America is getting an AI gold rush instead of a factory boom
(washingtonpost.com)64 points by voxleone 3 hours ago
64 points by voxleone 3 hours ago
> Most people in Western world also cannot imagine staying at conveyor belt or table doing the same assembly task whole week. I work in a factory and see this daily.
My family owns a small plastic manufacturing plant in the US. This is the biggest problem they face. The western worker's appetite for a low skill monotonous manufacturing job is very small. The business loses employees to Walmart etc. despite the factory having higher pay and significantly better benefits.
> The business loses employees to Walmart etc. despite the factory having higher pay and significantly better benefits.
Q: Do you ever use an online job service to advertise jobs and collect applications?
Asking because my 5 sons all learned that job portals auto-trash applications w/o a job history (1st time job seekers).
Other viable but never-seen applicants: Minimal or sporadic job history, the most minimal of criminal records, the wrong zip code.
Seen but never hired: Fully qualified people who are awful at job interviews.
> the best luck they had was getting in good graces with local probation officers & craigslist classifieds.
I appreciate the answer. And I get that you may not have more granular info than this.
But I am wondering what how jobs were advertised prior to utilizing ProbOff/CL. Maybe the answer is this. There was no avenue to get job listings in front of the most likely eyeballs.
I guess most of these jobs don’t allow for music or YouTube to be used during work?
I’m just thinking that people already spend a lot of time just consuming content, so if it were possible to watch YouTube while at the factory, maybe it wouldn’t be as unpopular.
Feels like there are a bunch of factories like that in the Midwest even now. There's a Honda factory near the Columbus, OH area where you have a bunch of employees doing absolute monotonous work all day like checking if a screw is the right shape or something. These jobs are slowly getting automated but it's not like no one would do them if they are available.
Should hire us autists and allow us to program via voice commands and augmented reality.. i would love something almost automate-able while doing something that also needs higher brain functions.
If they're losing employees, then they must not have that much higher pay or better benefits for it to be worth it to work there. I don't think you can easily blame it on the job being monotonous...
The job being monotonous is clearly enough of a downside that significantly higher pay and benefits are needed to attract talent.
Paying higher wages might help retain employees (or not! there are jobs people just won’t keep doing no matter the pay) but doing so could easily increase costs to the point where your product is uncompetitive in the market. It also might just be worth having higher turnover in order to keep prices low.
> The business loses employees to Walmart etc. despite the factory having higher pay and significantly better benefits.
Better pay + benefits than the most rock bottom lowest possible pay + benefits is really pathetic.
And based on the vagueness of your claims, we can assume full-time hours are also out of the picture, meaning no health insurance.
On top of that, tyranical small business owners are usually a nightmare to work for.
The slight problem with how AI is currently being marketed is that AI is going for the fun and creative jobs that people want to do, not the dull and repetitive jobs that nobody wants to do.
If every creative job is gone to the AI beast then there will be people willing to do factory work since nothing else will be available.
My guess is that investors expect AI to automate manufacturing, and are waiting to see where that tech goes before spending a ton of capital on soon-to-be-obsolete machinery.
The article is implying throughout that these two things are mutually exclusive, and while that makes some intuitive sense (only so much money to invest after all), the last chart [1] doesn't give any indication that data center investment comes at the expense of industrial investment.
[1] "Private sector spending on equipment, adjusted for inflation"
They are now open about it. Musk tweets about a new company Macrohard, which does not manufacture itself (https://xcancel.com/elonmusk/status/1977281341264740625#m):
"Our goal is to create a company that can do anything short of manufacturing physical objects directly, but will be able to do so indirectly, much like Apple has other companies manufacture their phones."
In other words, we are a knowledge economy and outsource like it's the 1990s with a bit of "AI" fantasies thrown in. The crash cannot come soon enough.
The US spent decades transitioning from a manufacturing economy to a service economy, deliberately.
Now there's a populist making political hay, throwing out numbers about trade deficits, which ignores revenue from services. Yes, there is have a trade deficit on goods, that was a long-term strategy because services were a superior investment.
Manufacturing is an inferior way to make money unless you're planning to go to conventional war, and since the US is a nuclear superpower it's never going to get into an existential boots-on-the-ground Serious War again unless it just wants to cosplay. Nukes make conventional war for survival irrelevant.
So: it took decades to burn the boats with manufacturing, and trying to rebuild them in a few years is a hilarious folly. It absolutely will not go anywhere, and honestly shouldn't anyway. There is real danger, however, that the US burns the boats on the carefully crafted service sector as well.
I actually think infrastructure and competence in AI is going to be huge for national security in a a few years.
Basically, I think future wars will be fought with AI drone swarms. If your AI is crappy, then your drones will suck and you'll lose the war.
It's true that today's use cases are about AI slop content. Then again, a lot of modern internet technology was spear-headed by porn sites.
It all drives ads
All we're doing is building platforms for ads, pits for advertisers to pitch dollars, nothing is getting made, all it does is drive consumerism. Google, Meta, Amazon, aside from now NVidia the whole economy is increasingly built around selling slop that we decreasingly know how to make anymore.
Wait. What if the AI gold rush contributes to better industrial robotics and ushers in an AI industrial revolution? China already has dark factories with no humans on the assembly line. Isn't that a possible outcome of the AI gold rush? (I mean omitting the fact that ChatGPT 5 Pro still says stuff like: "You’re right. I made a bad inference and defended it. That’s on me." We don't want that behavior on the assembly line.
I'm unclear on what people see in the current AI tech advancements that makes them think it will contribute to better manufacturing. The new feature of LLMs that makes them so interesting is their ability accept input and flexibly follow arbitrary instructions, meaning they're really good for varied work, especially when there are a wide range of acceptable answers ("creative work"). Everything I know about manufacturing at scale is that you want a person or machine that follows a tiny instruction set (at least in comparison to the potential flexibilities of an LLM) and nails the execution every time. This seems to me like the complete opposite of the strengths of an AI system like the ones that Wall Street are cheering.
Manufacturing robotics is all about movement. All movement exists on a spectrum of difficulty and context needed to perform. For instance, welding the steel plates together in an empty and repeatable consistent 3d space is now on the lower end of difficulty. Navigating through a partially manufactured vehicle cab to install a complicated dash assembly requires a lot of context and is incredibly difficult for a robot to do.
The more we can bring down all the difficulty of all these processes, the more we can accelerate manufacturing locally.
That's at odds with everything I know about manufacturing robotics, having worked with people doing that work. The complexity of the environment is irrelevant because the robot is programmed to make a specific motion and to adjust that motion in predictable ways based on the appearance of specific features. That is by design, not because (or at least not just because) the robot is incapable of planning its own motion. The whole system is designed to be predictable instead of adaptable because that's what you need to do to do the same thing millions of times.
I am not an expert in this, and don’t necessarily believe it. But the pitch is that existing manufacturing automation requires that specificity due to technical constraints. And that much of the factory automation that hasn’t happened is because it’s too costly to get to that level of specificity in that the existing automation requires higher scale to be cost effective. If you had more general purpose intelligence you could get around those constraints.
The video models are the ones that seem to be attracting the most attention in this area as it seems do similar to sight recognition.
I've heard that the general transformer architecture (not specifically LLMs, which imply a language model, but applied to sensory perceptions and outputting motor commands) has actually been fairly successful when applied to robotics. You want your overall assembly line to have a tiny, repeatable instruction set, but inside each of those individual instructions is oftentimes a complex motion that's very dependent upon chaotic physical realities. Think of being able to orient a part or deal with a stuck bolt, for example. AI Transformers potentially would allow us to replace several steps in the assembly that currently require human workers with robots, and that in turn makes the rest of the assembly much more reproducible (and cheaper).
Training these models takes a bunch more time, because you first need to build special hardware that allows a human to do these motions while having a computer record all the sensor inputs and outputs, and then you need to have the human do them a few thousand times, while LLMs just scrape all the content on the Internet. But it's potentially a lot more impactful, because it allows robots to impact the physical world and not just the printed word.
And it's a nice problem to solve with AI of many kinds because you can forward-solve the kinematic solution and check for "hallucinations": collisions, exceeding acceleration limits, etc. If your solution doesn't "pass", generate another one until it does. Then grade according to "efficiency" metrics and feed it back in.
As long as you do that, the penalty for a a slop-based fuckup is just a less efficient toolpath.
That's not how the LLMs should be used in manufacturing. It is still the current assembly lines robots that will do that. LLMs can be used by the humans who design the automation workflow, as coding assistants. That can lower the breakeven number of items that can be automated. Maybe if today it only makes sense to automate the manufacturing of a widget only if you can sell more than 100000 of those widgets, then with LLM assistance that number can be reduced to 1000. Whenever you have a 10x improvement of something, there's scope for a mini-revolution to happen.
Someone has to run the robots. And i bet it's not going to be the educated but spoiled workforce of the developed western world, but that will be outsourced to offshore destinations.
I think there's something cultural about wanting office jobs related to power over people, where you can always slack instead of waking up every day at 8 to go to the factory
Journalists keep conflating LLMs with AI. You don't use an entire DC with its own power plant to keep a line of robotic welders online and working.
FWIW journalists are just following the lead of tech executives and others hyping LLMs as "AI" so it's hard to fault the journalists specifically.
I beg to differ. Journalists are supposed to do their own investigation and analysis of the people, institutions, and events that they report on. If they just parrot the talking points of executives, then they’re producing advertisements, not journalism.
They've been trained by a decade of referring to advanced cruise control as "full self-driving".
What if it contributes to an evisceration of the middle class, instead? Hiring for new grads is already dead because of it, and it's not going to be coming back.
It's having the same sort of impact as unlimited immigration, except that in this case, the workers don't need weekends, or pay taxes.
China has been building robots and robotic manufacturing without AI. So why AI? Because the AI is a grift for those who can get exposure to its potential gains during the exuberance, while China builds actual capabilities. Profits and fiat are shared delusions, monetarily speaking, robots and factories are real, and will build real things.
Western executives who visit China are coming back terrified - https://news.ycombinator.com/item?id=45563018 - October 2025
Was Made in China 2025 Successful? [pdf] - https://www.uschamber.com/assets/documents/Was-Made-in-China... - May 5th, 2025
ASPI’s two-decade Critical Technology Tracker: The rewards of long-term research investment - https://www.aspi.org.au/report/aspis-two-decade-critical-tec... - August 28th, 2024
> Now covering 64 critical technologies and crucial fields spanning defence, space, energy, the environment, artificial intelligence (AI), biotechnology, robotics, cyber, computing, advanced materials and key quantum technology areas, the Tech Tracker’s dataset has been expanded and updated from five years of data (previously, 2018–2022) to 21 years of data (2003–2023). These new results reveal the stunning shift in research leadership over the past two decades towards large economies in the Indo-Pacific, led by China’s exceptional gains. The US led in 60 of 64 technologies in the five years from 2003 to 2007, but in the most recent five years (2019–2023) is leading in seven. China led in just three of 64 technologies in 2003–2007 but is now the lead country in 57 of 64 technologies in 2019–2023, increasing its lead from our rankings last year (2018–2022), where it was leading in 52 technologies.
You can show any number in isolation and it can mean anything.
Now try presenting it the distribution of typical job gains/losses!
Just my 5 cents. Running factory is damn hard job. 10 products built from 50 different parts having 70 different vendors is a small nightmare. So me people can manage that, but the most can’t. Most people in Western world also cannot imagine staying at conveyor belt or table doing the same assembly task whole week. I work in a factory and see this daily.