Can AI-Driven Robots Save Japan? || Sept 11, 2026

Japanese don't like Americans? Well, the line of countries that don't like Americans these days stretches around many blocks, all because of our infantile furor.

that’s not racism.

Imagine if we could pay LLM robot workers to do their jobs and then charge them for the electricity they need to function as well as other services, like maintenance, rent for their housing stations and so on.

T-800s with Japanese accents... woo-hoo...

Seems like Japan is pursuing the larger, more critical problem, while the hyper scalers are pursuing greed, and ought to get first dibs on the best chips.

Peter would you consider commenting on this please: https://www.facebook.com/reel/1478714260466198

This is how we get Gundams.

I think you are grossly underestimating the speed at which they will be able to train general purpose automations like you're describing. Tesla has been doing this for a decade and their cars still routinely choose to drive into walls-- not withstanding the general incompetence of that organisations leadership. Boston Dynamics would have solved it if it was that easy. What I would be expecting instead is expanded robotics for easily repeatable generalish tasks-- something akin to taking the specialist robotics they have and pushing it forward for one machine to handle maybe a handful of similar looking tasks. But it won't exactly be a "carer robot" or an "agribot" or something. And frankly it doesn't need to be. Hyper generalisation doesn't actually make a lot of sense. Also for those not in the thick of it, most of what people are calling models these days aren't just models anymore, and saying they're going to "train" a solution hides a massive amount of detail. At the frontier everything is a combination of a range of models stacked in novel ways taped and rubber banded together with what these labs consider software engineering. This is true of Anthropic and OpenAI-- less so of Google / Deepmind. Nobody is really releasing pure models for production use that aren't heavily harnessed. Our techniques aren't remotely close to being sophisticated enough for what non technical folks would expect when they hear the way the industry talks about it. With this in mind they'll probably be building specialist vertical centric software with a range of smaller easier to train models. An instruction or orchestration model, safety rating model, vision model etc. trained on the specific sandbox the thing will be working in. These then get wired together for the use case with a bunch of heavy safeguards.

I think that there's a missing piece here. LLMs are trained on digital data. The "data set" that is the physical world is vastly more complex and unpredictable. For example, to train the first driverless car that is considered safe enough to deploy took millions of hours (combined across platforms) to train through real world experience -- and compared to the kind of tasks that general purpose or even narrow purpose-built robots will have to navigate the task of training a car is relatively simple. After all, it's limited to one realm -- roads -- and to one rule -- don't hit anyone or anything. So it seems unlikely that anyone will unveil a more general purpose robot until a similar amount of real world training/learning has taken place. Just ask the folks at Boston Dynamics about how hard this is in real life. Their approach is motion-capturing human movements, taking that data and distributing it among X number of instances that then work independently and "genetically" combine to evolve a model that can be downloaded into a robot in order to perform the task. Not nearly ready for more generalized activities. And that doesn't even take into account the engineering challenges and cost of producing enough "platforms" to deploy in order to begin massive parallel learning in the real world.

Robots do not earn wages or buy consumer goods, creating a structural risk for market demand if automation replaces human labor. Japan will be tariffed for Robot exports

Look at how hard China is automating their factories. I suspect that Covid hit China way harder than they let on, with 200-300 million more deaths than reported. Quadrupled the number of creatorioums which all were running 24/7 flat out for over a year, plus the countless villagers quietly buried in the fields.

I would be hesitant to be cared for by robots run by LLMs or related technology. I'm American, and recent studies show LLMs value American lives less than lives of Chinese, Pakistani, or Japanese. If resources are short, it might decide to let me die and focus on others.

I think AI + Robots (automation) is going to help a lot more than Japan. I'm mainly thinking China. Really all of us. Maybe the demographic time bomb isn't going to be as bad as predicted.

Two points... 1. The Japanese had industrial robotics mastered years ago. If you've ever seen a Fanuc robot in operation, you were impressed. Fanuc opened the first Lights Out (dark) factory 25 years where they used robots to make other robots. 2. The Japanese are generally polite, but definitely don't like non-natives. What surprises me is how many Koreans have been admitted to make up for the lack of native Japanese workers. I have a friend who's a resident alien from Japan and the Koreans and Chinese are her favorite groups to hate on. I've never asked her about Filipinos or Indians. Lol.

Japan is automated at a frightening pace. I restaurants, you order from a tablet, then the food is brought by a robot. I have been at a restaurant that didn't even have a cashier, you were supposed to pay according to the honor system.

This is being done all over and Japan is late to the game at this point. What makes Japans entry into this realm promising? They don't exactly have a history of great software companies. https://youtu.be/1cllCVK-9lo?si=-V6UA0l5BjXyKNoA

@Mark Meyer: Japan's unemployment rate stands at 2.24% and falling. For every person looking for a job in Japan, there are 1.18 openings. People in their 20s are not struggling to find work, but they might not be able to find a job matching their high education. @Pater: There are labor shortages in construction, elder care, logistics, hospitality, manufacturing and IT. Many jobs are now filled by "temporary" foreign workers, despite public opposition.

Oh shit we've entered an alternate timeline, the Mandela effect is real... Peter just gave us promising news.

You stated there is a workforce shortage because of older people aging out and yet people in their 20s are struggling to find work. Any thoughts on why that is?

Having been living in Japan for more than 35 years, and working in the field of robotics and industrial networks, this video was very reassuring to see.

en vid eeee ah