Hacker News new | past | comments | ask | show | jobs | submit login
‘Gods’ edging out robots at Toyota facility (japantimes.co.jp)
248 points by marcusgarvey on Apr 23, 2015 | hide | past | web | favorite | 117 comments

This is a fundamental shift from how many other factories work.

In general, the workers are not given enough information to replicate the product. They are shown how to attach a specific sprocket to a widget. The tasks are kept so simple (nothing to do with pleasantness) that no one can learn how the whole device is created or suggest improvements. This model lends to people being replaced by robots because human capital is one of the most expensive parts of running a factory.

In the case of Toyota, it sounds like these are high skilled engineers who are watching, understanding and making improvements to how the robots behave.

While the article itself is good, the title feels like link bait to me because the engineer's job is not to replace the robot but to enhance the robot. The dynamic isn't human vs robot as in the first scenario, instead it is human & robot cooperating.

With these types of factories, I think we'll likely see a resurgence of factories in rich countries as the most expensive component of the factory shifts from human capital to transportation of completed goods.

[edited for grammer, sic]

There are two distinct groups involved here.

First, there is a specific role in Toyota group companies with the title "Production Engineering", which is tasked with creating/improving/redesigning/etc. the production process of particular products, new and old. They are the most important category of engineers hired as undergrad and masters engineering graduates, and everyone knows that they are the group best positioned for career advancement through the engineering side of the company. The exact title is "Seisan Gijyutsu 生産技術". If you see a Toyota group company CEO with an engineering background, chances are he came up through this division.

Second are the "craftsmen" who are typically HS grads from the Aichi area, sometimes from Toyota branded schools. They are the ones with the deft touch and individual skill that are often lauded in these western oriented articles.

It's only when you have both sides working together that you will have scalable, transferrable progress.

disclosure: I worked at a Toyota group company as a new grad in Aichi and spent 3 months in the factory as part of my training.

That's a good point.

Here in silicon valley, the best performing companies are rarely the ones who have only engineering talent. A good engineer often lacks domain experience in the space they are trying to innovate in. It is important to find and work with those who have domain experience in order to progress quickly.

Thanks for pointing this out. :)

The appropriate analogy for silicon valley would be if your programming rockstars were not the same people as your architectural rockstars and neither had more than the most basic experience in the other area. Imagine if you had a bunch of people who could theorycraft a great system really well but didn't understand much past hello world, and you had some really amazing programmers who could code just about anything efficiently but didn't understand system architecture.

Without those two areas working together and understanding each other, you end up with a broken system. This is precisely the sitaution that you face in mechanical design + mechanical manufacture.

I am a mech engineer by profession and it's critical for me to engage early and often with the tradestaff who manufacture my items, so that they can guide me. I'll understand the design fundamentals far better than they will, but equally they'll understand the tooling and production processes that it's necessary to respect.

I can propose a design that will work, but they're the ones hwo can make it result in half the material wasteage with a few tweaks. If I can ensure that such aspects are successfully incorporated into the design then the part is better than either of us could achieve individually.

In reality, the programming/IT side of things tends to blur and you tend to have quite significant experience in both areas, so the problem is less pronounced. It's far more distinct in the mechanical world.

This is a great perspective.

You've done a pretty detailed analysis of the company process.

So is the Production Engineer "Seisan Gijyutsu" focused on finding improvements for the Craftsmen to develop or does the Production Engineer "Seisan Gijyutsu" develop the improvements and teach them to the craftsmen?

The production engineer is there to optimize the production line - this is where Iso9000 and BS5750 really come into their own.

eg if i replace this machine tool with x% scrap/rework rate with a new one that has y% and the new machine costs Z$ million - how long does it take to payback.

I would suspect that the engineers theorise what can be done and produce a small scale demonstration and then the craftsmen take the concept and refine it into something that's reproducible at scale too.

Thanks for the nice summary. No way was I going to click on a headline that read like this one!

There are two kinds of jobs that robots will have a hard time replacing: these sorts of high-skill, creative, intution-and-experience heavy tasks; and low-skill but mechancially awkward tasks.

In both cases, the key to job survival will be that it's something of a niche. Anything remotely generic will be taken over by machines. So for example we'll still see certain kinds of literature generated by humans, but 100% of financial reporting, political reporting and sports reporting will be automated. More sophisticated commentary of the kind that Nate Silver does might last, but the Paul Krugmans and David Brooks of the world will be engineered out in the next decade or so by bots that troll the web for stories and generate canned ideologically-loaded commentary based on a smattering of data sources.

At the other end, jobs like fruit picking will be completely automated because it's a common task that can be implemented with a relatively generic machine, but cleaning houses will remain the domain of humans because houses are such awkward, complex spaces. Once we have a humanoid robot with very good AI that'll change, but I'm a little doubtful about seeing that in the next few decades.

>There are two kinds of jobs that robots will have a hard time replacing:

Three - add anything which is primarily social.

I think hairdressing will be one of the most resilient jobs. Not only is there a huge market for it, but it requires a certain basic level of dexterity, and it's primarily social. The talking is at least as important as the hair cutting, and it's going to be a while before robots can replace it.

Hairdressing may be resilient for some people, but my eyes were opened in Singapore when I went to "K-Cuts" - which is as close to an automated hair-cut as you can get. Full haircut, in 10 minutes for S$10. Nobody speaks english, so talking isn't an option. They use vacuums to collect your hair. I hold up three fingers, point to the top of my head, two fingers, point to the side. That's the sum total of guidance.

And these places are busy - my guess is there will be an automated haircut place in <10 years, and they will pick up 90% of the business within 5 years.

People will always go to custom hair cut places, but I bet there are a lot of people who will be keen to pay a small amount of money to get a reasonable hair cut.

Your anecdote is good to keep in mind, I learned something. While I don't think the US has any auto-cut places (oh the liability!), we certainly have had infomercials selling vacuum haircutting machines for a while[0]. I'm not sure they've made a huge dent in the beauty industry to date.

Beauty is not an industry that a bit of tech pixie dust will magically transform. Haircuts aren't just a utility, they're a personal statement and a piece of one's identity. And as a previous poster indicated, it's also social and cultural. Plus the variety of people's hair and variety of desired cuts make it an incredibly complex and fine motor skill, one that robots will suck at for a very long time. And people rarely know exactly what they want when they go into a hair salon, so there's a service component there too. Beauty just isn't ripe for disruption because people and culture are so involved.

I think this is similar to the soylent discussions. Food-as-utility people think its great, but food has cultural value far beyond calorie intake. So its found its niche among products that already existed (sans tech pixie dust) - the nutritional shake market.

90% is an order of magnitude off in my opinion. But I agree that a place that offers reasonable hair cut for cheap is what many people would want, so I do look forward to your future where that is offered!

[0] http://www.flowbee.com/ http://www.haircut.com/ http://www.aircut.com/

I'm really interested in finding out - at $10/haircut/10 minutes - It's the sort of thing you can do casually every couple weeks without even thinking about it.

Definitely agree that there will be some people who absolutely will want to go to a hair salon - and, honestly, my 90% was just a thumb-in-the-wind estimate. You're probably right that it's high, but, never underestimate the desire for people to get a good deal.

I think your Soylent comparison is excellent - and is exactly the one I would make. People who see food in a utilitarian perspective see the value offered from soylent. Likewise, people who consider haircuts as the process required to rid themselves of the excess hair, will get a lot of value out of an automated hair cut establishment.

There's probably a market for very cheap, simple silent haircuts.

There is most definitely a market, and these guys are killing it in East Asia: http://www.qbhouse.com/

A better link is http://www.qbhouse.com/sg/about/ which kind of explains what it is. And as far as I can see it's all about cheap, efficient haircuts. I can't see how this is a thing! Of course, I'm not sure how Krispy Kreme got to be such a phenomena here either, so what do I know. But, the duration of my haircut never struck me as something I really needed to optimize.

Makes me happy I learned the trade of barbering in my dad's barbershop as a teenager. There is still a part of me that some day would like to unglue myself from the computer and return to the shop.

I understand your feelings. A part of me wants to unglue from the computer and be a baker.

I would definitely buy a robot hair cut if it was fast, cheap, and worked well.

> but the Paul Krugmans and David Brooks of the world will be engineered out in the next decade or so by bots that troll the web for stories and generate canned ideologically-loaded commentary based on a smattering of data sources

Characterising a Nobel-winning economist as a content troll says more about you than him.

Krugman is a Nobel-winning economist. He is also a loudmouth ideologue who writes a column, in which he makes authoritative statements that have nothing to do with the research that won him a Nobel. He trades on that prestige, but he's no better than a highly-ideologically-driven professional-economist-but-out-of-his-area-of-specialty in most of what he writes. I have seen him rip Republicans for the same behavior that he lauds Democrats for.

I wouldn't call him a troll. I would call him a shill. And no, that's not not saying more about me. It really is about him.

I doubt you read his column. Whether you agree with him or not he provides tons of economics data for his arguments. His topics are frequently technical. I'd say at least 70% of his topics have to do with what he is a Harvard professor for.

> I doubt you read his column. Whether you agree with him or not he provides tons of economics data for his arguments.

Yeah, that post seems kind of bizarre to me. Is Krugman partisan? Absolutely, and he admits it. Can he be - compared to most of the professional commentariat - abrasive and rude? Yes, and he admits it. [1] At the same time he is a cheerleader for Nate Silver and data-driven analysis. This is definitely reflected in his blog posts though somewhat less so in his columns (which do tend to be fluffier for a popular audience.) Half of his writing since the beginning of the recession has been deriding narrative-based analysis and demonstrating that the data contradicts it.

He does of course sometimes write about subjects outside his expertise - he wrote a lot about the Iraq War, for example - and he sometimes shades his economics with political analysis, but I wouldn't go so far as calling him a shill. Krugman is obviously derisive of most Republicans, but he has been decidedly unhappy about most post-2008 Democratic policies.

Krugman is basically the only mainstream political pundit I read at all, because you can actually learn a lot from what he writes.

I can see machines replacing David Brooks, but it will be a tragic day when we lose such a fitting (though well-intentioned) subject for mockery.

[1] He gives his reasons here: http://krugman.blogs.nytimes.com/2014/09/14/wild-words-brain...

I know that this is a useless point to make because of the prevailing political propaganda, but Krugman is a highly trained, accomplished and creative academic economist, he provides exactly the kind of "sophisticated commentary" that you say might last. David Brooks I agree could be replaced by a robot.

I understand your point, but I disagree.

It may take some time, but many of these 'awkward' jobs too will disappear. Once the house is the machine, the problem of maintaining it becomes a lot simpler. In fact, these environments may look nothing like a house (I would expect the first iteration to focus on low-cost functionality over form). Such machines could eliminate the need for independent robots or human labour.

"No way was I going to click on a headline that read like this one"



The explanation that I get from article for the shift to training more metal workers is that a highly skilled metal worker winds-up being more akin to an engineer - that aspects of metal work are too complex to be understood by just theoretical knowledge of the engineers. Thus the robots in the factories duplicate the processes which both metal workers and engineers create. But if you get rid of the metal workers, you are left not able to change your processes or create new, effective processes.

Some tradestaff do progress in to 'engineering' roles and in many professional engineering organisations there's provision for recognising this. They tend to be given a title like 'technical officer' and their role is to do the engineering with a bit less maths and a bit more hands-on experience.

Recognising and upskilling the best tradestaff into these roles is one of the most important aspects of a successful mechanical design organisation, as far as I'm concerned. They deliver a critical insight that your degree-educated, never-worked-away-from-a-desk engineers simply lack, because those engineers haven't spent years bending and cutting steel to understand exactly how far you can push the processes.

Example: I've just finished redesigning a proposed lifting device where some idiot engineer had specified steel bent in a radius of half its thickness. Any manufacturing tradestaff will tell you this is complete insanity as the rule of thumb is ~1.5-2* thickness as a minimum radius - any tighter and you rapidly start tearing the material apart along its grains. It's just not done. Yet there's clearly an engineer out there who didn't even consider it and sure enough, as soon as we crack tested the bent area, about 1/3 of the items had cracks.

This is the critical stress area of a proposed lifting device so that's basically a cardinal rule broken, because the manufacturing defect (the tear in the metal) will propagate into a crack over successive lifting cycles until the whole thing breaks in half and the several tonnes of load it's supporting falls back down. Had the engineer talked to literally any tradesperson they would've known to avoid this, yet here we are.

er yes their always has been a vocation route it normally goes

Trade Apprentice -> Technician (associate professional) -> Engineer -> Ceng (PE)

Some Cultures make it harder to actually progress all the way Germany used to be realy bad.

Until the machines learn to learn...

I seem to recall a very old science fiction story about a future where nobody has to go to school. Instead, they play all day until they come of age, at which point a computer programs their mind with all of the expertise needed for whatever profession they'd be best at and would be most useful. The problem turned out to be that, while this is great, the programmed minds couldn't innovate further improvements, and so society grew stale. A small percentage of the population, including of course the protagonist, were chosen to learn stuff the hard way in order to be able to come up with new advancements. For the life of me, though, I can't remember what story that was. Any ideas?


"Profession" by Asimov (1957)

That is exactly it! Thank you so much for finding that for me.

I vaguely recall something similar. I think it was called "Demon on my back" or the like, but I can find no such novel listed anywhere. It was a YA story.

[Edit: Looks like it was "Devil On My Back" http://en.wikipedia.org/wiki/Devil_on_My_Back Hope that is the one you are looking for]

I think this is a real indicator and validation of what the next economy will look like: humans at the apex of the arts, where imagination and craft will be highly valued in the everyday folk worker, and mundane or trivial tasks will be assigned to automatons. We will choose our own employment adventure.

There will be few people needed at the apex of the arts, once the best of the best can do the jobs of most everyone else combined. We still need an economic plan for everyone else.

I think you're vastly underestimating the diversity of tasks to be mastered. Everything is a hard problem if you want to do it well enough, and the current economy consists of a vast number of distinct kinds things made sloppily.

Japan is naturally doing this through their low birthrates.

Their amino acids will make an excellent source material for the 3D food printers.

Nah, I'm pretty it will be more efficient to just grow meat in a lab.

Out of what?

Try googling "in vitro meat" and you may find some answers, but I don't know if this problem is solved yet. This article mentions algae as a possibility:


I don't look forward to my future as soylent green

I agree, there will always be fewer people needed at the apex of the arts, and in fact that's the definition of the apex!

But an economic plan that revisits the idea of what an apex is, I think would accomplish that 'for everyone else'.

That's why the idea of micro farming is so interesting: in an agrarian system where people grow their own food, as a pastime and 'art', then you can afford many, many apexes - at least one per farm/garden.

To enable that, we need to have an economic system that redistributes the resources so that we can all have a shot a living a decent life, reducing the working hours of the many rather than the wealth of the few. Has there been a time in history where so many people have been working so much for so little?

Of course there's been a time in history where so many people have been working so much for so little. Plantation owners used to work their slaves from dawn till dusk for nothing. Farmers used to work the land from sun up to sun down just to feed their family. What a ridiculous question.

Why do so many people think that times have never been so bad when, in all honesty, times have never been so good?

Because the specific subset of people who post in these forums, mostly middle-class North Americans and Europeans, are worse off than a couple of decades ago - coupled with the fact that the type of jobs that are lifting hundreds of millions out of poverty make us feel dirty and self-righteous.

Worldwide, there are more people in slavery today than there was when slavery was legal. Plus all kinds of manufacturing quasi-slavery in south and south-east asia with crazy hours.

Because we used to live in the garden of eden where everything was perfect, but then we sinned against god and if we don't repent we're all going to hell.

I'm joking a little but I'm also serious. I kind of think that humans are somewhat hardwired to think this way given that so many people believe some version of this story.

It's not clear. Are you advocating for socialism?

Capitalism is a method for distributing scarce resources. Remove the scarcity and you remove the need for capitalism.

Not all resources will be non-scarce, though, so this kind of one-size-fits-all thinking is misleading.

Nor are all alternatives to capitalism socialism. Socialism is also known to have very severe problems in application, particularly how to incentivize socialist managers to behave like capitalist ones in most circumstances, driving innovation and maximizing productivity. Socialist managers have somewhat different contstraints on them from capitalist managers, and nominally different long-term goals, but they are more similar than you might think.

The thing is: we know how to make capitalism work in a range of varieties, from the Scandanavian social democracies to the mess that's Texas. We have never seen a successful system of public ownership of the means of production (which is what socialism is, and nothing else). It has been tried in a variety of places and times, and it has always run up against fundamental problems of managerial incentives and political corruption at levels that make most capitalist political economies look pretty good (which is saying something!)

So looking for non-socialist means of distributing wealth in a society that is unequally post-scarce is a good idea. Today wealth is distributed almost exclusively through paid work, but that is a relatively recent invention. I'm not suggesting a return to feudalism or anything like it, but think we should be willing to consider variations in social order that are at least that large-scale, because the difference between what is coming and industrial capitalism is going to be at least as big as the difference between industrial capitalism and the semi-feudal societies that preceded it in many places.

There are post-scarcity examples in human history (see gift economies and potlach societies - common among native americans and in the pacific islands) where social pressure caused any accumulation of wealth to be shared with the entire society (this is a bit similar to Andrew Carnegie's dictum that to die rich is to die disgraced, now leading to the billionaire's pledge). Of course, this only worked because the societies were individual tribes/villages and fit within Dunbar's number, where every individual knew every other individual in the society.

If post-scarcity is brought about by automation, then there will be no need for most people to work, and socialism won't be the appropriate economic model (since people will not be directly involved in production). At this point we'll probably need to return to a gift-like economy.

Sweden is a capitalist country with a strong social safety net. It's not remotely socialist, at best a social democracy. In no way could you say that, in Sweden, 'to each according to his needs, from each according to his abilities'.

The majority of production in Sweden is not state owned.

>which is what socialism is, and nothing else

Actually, socialism is the ownership by the workers of the means of production.

A system of co-operatives would also be socialist.

You can't really remove the scarcity; if machines make mass-produced stuff non-scarce, the rest (including intangibles) will become more important. In Star Trek, everyone could probably have their own luxury car, but there was still only one captain for each whole ship.

"The Australia Project" from the novel Manna is one possible future in a post scarcity economy, it also shows one of the others.

It's a terrible method.

That is shockingly ignorant, and shame on whoever taught you that.

Free markets and trade have lifted the most people out of poverty and improved the living standards of the largest number of people in history.

It is an imperfect way of organising the needs of people, but it is superior to everything else that has been tried, by an order of magnitude.

> It is an imperfect way of organising the needs of people, but it is superior to everything else that has been tried, by an order of magnitude.

This is extremely important to remember. The fear of the unknown should not stop us from trying new economic systems that may be superior to capitalism.

Yes; but even less radically so, I'm thinking about Keynes' failed predictions.

You shouldn't advocate for socialism, even in jest. It is a terrible system which has caused more suffering than all the despots in history together.

Forward thinking will only happen when everyone walks away for the wreckage of socialism and realises it is a dead fork in the road that must be rejected utterly.

...If we build framework to support such an economy, it may be possible.

That's the crux of the issue. It seems that the current economy will rather drive itself to the ground, taking all of humanity with it, rather than transitioning itself to a different value system. It looks like the post-scarcity world is not on the phase-space trajectory we're currently following.

Right now I feel every reason to be gratuitously optimistic.

- Solar photo-voltaic is or is about to be the cheapest form of electricity, we will solve the storage problem

- Electric cars are getting closer to the tipping point, and they are going to drive themselves

- Biotech is making incredible gains, see cancer immunotherapy or NSI-189

- The pace of technological innovation has accelerated and continues

- The intellectual capital of the human race has never been anywhere near this large, free, accessible

- Robots/a lot of weak AI is or is about to get to a point to change the game

We're on trajectory technologically but not sociologically.

  "We eat in restaurants, buy branded toiletries, build 
  skyscrapers, create legislative institutions, travel in 
  flying machines, write poetry, and search for meaning in 
  relationships, temples, and scientific books. Humans have 
  discovered antibiotics, sent probes into space, decimated 
  rainforests, shared a billion views of clips of kitten 
  behaviour, and decoded their own genomes.

  But there is one thing that humans have singularly failed 
  to do, and that is to properly understand their own behaviour. "

  — Robert Aunger and Valerie Curtis:
  Gaining Control: How human behavior evolved
[0] https://global.oup.com/academic/product/gaining-control-9780...

It's hubris to say this, but we're not as far off as you might think.

If "they" can fix economics ( big if ) that alone would make a big difference. But the US only got off the gold standard in my lifetime.

There's a Churchill quote about Americans; I think it generalizes to humans - "Americans can always be counted on to do the right thing…after they have exhausted all other possibilities."

On a macro scale it's momentum and incentives at play. Daunting, but malleable.

Locally, we're close to achieving sensible cities/communities with renewable energy, easier access to rapid prototyping, and worker co-op business models.

Do what you can with what you have to make your city and friends have access to healthy enjoyable lives.

From where I'm sitting, it doesn't look like we're remotely close. Renewable energy and effective urban planning are not at all common, and worker co-op business models are quite rare.

I like your vision of the future, but I don't see much evidence that we're headed there.

I agree it's what the 'next economy' will look like, but it's also what the 'last economy' looked like. Prior to the entire tech movement was agriculture, resource harvesting and material manufacturing.

Previously farming was done by hand, now a single machine replaces a hundred workers.

Previously coal was mined by hand, now a single machine replaces thousands of workers.

Previously trees were axed by hand and sluiced by river, now you can clear a forest with a couple machines and a couple skilled operators.

Previously producing cloth required massive numbers of skilled workers, then the loom was invented.

I often see this argument that we're in some unprecedented state of replacing labor with automation. I don't see any proof of that. I think we replaced labor with automation at a much faster pace during the industrial revolution and made it out just fine.

I'd love to see some stats on the differences if anyone has them.

I didn't say replacing labor with automation, I said replacing mundane, inglorious labor with automation. I agree with you. What I am saying is that we will choose which labors we do by hand, based on an emerging, re-emerging set of ethics and values.

or... previously everybody had to produce the food they ate, now only 1% does. previously many people had to dig out coal, now a machine has relieved thousands of a shitty job......

replace "automatons" with "foreign workers" and this economy is already in place in USA. It doesn't seem like pleasant to the humans at the apex.

most will choose their own unemployment adventure.

The problem is the current size of the world population is designed and fit for the Industrial Age and not for the Information Age.

What do you suggest we have to do to the "surplus" of manpower now?

Please no retraining nonsense as jobs will dwindle too and mirror the trend elsewhere.

Please show some political sense in your reply.

Invent new enemies every few years and new wars to send all our excess manpower off to?

“To be the master of the machine, you have to have the knowledge and the skills to teach the machine.”

That is an awesome quote.

Thanks for posting the article it was an interesting read.

This is so applicable to our industry. I heard Chuck Moore say once that most microcoded division routines are bad, because it's the first and only time that engineer writes one. He said his tenth division routine was far better than his first.

It concerns me that in the rush to build abstraction upon abstraction, we're neglecting the tools and practices that make it possible to work on and improve the fundamental layers.

Considering that one of the current trends is the development of unikernels (like MirageOS), network stacks in userspace, etc, I'd say we're still thinking about the lower layers.

Back in the early 90s, I was working in a Sony plant in Singapore that produces CRTs. It's a fully automated plant and my section are in charge of making the famous Trinitron apperture gril. Anyway, I still remembered there was once a group of Japanese craftsman came to visit our plant. And some some of my colleagues were assigned to attach with them and learn. These craftsman turn out to be ladies (wow!) and they showed us how to do some of the work manually (WTF!) and the skills they have really put to shame the local guys, we stand and watch and totally stunned. The crafmanship and professionalism in their skill is truely awesome.

This is nice PR but to imply that this is anything more than that is silly. These humans are clearly just improving the robots.

I don't see that this is just another PR article, more that there will always be a place for craftsmen in Toyota, even in the face of the wave of production automation.

One thing I admire from the Japanese culture, is that they take `craftsmanship` very seriously in all aspects of work.

I gain this appreciation after working part-time as a kitchen-hand in a number of Japanese restaurants throughout my uni days.

Those that are `authentic` Japanese (ran by Japanese owners) has a very methodical process and obsess over every single detail.

Even a simple task such cooking rice has an elaborate process and specific technique that I must always follow through. Just two examples:

* the amount of water to put into the rice depends on the season (i.e. dry vs wet season at harvest), must be exact to the millilitres.

* the specific motion of washing (or more appropriately `grinding`, as the water has been sieved out).

The chef always check the rice after it cooks and gave me feedback, every single time.

I completely disagree, the type of craftsmen you're referring to will cease to exist at the company in the future. The very concept of craftsmanship is antiquated by the idea of intelligent machines. That is, machines that can learn.

No matter how hard a human tries to "obsess over every single detail" they will simply be unable to compete with a machine intelligence in this regard. All the things you listed are so obviously suited to be tasks that intelligent machines can excel and surpass humans at. Even unintelligent machines would fare quite well against a human at all the tasks you listed.

You may argue that there's an art to craftsmanship that machines "will never duplicate the creativity of humans." I hear that like 20 times a week and it's not true. This falsehood lies in a flawed premise. The premise that the machines doing art/crafstmaship will seek to "replicate" human creativity is simply false. Human creativity doesn't necessarily replicate the creativity of other humans; why would a machine creativity need to replicate? Many intelligent machines are already making art online. Machines do not need to replicate. Intelligent machines can create on their own, independently. People are doing this with twitter bots now.

Notice I use the word "machine intelligence" and not just machine. A machine is process that has been animated in the physical world. A machine intelligence is fundamentally different. It learns and is better than we are at most (if not all) things. I understand that this eventuality is terrifying to people. Being scared of the inevitable doesn't stop it. We need to be thinking about this now so we can make plans to figure out what do with humans once these machines proliferate through our society.

As an aside I also love Japanese culture. I had the opportunity to travel to Japan after high school. It was great.

What I don't understand is why the "Gods" don't say "No we won't train robots to do our jobs."

One of the biggest hurdles to automation in general - especially where replacing skilled workers is concerned - is in teaching (algorithmically or otherwise) the machine to replicate what the skilled worker is doing.

If there really is the existential crisis of "machines taking our jobs" then why are the robot trainers not pushing back? Isn't is immediately clear what they are doing?

I personally think everything should be automated with no mandatory human inputs for the majority of tasks - but that is hundreds of years away if ever. The key sticking point though is whether people will be willing to be "the last human to hold this job."

At least in Toyota, there seems to be no existential crisis of machines taking over our jobs. What the article mentions is just a strategy to improve quality and encourage innovation. These people are not particularly worried about teaching the machine to imitate the skilled worker; they're too busy figuring out what the skilled worker should be doing in the first place.

Also, gods of engineering are probably different from other gods. Like passionate programmers, they're having too much fun making machines perform new and exciting tasks.

As I understood it, under this plan there won't be a 'last human' because...kaizan...the humans will always be improving and innovating.

...Until profits are in danger and you have a conservative board. Then you might see the last human, shortly followed by corporate irrelevancy.

Would you rather program robots, or work on a car assembly line?

Programmers don't seem to have a problem automating aspects of their own jobs. Efficiency for it's own sake is a motivation.

The recall problem was caused by software -- where are the software gods who write machine code instead of using compilers? ;)

I like the joke, but I feel I should say software is still a craft and not automated. Just that the tools have improved.

Yeah and if you have to drive a nail, you should use your fist instead of a hammer.

If you only have a hammer everything looks like a nail.

The best programmers I know all find an excuse to write some assembly every so often, even if just as an exercise.

Reminds me of the John Henry myth:


probably will end the same way.

This is the normal backwash of Toyota.

Toyota was one of the(or probably the) first company to totally replace humans with computers in their production lines.

So, as a result, they are the first to say: "Well, it looks like some things computers do very badly compared to humans".

Today little things quality inspection is done by humans but most of the work is done by machines, so take whatever you read with a big grain of salt.

Yeah, there are professionals needed, my company is one of those. But most of the jobs those companies created are forever gone.

It seems to be working:

> "In an area Kawai directly supervises at the forging division of Toyota’s Honsha plant, workers twist, turn and hammer metal into crankshafts instead of using the typically automated process. Experiences there have led to innovations in reducing levels of scrap and shortening the production line 96 percent from its length three years ago."

That's how it was done at Chrysler in 1935.[1] Toyota's Honsha plant has had an automated crankshaft forging line since 1988.[2]

I suspect this is a misconception. The Honsha plant was downsized in 2007. They have a lot of space there, and a "Museum of Production". They may have a manual forging setup as a training and demo operation, to give their employees a feel for how heated metal behaves during forging. A 1988-vintage automated forging line probably has too little feedback and too much brute power, and maybe too many die stages. Better employee understanding of the process may lead to a better. It's unlikely that Honda is doing manual forging for volume crankshaft production.

[1] https://archive.org/details/0555_Master_Hands_18_27_28_00 (start at 00:11:45)

[2] http://www.toyota-global.com/company/history_of_toyota/75yea...

which is exactly what I think the article is saying. So you are right.

Sometimes I do find programmers that do not know how to, say, install a linux distro, or at a lower level, don't understand what a data bus is, or how one bit of information is actually stored in a computer.

> Learning how to make car parts from scratch gives younger workers insights they otherwise wouldn’t get from picking parts from bins and conveyor belts, or pressing buttons on machines.

This is also the spirit of the nand2tetris [1] course, currently running on coursera [2]. The book is a really entertaining read too. Depending on how familiar you are with electronics you may be able to outpace the video course quite a bit.


> “Fully automated machines don’t evolve on their own,”

not yet... :p

1: http://www.nand2tetris.org/

2: https://www.coursera.org/course/nand2tetris1

I know a computer science professor that asked me how to make a website as recently as 2007

Not entirely sure what your point is here but I assume that you mean he should have known.

Can't imagine why a computer SCIENCE professor would be expected to know how to make a website any more than my Newtonian mechanics professor would have been expected to know how to build a car.

There is a big implicit "for now" after nearly every sentence in the article. Sure, we don't have machines that can autonomously optimize themselves. For now.

Soon...soon... pets robotic white cat

Looks like the world is moving in the right direction again.

Several Toyota employees indicate the direction of the world?

Manual labor is the right direction? Wouldn't you rather have robots automating all the difficult/dangerous/demeaning work?

Encouraging skill growth, problem solving skills, and wisdom from the workforce, rather just using humans as cogs.

It's not the "moving in the right direction", it's just slowing down the pace of going in current one, and I'd say it's just a stopgap. Even if you temporarily save a few high-skilled jobs (which are least endangered anyway), you're still waiting for software "gods" to grant machines the gift of self-improvement.

So should developers learn more about how low level libraries work so they can make better code? ie, should we be fostering more Kami-Sama developers? Or do you really not need to know that stuff, and focus instead on putting together higher level libraries into awesome apps because business is business and we have deadlines?

> “If there is ever a technology that’s flawless and could always make perfect products, then we will be ready and willing to install that machine,” Kawai said. “There’s no machine that is eternally stable.”

This seems like a nice reformulation of Turing's Entscheidungsproblem

No, it's definitely a separate observation. The Entscheidungsproblem describes a problem space that we cannot build a machine to solve.

What Kawai is saying is a more physical thing - even for problems we can build a machine to solve, the machine will eventually stop working under normal use, and it's unlikely even to produce consistently correct results.

Somewhat tangential, but give 1909's "The Machine Stops" by E.M. Forster a read. It's excellent and speaks to this point.


The idea reminds me of this SciFi story from 1949: http://escapepod.org/2013/10/28/ep419-expediter/

It's an avowed aim of the Toyota Production System

Automation with a human touch


I guess a programming analogy might be the "radical" idea that it might make sense to let highly paid engineers read through the source code of libraries that you use.

Headline term lost in translation; but well written concept.


Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact