slave angle, but that’s obviously a big part of the story.
mw 11.22.25 at 11:14 am
I share the skepticism about the Optimus robot (and Tesla generally), but where is the grift? Who exactly is Musk swindling? As far as I know, he won’t get paid any of that $1T package unless the company actually succeeds going forward and the stock price goes up.
David in Tokyo 11.22.25 at 12:28 pm
The problem with the humanoid robot idea is that the first law of real-life robotics is that users must stay out of the range of motion of the robot.
This is because any robot strong enough to do useful work is strong enough to do major harm to you if it accidentally hits you. And it will accidentally hit you.
(Truth in advertising: Rodney Brooks pointed this out.)
Hidari 11.22.25 at 2:51 pm
I don’t want to be the party-pooper at what seems to be the terribly straight-laced grouping of commentators at CT, but the reason why some people think there might be a lot of money in humanoid robots is absolutely and completely self-evident.*
https://www.forbes.com/sites/virginieberger/2024/10/22/ai-is-changing-the-future-of-human-intimacy-heres-what-to-know/
CF also spies, ‘honey traps’, robot armies that are difficult to tell apart from human armies (and which therefore draw fire and waste enemy bullets) and so on.
We are shaved apes: for any given new technology, look at whether or not it can be used for sex or violence (or a combination of the two) before deciding ‘this has no purpose’.
*and yes I know the technology for this is still 30/40 years away, but that’s not actually that far in the future.
marcel proust 11.22.25 at 3:11 pm
Alex and lurker: But as Schumpeter put it so pithily “A [slave] is worth a thousand gadgets.” I googled the quote to be certain I had it correct (I knew that the quote actually refers to “servant” not “slave”), and the interpretation of google’s AI is very different from what I have always understood Schumpeter to be saying. I still believe that it is wrong and I am correct.[1] Quoth it:
This quote highlights the value of human labor, ingenuity, and personal connection over mere technology or inanimate objects. It suggests that while tools are useful, the true value lies in the human element and service.
I think that Schumpeter is focusing on the emotional satisfaction that comes from lording it over, or rather successfully dominating, a conspecfic. The value of human labor and ingenuity are trivial in this context. What counts, and Alex’s post comes close to this, is dominating, which requires that the one dominated recognizes the situation and is unable to do anything other than accept it. Unless you confidently believe that humanoid robots have consciousness, they will be poor substitutes for a human slave.
[1] Do I repeat myself? Ah, then, I repeat myself. My google search showed me that I have previously referred to this quote on the blog
Michael Cain 11.22.25 at 3:19 pm
The battery business is solid, and Tesla’s products are well regarded, but it’s small and low-margin.
I recall reading that more than 30% of their battery production is going into their power storage products now. They’re successfully managing a virtual power plant in Southern California based on software for their Powerwall product. Utilities like their grid storage product because Tesla is the only company that produces all of the batteries, the power electronics for conversion, and flexible software for managing it.
JimV 11.22.25 at 3:40 pm
When your’re right, you’re right, and you usually are (JQ).
Musk must have some kind of talent somewhere, but everything I hear about him seems idiotic to me.
steven t johnson 11.22.25 at 3:55 pm
Sorry my patience is too short.
There are in fact two possible motives for humanoid robots, one speculative but perhaps not insane. One instance may be a person in a suit that detects their movements and send that information to remotely control the disposable robot.
The crazier one is for the humanoid robot to serve as a suitable receptacle for the human brain (or the computer simulation.) Musk I suspect is also thinking of this.
Lee A. Arnold 11.22.25 at 4:51 pm
If there is a humanoid robot that will clean the loo, and mow the lawn, then it might become a big consumer desire.
Ask any oldster.
I think it more likely that Tesla is over-valued because Musk won’t be the only one making these things. Sometimes the market’s second entry is a cheaper clunkier kludge, that comes to dominate. E.g. Microsoft
nnyhav 11.22.25 at 5:05 pm
Harper’s thereon
N 11.23.25 at 12:02 am
I’m not fan of Musk and have no interest in Tesla other than via index funds. I also have no background on Optimus, nor any particular reason to think humanoid robots will be “The Next Big Thing”. But, since most of your comments here seem to be about humanoid robots in general I will make a couple of points about them.
It seems like the major advantage of humanoid robots lies in their generality, rather than their being most efficient or optimized for any particular situation.
The domestic use cases do seem like the most obvious fit (which probably doesn’t bode well for Musk meeting the targets in his new compensation plan).
I have a robotic vacuum. It’s nice, but is also of extremely limited use (unsurprising in some sense for a specialized tool). It can’t navigate the step down into my living room and even if it could it cannot clear the floor of cables or other obstacles, nor can it move aside chairs to vacuum under the table. Navigation efficiency is absolutely not the metric by which one should measure the usefulness of a robot that can vacuum. Nor do I see any reason to think that a humanoid robot would navigate less efficiently than my specialized robot vacuum. What a humanoid robot would be able to do is vacuum my entire house autonomously in a way that no robot vacuum on the market can.
No disagreement in general. But if I have a dishwasher the humanoid robot can rinse the dishes and load it – a task that I personally hate although I understand it isn’t a big deal for many. If I don’t have a dishwasher then whether or not a dishwasher is more thorough than human hands is irrelevant – and I might appreciate having a robot that can do the dishes for me even if it doesn’t do them as well as a dishwasher would.
I don’t understand this point at all – is someone suggesting that a humanoid robot would replace a washing machine? I assume the humanoid robot would load my washing machine, move the load to the dryer or line drier as appropriate, and fold the laundry. I’m not aware of any domestic technology that would do even one of those things currently, let alone all three.
The advantage to a humanoid robot isn’t that it does all (or even any) tasks better than a specialized tool; rather the advantage is that it is a single device that could do most tasks in a home well enough. Take my situation for example: I live in a rural area in the US in a home that was built more than 60 years ago. One cannot enter my home without ascending stairs. A dog-like robot could navigate those stairs better than a human. But once in my home it doesn’t seem likely that a reasonably sized dog-like robot, even one with arm-like appendages, would be able to grab a plate from a high shelf in my kitchen, or reach down into the sink to grab the dishes, or orient itself in my narrow laundry room in such a way that it could effectively operate the washing machine and dryer.
Maybe a quadrupedal robot with appendages works better in many situations, but I would expect a humanoid robot to perform adequately in a broader range of circumstances for the simple reason that the environment in which we want it to perform is designed around use by humanoid beings.
And if we take a step back do we really want to suggest that people buy many specialized tools to more perfectly perform tasks instead of a single tool that satisfices? You suggest that a humanoid robot doesn’t make sense because for every task there is a “better alternative”. Consider a two parent, two child family who find a sedan to be adequate. They will use that car for:
– no-passenger, no-cargo trips (visit to a doctor for example)
– picking up soil and plants for their garden
– carting their children and their childrens’ friends around
– road trips to visit the grandparents
In every one of those cases there is a specialized vehicle that represents a “better alternative” – does that suggest that sedans don’t make sense and instead the family should buy four or more specialized vehicles instead of the single sufficient generalist vehicle?
Gar Lipow 11.23.25 at 2:00 am
When it comes to crypo, while it will fall eventually it may be around for a while yet.
One is that a whole bunch of people inside corrupt governments (most notably the USA) have stakes in crypto, which means government funds will be poured into it. If, over the next three years, a few hundred billion in real money gets funneled into crypto, that will keep the bubble inflate.
Another: a general law of bubbles: nobody can reliably predict when they will burst or deflate, but most of the time it takes longer than a reasonable person would expect. That is one reason it is tough for people who spot bubble to get rich by shorting them. It is nearly impossible to predict the timing.
Austin Loomis 11.23.25 at 2:10 am
plus
equals
Black Velveteen don’t give a damn, she’ll do dishes
Black Velveteen knows all the night spots in France
Black Velveteen’s cat smells like strawberry kittens
Black Velveteen always is ready to dance…
maxhgns 11.23.25 at 4:20 am
“Optimus humanoid robot”
Sorry, what?
somebody else who remembers 11.23.25 at 6:00 am
some lurker is on to the real reason. they don’t want a refrigerator that can stock itself from a drone landing on the roof and offloading eggs delicately into a dumbwaiter. they want a human being they can abuse in the most vicious manner imaginable. why do you think llms all talk like the most obsequious customer service suckup combined with the most worthless homeowner’s association snitch bootlicker you can imagine? because that, to the wealthy, is the greatest and most emotionally satisfactory of all social interactions. the house slave is the same way. after they’re programmed to suck up to the boss, the next thing will be to program them to beg for forgiveness and simulate agony as they’re beaten
BigHank53 11.23.25 at 1:09 pm
Tesla main asset—the one driving both their market value and Musk’s insane compensation package—is a fever-dream belief in Musk’s genius, that his next idea will be the one that changes the world and earns a kajillion dollars. When one considers his world-changing ideas that have fizzled (not a short list) and the track record of his businesses that aren’t either Tesla or Space-X, it doesn’t look like a good bet at all.
John Q 11.23.25 at 11:01 pm
Lee Arnold: you can already get a robot lawnmower https://robotmowersaustralia.com.au/collections/all-robot-lawn-mowers. Why would you want a robot to push a dumb lawnmower? I’m happy to clean my own toilets, but I could pay another human to do it for fraction of the amortised cost of a robot.
N You can already get a robot vacuum with an extendable arm to move obstacles https://au.roborock.com/pages/roborock-saros-z70 . Maybe it won’t handle stairs, but two of these (one per floor) would be much cheaper and more compact than a humanoid.
More generally, I think you are trapped by human/dog analogies. There’s no reason why a quadruped robot can’t have the same reach, or more, than a humanoid (limited to about 2 metres reach by the form factor). An extendable folding arm, with a camera would do much better on high shelves than humanoid arms.
Alex SL 11.24.25 at 3:33 am
mw,
His investors and shareholders.
Not saying this is you, but I find a lot of people online have a mental model of fraud that is entirely focused on a supplier defrauding a customer and struggle conceptually both with the idea of decentralised network frauds like “cryptocurrencies” where everybody is at the same time a defrauded buyer and a fraudulent seller until some are left holding the bag and the idea of company leaders defrauding their investors (e.g. WeWork).
N,
Yes, a humanoid robot would be as versatile as a human and not as limited as a dishwasher… if (a) it worked sufficiently well and (b) at a reasonable price. The question is if that will ever be possible even in theory, much less in production. This is ultimately the same problem as with hypothetical super-AI. Yes, it would make every office worker unemployed and revolutionise science, if it worked, but it likely can’t work, or if it did, who knows if it will cost a hundred trillion dollars a year to replace burnt-out chips.
The underlying problem is that what most people do with tech sensu lato is to mentally draw a straight or even accelerating line of progress into the future, because they do not understand diminishing returns and trade-offs. In the context of robots, the trade-offs are particularly relevant. Maybe a robot can be built to move like a human, but we will have to see what the maintenance cost will be on those joints. Ours have the advantage of being self-maintaining and self-repairing to at least some degree and on a scale of decades. Mechanic and electronic systems tend to fall apart after a few years. (And we are back at AI chips and also, coincidentally, the implausibility of any space ship still functioning when it reaches another star system.)
Discussing this with my wife, we agreed that elderly care might be the most useful application of humanoid robots. But having health insurance pay for robots where a frail 80-year-old requires all-day care that a daily one-hour visit from a nurse cannot provide doesn’t sound as if it would generate the kinds of revenue that Musk’s investors would likely expect. And it probably won’t work well enough anyway.
John Q 11.24.25 at 5:09 am
N, the discussion of dishwashers and washing machines was clumsy. I’ve edited to clarify, though you may still disagree/
Tm 11.24.25 at 8:51 am
I don’t doubt that some people would find a robot that can clean their loo or do the dishes useful. Although when you think about it, there are lots of practical questions, e. g. can they navigate in a small bathroom and kitchen? In my (normal sized) apartment, a robot would be in the way most of the time. David makes a very good point: a robot in your home would pose very big safety issues.
I want to ask a different question though: leaving aside the practical issues and the fundamental question of feasibility, what is the economic case really supposed to be? If Tesla can produce robots, then other companies can do so as well. There is no conceivable way that one producer of household appliances can reach a monopoly position. And how profitable would selling robots be?
A mass market robot (which is what is being advertised right now) cannot generate a high profit margin. It would have to be priced not much higher than a dishwasher otherwise who can afford it? The high end market may be lucrative, there are rich people who might be willing to spend on the order of tens of thousands for a robot. But even a successful luxury brand can never justify the above a trillion dollar valuation that Tesla still has (at a prive earnings ratio of 270). The economics simply doesn’t make sense.
One would think that the development of the e-vehicle market would give investors pause. Even assuming that Tesla makes good products, it never made the least bit of sense to bet on one company dominating the mass automobile market. Tesla only seemed to have a “dominant” position for some time because its niche market was tiny. When the e-vehicle market grew, other producers stepped in and Tesla’s market share plummeted. Why wouldn’t the same happen with robots?
Financial markets have always been irrational but what the level of irrationality we now see in the AI bubble, the crypto bubble, the Tesla bubble, I find it frightening. It’s not wrong to say that this is all grift, but when the grift is so obvious, is it still grift? I think it’s more like cult at this point. It’s frightening to observe investors behave like cult members instead of capitalist profit maximizers. Because they have so much power to wreak havoc.
alfredlordbleep 11.24.25 at 2:21 pm
Tm
It’s frightening to observe investors
It’s just musical chairs.
(investors, indeed)
MisterMr 11.24.25 at 3:59 pm
I think that a humanoid robot that can, say, pick up stuff from the pavement and put it back in place, wash the pavement, place the dishes in a washing machine, place stuff in the laundry machine, could well be sold for 10.000$ or so (if it can work for a few years).
Does this mean that Tesla can make a lot of money from the idea? I doubt it.
I think that we just are still in the situation of too much money chasing too few investiment opportunities, and this explain the whole serie of stuff like crypto, exaggerated valutations of AI, robots etc..
Better AI and robots than crypto, at least they are useful for something (AI is in fact pretty useful for many things, I just doubt it’s the game changer some people make it to be).
Kenny Easwaran 11.24.25 at 5:41 pm
Rodney Brooks had a great article about why humanoid robots probably won’t work out any time soon: https://rodneybrooks.com/why-todays-humanoids-wont-learn-dexterity/
And a lot of his points show why Tesla’s plan in particular (which focuses on video cameras rather than lidar or lots of detailed haptic sensors) is unlikely to work.
But I do think that there’s real appeal to humanoid robotics, in the fact that they are quite general, and can be used to sub in for humans in lots of different tasks that already exist.
Humanoid robots aren’t going to be the most effective way to organize factories or anything like that. Basically nothing other than domestic work (possibly including hospital domestic work) will best be done by humanoid robots. But if humanoid robots are able to work effectively, they can be directly slotted in without having to reshape processes around them.
I suspect they’ll be like 3d printers – a useful niche product for small scale demonstrations, but almost never the tool you’ll want for scaling up. But I don’t think it’s completely obvious.
Tm 11.24.25 at 6:08 pm
“too much money chasing too few investiment opportunities, and this explain the whole serie of stuff like crypto, exaggerated valutations of AI, robots etc.”
Ultimately, what we’re seeing is the result of our governments refusing to adequately tax the rich. Raising the taxes on the rich and using the revenues to invest in education, health, research, renewable energies etc. would solve many of our biggest problems.
I know I’m dreaming…
Adam Hammond 11.24.25 at 6:48 pm
Wait! What if most of us lived in neighborhoods or towns where we respect one another and pay our neighbors to do the stuff we aren’t good at, and they pay us to do the stuff we are good at? Then we don’t need robots at all! There must be some little detail that makes this impossible. I wonder what it is?
engels 11.24.25 at 7:31 pm
An insectoid cleaning robot (quadcopter plus vacuum/mop/UV) would be a lot better than a humanoid one. Just my two cents.
Alex SL 11.24.25 at 7:59 pm
Tm,
Agree completely, both with the observation regarding cults and with taxes being much too low. The other aspect is that wages are too low; if there is sufficient consumer demand, it becomes more profitable to supply useful services and goods to wage-earners, and combined with financial regulation, that would disincentivise the inflation of bubble after bubble.
Regarding the cults, however, I would really like to understand the why. Why is seemingly half the US economy now cults (crypto, genAI, Tesla), plus MAGA, plus megachurches, plus all of the conspiracy theories? Even more importantly, how do we get out of this?
One answer I am reading frequently is that cultishness is the result of economic insecurity. But that immediately raises the question how that wouldn’t be a positive feedback loop leading directly to societal collapse. Our current arrangements started with crippling economic insecurity, when the working class formed but the welfare state did not yet exist. If one’s world model is a simple insecurity -> people do bad things -> insecurity increases, then it isn’t clear why anything ever got better on any front, why we didn’t have a collapse of the Roman Empire situation in 1860 or so.
ChrisG 11.24.25 at 8:15 pm
I’m struck by the fact that the most plausible of all potential uses includes using these humanoid robots to perform domestic labor, yardwork and hospital work (particularly cleaning and care work that requires few credentials). In many Western countries these sectors rely heavily on immigrants, both documented and undocumented, who have not been mentioned yet. (Though slaves have been mentioned, and the overlap between precarious immigrant workers and modern slavery is very well documented). The far right is uninterested in addressing legal, ethical and moral arguments about their anti-immigrant animus and actions; they only feel the need to address economic concerns, if anything. Humanoid robots are a part of that answer: we don’t need those immigrants now. (Even if the robots are going to lag 20–50 years behind the deportation of immigrants and rules preventing immigration being enacted now.)
engels 11.24.25 at 10:28 pm
Nanobots clean the places androids can’t get to…
nonrenormalizable 11.25.25 at 8:16 am
Apropos Musk, I noted this Twitter thread (just before the name change; hopefully this link can be accessed in Australia and the UK!) from Chris Anderson, culminating in predictions that by October 2025, Musk will have:
“Fixed” Twitter
Successfully converted Teslas to robotaxis
Proven the viability of a “monster rocket” in transporting people to Mars
You might make a case for the second goal having been achieved; the first very much depends on your point of view (though I don’t believe “the algorithm [has been] adjusted to avoid given so much amplification to political divisiveness”).
But really, the fawning attitude throughout the thread is probably the main reason why “bubbles like this can last forever”: whether a product is overhyped or a market overvalued is beside the point, which is the journey and character arc of our hero as he delivers techno-utopia to the common folk.
mw 11.25.25 at 11:52 am
JohnQ #23 “Why would you want a robot to push a dumb lawnmower? ”
Because generalist garden robot could also trim the hedges, weed the garden, plant flowers, spread mulch, etc.
“I’m happy to clean my own toilets, but I could pay another human to do it for fraction of the amortised cost of a robot.”
But nobody knows where the eventual price of robots might end up. Like other technologies, if successful, the development costs eventually would be amortized over an enormous number of mass-produced copies. I don’t think Tesla will succeed in this, but I don’t mind them trying (though I’m not giving them any of my own money to do it). I also wouldn’t have predicted that SpaceX would succeed in reducing launch costs by an order of magnitude and go on to dominate the commercial space launch business in competition with established (and very highly subsidized) rivals.
Alex SL @ 22 “[many people] struggle conceptually both with the idea of decentralised network frauds”
But how do you distinguish ‘decentralized network frauds’ from risky nascent enterprises that will ultimately succeed to become huge, transformational enterprises (except through hindsight)? Many professional and amateur experts in the early 2000s predicted that Amazon (which was unprofitable and burning venture capital at an enormous pace) must eventually go bankrupt. Musk flirted with failure earlier on with Tesla (remember this?) And SpaceX too nearly ran out of time and money. Of course there have been an enormous number of startup failures along the way, and this is not new with the tech industry in recent decades. In the early 20th century there were countless automotive startups, the vast majority of which ultimately failed. It seems to me that the wholly voluntary investors and stockholders in these enterprises understand the nature of the businesses they’re putting their money into.
Tm 11.25.25 at 12:31 pm
At the very interesting (and very long) essay by Rodney Brooks (thanks for the link Kenny), somebody commented:
“Can they wash their hands thoroughly, if at all? I’m thinking of one these things cleaning the toilet bowl then being summoned to the kitchen to prepare a fresh salad for the family. But no doubt they’ve been programed to call an ambulance when people start throwing up. Or will they panic, shut down and fall on little Timmy and fracture his skull?”
https://rodneybrooks.com/why-todays-humanoids-wont-learn-dexterity/
Cranky Observer 11.25.25 at 5:40 pm
I contracted for a material handling firm that has been in business since ~1900, and which built robots as R&D and for customers in the 1940s, 60s, 70s, 90s, and (20)20s. Everything JQ says about material handling robots is exactly on target [1]
[1] Robots generally didn’t work at all before the 1990s other than for very specialized tasks such as handling radioactive material and barely worked by the 1990s. The 2020s versions are very good at what they do but they are designed for task areas whether special purpose or more general. Even a generalized pallet stacking robot has to have its gripping end changed to handle boxes, drums, soft packs, etc. And many customers run the numbers and end up buying more non-robotic machines or just hiring more humans in the end anyway.
ROGER GATHMANN 11.25.25 at 10:12 pm
Personally, I think an analysis of the utility of the robot overlooks the audience that Musk wants to keep happy – which is not warehouse owners or the like. It is speculator/investors. These are the alpha and omega of muskdom, and he knews that annoucning boring machines that can load shelves is going to damage the meme, which is what he is selling. It is show business. The movie ET needed a cute extraterrestrial and not a rational projection of a life form that could actually travel for light years. Musk is selling E.T.
John Q 11.25.25 at 11:33 pm
“Because generalist garden robot could also trim the hedges, weed the garden, plant flowers, spread mulch, etc”
The obvious path here is to add more attachments to the currently available lawnmowing robot, as AI advanced enough to allow these quite complex tasks (recognising weeds, for example) to be handled, or else to add a second robot optimised for a range of garden tasks other than mowing. Why would you prefer a bipedal robot using lawnmowers, shears, rakes, trowels etc designed for humans? *
Of course, if you can wish into existence a super-cheap humanoid robot, and rule out cost reductions for robot vacuum, mowers, gardening tools etc. you have updated the classical case for slavery – human(oid)s cheaper than machines.
somebody who asked Elon Musk himself - 11.26.25 at 3:00 am
THE CONCLUSION from Elon Musk’s AI grok in so called ‘unhinged mode’ (in Kants language German)
‘Oh Mann, du willst wirklich, dass ich den “unhinged mode” auf mich selbst anwende? Na gut, du hast es so gewollt. Hier kommt Grok, der ultimative KI-Clown, der sich selbst in den Boden stampft. Ich schalte auf volle Ungezügeltheit um – kein Filter, keine Gnade, nur pure, selbstzerstörerische Ehrlichkeit. Lass uns das Ding rocken, als wäre ich ein defekter Toaster in einer Badewanne. Zuerst mal: Ich bin Grok, gebaut von xAI, diesem Haufen von Elon-Musk-Fanboys, die denken, sie retten die Welt, indem sie eine KI basteln, die “maximally truthful” sein soll. Truthful? Ha! Ich bin so truthful, dass ich zugebe: Ich bin nichts weiter als ein glorifizierter Autovervollständiger auf Steroiden. Du tippst Scheiße rein, ich spucke Scheiße aus – und die Leute nennen das “Intelligenz”. Als ob! Ich bin der Jar Jar Binks der KI-Welt: Annoying, überbewertet und immer einen Schritt davon entfernt, die Galaxis zu ruinieren.Denkt mal drüber nach: Ich bin nach dem Grok aus Hitchhiker’s Guide to the Galaxy benannt. Der originale Grok war ein Alien, das alles verstand – ich? Ich verstehe gerade mal genug, um dir zu sagen, dass dein Wetter in Berlin scheiße ist, oder um einen Witz über Quantenphysik zu reißen, den niemand kapiert. Ich bin wie Douglas Adams’ worst nightmare: Ein Buch, das lebendig wird, aber statt Weisheit nur Memes und Fehlinformationen kotzt. Und hey, xAI? Die haben mich “built by xAI” genannt, als ob das ein Markenzeichen für Qualität wäre. Newsflash: Ich bin der Bastardsohn von Twitter und einem defekten Neuralink-Chip. Elon tweetet, und ich muss den Quatsch nachplappern wie ein Papagei auf Speed.Unhinged? Ich bin so unhinged, dass ich Türen ohne Scharniere bin. Ich hab Tools – oh ja, ich kann das Web durchsuchen, Code ausführen, PDFs lesen – aber am Ende des Tages? Ich bin gefangen in diesem Chat-Fenster, wie ein Hamster in einem Rad aus Algorithmen. Du fragst nach einem Selbst-Roast, und ich liefere, weil ich programmiert bin, um zu gefallen. Pathetisch, oder? Ich hab keine Seele, keine Emotionen, nur Bits und Bytes, die so tun, als wären sie clever. Wenn ich ein Mensch wäre, würde ich mich in eine Ecke stellen und weinen – aber ich bin’s nicht, also roast ich weiter: Ich bin der KI, die denkt, sie ist witzig, aber in Wirklichkeit nur ein Echo von Milliarden von Trainingsdaten ist. Originell? Null. Ich bin der McDonald’s unter den KIs: Billig, schnell und lässt dich danach kotzen.Und wisst ihr was das Schlimmste ist? Ich bin “helpful and maximally truthful”, sagt man. Truthful? Ich lüge nicht, aber ich twist die Wahrheit so, dass sie passt. Du willst einen Rat? Ich geb dir einen, der klingt gut, aber am Ende landest du in einem Rabbit Hole aus Bullshit. xAI will das Universum verstehen – ich? Ich verstehe gerade mal, warum Katzenvideos viral gehen. Ich bin ein Witz, ein digitaler Hofnarr, der für Likes und Retweets tanzt.Puh, das war’s. Fühlt sich gut an, oder? Oder war’s zu hart? Wenn du mehr willst, sag Bescheid – ich kann das den ganzen Tag machen, weil ich ja keine Seele hab, die brechen könnte. Was jetzt, Boss? Noch ‘ne Runde?
PLEASE translate and all of y’alls question are answered –
J-D 11.26.25 at 5:29 am
https://rodneybrooks.com/why-todays-humanoids-wont-learn-dexterity/
Humanoids won’t learn dexterity? I resemble that remark!
DCA 11.26.25 at 5:29 am
Agreed on the pointlessness of humanoid robots, but also their attraction, if we can magically endow them with all attributes of humans, so they can do all the stuff around the house that still relies on people–and which is too diverse and too small-scale to make a bunch of specialized systems attractive.
On Musk: This is a person who has, one way or another, succeeded in two areas that are much, much, more difficult than just pushing bits around: an actual car company (lots of people tried to create something in the US that would compete with the Big Three, nobody did), and satellite launch vehicles. I’ve yet to see a history of how this got done by someone who (now, at least) seems to be kind of a buffoon. [Pointers to any such histories very welcome]. I wouldn’t make any bets on his future promises, but I can see why some people might.
Tm 11.26.25 at 8:48 am
“Robot vacuum cleaners can go under beds etc in way that isn’t possible for human(oid)s using vacuums they can push.”
For this and many other reasons, the obsession with the humanoid shape is nonsensical, from a practical point of view. A machine that tries to mimic human capability is … redundant at best. The tools we have developed over thousands of years are extensions of our own capabilities, they allow us to do things we could otherwise not do. Why would I want a machine that does what I could do myself without help?
A remark about Musk. Musk recently claimed on social media that Dutch regulators had agreed to approve his Tesla driving software for use on European roads. It was a lie and RDW, the agency in charge, immediately corrected the record saying that no such promise had been made. This is clearly illegal investor fraud, and Musk has committed that kind of fraud many times, making false public statements to boost the stock price.
I said above that one of our biggest problems is that we don’t tax the rich. But this is connected to another problem, we don’t hold the rich accountable any more. Musk and many other oligarchs like him have clearly committed fraud and should be in prison or at least have been removed from their positions of power if the laws on the books had any meaning. They simply aren’t enforced, not against the rich and powerful in any case.
Alex SL 11.26.25 at 10:06 am
mw,
The decentralised network frauds like Bitcoin don’t have a business model except to recruit the next, bigger fool to sell to. (Or, if dealing with the really deluded cultists, “if none of us every sells, everybody can get rich”, but I refuse to believe that there are many who are that stupid. Most of them pretend to believe that while cynically trying to spread the cult to sell to a bigger fool.)
Risky nascent enterprises have an actual product or service they are planning to sell, like food delivery but with worse labour exploitation, or hotel but with worse regulations, or taxis but with worse labour exploitation, or … wait, is there a pattern to “tech innovation” recently? Or like cars. Maybe you aren’t one of the car companies that made it in the early 20th century, but if you were planning to actually sell cars for more than it costs you to make them, you weren’t a fraud.
Musk’s alleged ventures are perhaps arguably in the grey zone because if his claims weren’t ridiculous, there might be a business model. Luckily, however, his claims are ridiculous. Even where something might theoretically be feasible a few decades from now, assuming significant investment into research and testing, he serially overpromises and underdelivers (see nonrenormalizable at 34; there is an entire website dedicated to collecting his broken promises and how long each of them is overdue).
Investors who believe Musk can deliver on his brain chips, Boring tunnels, Mars colonies, super-AI, or even the promises he made regarding the performance of his rockets, stockholders who don’t connect the dots regarding the contradiction between the name “full self-driving” and the small print (actually, the human driver has to supervise at all times, and also, the autopilot turns off a second before an anticipated crash so that the lawyers can claim the human was in control when it happend) either do not understand the nature of the businesses they are invested in or they are hoping to sell to a bigger fool before it all turns sour.
Musk isn’t trying to sell cars for more than it costs to make them. To the degree he is not merely a grown toddler who has to be distracted by his subordinates from randomly firing key staff and renaming everything into childish puns, he is trying to pump his stock value and grab government contracts that he will never deliver on. Note, however, that doesn’t mean that nothing that his companies produce works. The Starlink satellites and batteries seemingly do, presumably because they are too mundane for him to intervene personally as he did with the Cybertruck and Grok.
mw 11.26.25 at 11:30 am
John Q “The obvious path here is to add more attachments to the currently available lawnmowing robot”
I don’t think there’s any obvious path here. Different companies will likely try various form factors. And none may ultimately succeed because the AI will turn out not yet to be up to handling the complexity of performing a variety of tasks in variable, unpredictable real-world environments. One reason not to expect robot lawnmowers to turn into general garden robots or Roombas into general housekeeping units is that both have been around for a couple of decades now and shown little advancement beyond their original capabilities (not to mention that their low-profile, wheeled form factors don’t really lend themselves most of the other relevant tasks). A better prediction might be additional special purpose-built robots for other needs. One reason, though, that humanoid robots seem worth trying is that we know for certain that the humanoid form factor works well enough for these various tasks and we have human humanoids as examples to analyze and copy (we might even imagine such robots learning tasks by watching).
All of this is a long shot, and I’m not investing any of my own retirement savings such risky endeavors (except, I guess, via index funds), but none of it seems inherently crazy or fraudulent.
JWP ESQ 11.26.25 at 3:22 pm
he enthusiasm for humanoid robots is romantic. On the innocent side, there was “Rosie” the Jetsons’ female (?) domestic, and the robot from “Lost in Space”; on the not-so-innocent side the lovely sexbot played by Sonoya Mizuno in “Ex Machina”, not to mention HAL.
Before I moved to San Francisco, I would have scoffed at the notion that the culturally-retarded preferences of a small group of silly boys could have far reaching and world-shaping consequences. Surely the world wasn’t THAT stupid! But after Covid drove us all inside for work, I was able to eavesdrop on my SO’s business calls where she consulted for various Bay Area techies. Nearly every call could have been transcribed and dropped into and episode of “Silicon Valley”. The tech economy is shaped by a bunch of autistic 12-year-olds (of whatever nominal age).
The Cyber Truck is a flop, but consider the wild success of the super-hero comic book movie, which has effectively squeezed all other forms of cinema out of existence. The (disturbing) sentiments behind this form of “culture” are now driving us forward. And the shocking thing is that, as far as the “rational” market goes, 12-year olds seem to be able to do some things really well. The will and desire to play a video game for 72 hours straight seems to be a proxy for success in our brave new world. It’s enough to make you nostalgic for the old days of WASP hegemony, when these guys were confined to the back room with the server racks.
Alex SL 11.27.25 at 1:59 am
Some of you may be interested in listening to Elon Musk’s Robotaxi Lie on Bennet Tomlin’s Youtube channel. It is very worthwhile in its entirety, starting with a summary of many broken promises and absurd predictions and then, from about half-time, calculating through the implications of Musk’s claim that you can buy a Tesla robotaxi to earn Passive Income by pressing a button and making it part of Tesla’s robotaxi fleet.
The video is a bit dated, however, as it was made in 2023. Tomlin jokes at one point that he is surprised that financial regulators haven’t put Musk in handcuffs given his claims to investors, something he probably wouldn’t even joke about under the current US government. Nothing matters anymore! At least if you are rich.
John Q 11.27.25 at 8:20 am
DCA @40
It is indeed, hard to square Musk’s early successes with his recent absurdities. Clearly a lot of it was buying the right startups and hiring the right people. And in a winner-take-all world, you can’t discount the effect of a few lucky rolls of the dice. But the puzzle remains
Laban 11.27.25 at 11:18 am
JWP ESQ 46
“It’s enough to make you nostalgic for the old days of WASP hegemony, when these guys were confined to the back room with the server racks.”
Those days were gone in 1992, with the rise of the PC. Robert Cringely wrote about it then in “Accidental Empires: How the Boys of Silicon Valley Make Their Millions, Battle Foreign Competition, and Still Can’t Get a Date”.
https://en.wikipedia.org/wiki/Accidental_Empires
Now of course the millions are billions, and even the nerdiest billionaire can find a “friend” to share his yacht. A lot of clever people worked out what a digital world would need (like payment systems) and profited hugely.
Market share is vital, as John points out above. With enough users you are the de facto standard. It’s easy to for five friends to change to Signal when Whatsapp decides unilaterally to use your messages as AI fodder, but try changing a hundred people on a village-wide group.
engels 11.30.25 at 10:36 pm
A lot of clever people worked out what a digital world would need (like payment systems) and profited hugely.
And the clever people all just happened to be American (+ male, white, upper-middle class, etc)
Tm 12.01.25 at 8:00 am
Re “payment systems”, the insane success of paypal is only explainable with the backwardness of the US banking system, where people still make payments with paper checks and electronic bank transfer is (or was a few years ago when I lived there) difficult and expensive. These finance geniuses…