Japanese White-Collar Workers Are Already Being Replaced by Artificial Intelligence (qz.com)

Most of the attention around automation focuses on how factory robots and self-driving cars may fundamentally change our workforce, potentially eliminating millions of jobs. But AI that can handle knowledge-based, white-collar work is also becoming increasingly competent. From a report on Quartz: One Japanese insurance company, Fukoku Mutual Life Insurance, is reportedly replacing 34 human insurance claim workers with "IBM Watson Explorer," starting by this month. The AI will scan hospital records and other documents to determine insurance payouts, according to a company press release, factoring injuries, patient medical histories, and procedures administered. Automation of these research and data gathering tasks will help the remaining human workers process the final payout faster, the release says.

Comments

As if this is new

[ranton]

As a software developer of enterprise software, every company I have worked for has either produced software which reduced white collar jobs or allowed companies to grow without hiring more people. My current company has seen over 10x profit growth over the past five years with a 20% increase in manpower. And we exist in a primarily zero sum portion of our industry, so this is directly taking revenue and jobs from other companies.

People need to stop living in a fairy tale land where near full employment is a reality in the near future. I'll be surprised if labor participation rate of 25-54 year olds is even 50% in 10 years.

Re:As if this is new

[Oswald McWeany]

I completely agree. Even jobs which a decade ago looked irreplaceable, like teachers, doctors and nurses are possibly in the crosshairs. There are very few jobs that AI can't partially (or in some cases completely) replace humans. Society has some big choices to make in the upcoming decades and political systems may crash and rise as we adapt.

Are we heading towards "basic wage" for all people? The ultimate socialist state?

Or is the gap between haves and have nots going to grow exponentially, even above today's growth as those that own the companies and AI bots make ever increasing money and the poor suckers at the bottom, given just enough money to consume the products that keep the owners in business.

Re:As if this is new

[Grishnakh]

Society has some big choices to make in the upcoming decades and political systems may crash and rise as we adapt.

Are we heading towards "basic wage" for all people? The ultimate socialist state?

It depends on the country, I think. I believe many countries, like Japan and Finland, will indeed go this route.

However, here in the US, we are vehemently opposed to anything that can be branded as "socialism". So instead, society here will soon resemble "The Walking Dead".

Re:As if this is new

[EvilSS]

I think even in the US it will hit a tipping point when it gets bad enough. When our consumer society can't buy anything because they are all out of work, we will need to change our way of thinking about this, or watch the economy completely collapse.

Re:As if this is new

[alexgieg]

The problem is that AI is becoming faster at learning the new job opportunities than people are, thereby gulping them before people even were there to be replaced. And this speed is growing. You cannot beat an exponential growth with a linear one, or even with just slightly slower growing exponential one.

Re:As if this is new

[unixisc]

My original comment was not about home schooling. It was about parents spending all time w/ their kids once kids are out of school - no daycares. That would include being involved w/ helping their kids w/ both homework and extra curricular activities.

Re:As if this is new

[Grishnakh]

I'm going with the latter (complete economic collapse). There's no way, with the political attitudes and beliefs present in our society, and our current political leaders, that we'd be able to pivot fast enough to avoid it. Only small, homogenous nations like Finland (or Japan, even though it's not that small, but it is homogenous) can pull that off because they don't have all the infighting and diversity of political beliefs that we do, plus our religious notion of "self reliance".

Re:As if this is new

[Anonymous Coward]

Nah, the economy won't completely collapse. Here is how things will play out:

1) Labor automation continues to make the teeming masses of poor people obsolete.
2) Unemployment continues to rise due to this.
3) Crime rates continue to rise due to unemployment.
4) Incarceration rates continue to rise; the prison industry continues to grow and thrive on tax dollars paid by rich people.
5) An entire generation of poor people dies off in prison.
6) The population shrinks back down to a workable number due to this.
7) The rich families thrive in an interlocked economy of ownership of labor automation, while everyone else dies off in prison.

so you see. we will have something like a living wage, but it will actually be residence in prison where you are not allowed to breed.

The end.

Obviously

[Anonymous Coward]

People that do trivial tasks like looking at numbers on documents, something a computer can easily do, are prime for getting replaced.

Face it, if you aren't creating new things, you're the first to go. Maintaining a process is basically pattern recognition.

Re:Obviously

[kwerle]

SInce this is very very similar to what my partner does, I feel like I'm a little qualified to speak on the subject at hand.

Yeah, pattern matching should nail this - but pattern matching only works if the patterns are reasonable/logical/consistent. Yes, I'm a little familiar with advanced pattern matching, filtering, etc.

Here's the thing: doctors are crappy input sources. At least in the US medical system. And in our system they are the ones that have to make diagnosis (in most cases). They are inconsistent from one doctor to the next. They are inconsistent from one day to the next. They are inconsistent from one patient to the next. They are inconsistent *within a patient* when the original diagnosis was wrong. And what's possibly worst of all: they disagree.

In the US we do the same kind of thing - base payouts on what the doctor diagnosed. They need to write specific magic words in the right way. So my partner looks at medical records and then confronts the doctor - somehow trying to suggest what they left out without making a diagnosis (because she's not a doctor, so she's not allowed to).

As you can imagine this is a delicate dance. Some doctors have egos. In any case, many of things she does are fixing errors [of omission, often], but others are a lot more complicated and sometimes very rare (some medical conditions just don't come up very often).

Finally, if you think having a person hound a doctor to get something corrected might be tricky - imagine having a machine try to do the same thing. Some doctors may be more resistant to that...

The easy answer to this is: that process is crap. Fix doctors/the system/whatever.

I agree.

Good luck with that.

Queue the chicken littles

[sjbe]

Calm down everyone. This is just a continuation of productivity tools for accounting. Among other things I'm a certified accountant. This is just the next step in automation of accounting and it's a good thing. We used to do all our ledgers by hand. Now we all use software for that and believe me you don't want to go back to the way it was. Very little in accounting is actually value added activity so it is desirable to automate as much of it as possible. If some people lost their jobs doing that it's equivalent to how the PC replaced secretaries 30+ years ago. They were doing a necessary task but one that added little or no value. Most of what accountants do is just keeping track of what happened in a business and keeping the paperwork flowing where it needs to go. This is EXACTLY what we should be automating whenever possible.

I'm sure there are going to be a lot folks loudly proclaiming how we are all doomed and that there won't be any work for anyone left to do. Happens every time there is an advancement in automation and yet every time they are wrong. Yes some people are going to struggle in the short run. That happens with every technological advancement. Eventually they find other useful and valuable things to do and the world moves on. It will be fine.

Re:That's weird

[phantomfive]

"Watson" is a marketing term from IBM, covering a lot of standard automation. It isn't the machine that won at Jeopardy (although that is included in the marketing term, if someone wants to pay for it).

IBM tells managers, "We will have our amazing Watson technology solve this problem for you." The managers feel happy. Then IBM has some outsourced programmers code up a workflow app, with recurring annual subscription payments.

Re:Japanese workforce is growing old

[avandesande]

Japan is notorious for being far behind on office automation.

That is "automation". AI is something else...

[gweihir]

When you have people basically implementing a process without much understanding, it is pretty easy to automatize their jobs away. The only thing Watson is contribution is the translation from natural language to a more formalized one. No actual intelligence needed.

Failure of imagination

[sjbe]

As a software developer of enterprise software, every company I have worked for has either produced software which reduced white collar jobs or allowed companies to grow without hiring more people.

You're looking at the wrong scale. You need to look at the whole economy. Were those people able to get hired elsewhere? The answer in general was almost certainly yes. Might have taken some of them a few months but eventually they found something else. My company just bought a machine that allows us to manufacture wire leads much faster than we can do it by hand. That doesn't mean that the workers we didn't employ to do that work couldn't find gainful employment elsewhere.

And we exist in a primarily zero sum portion of our industry, so this is directly taking revenue and jobs from other companies.

Again, so what? You've automated some efficiency into an industry that obviously needed it. Some workers will have to do something else. Same story we've been hearing for centuries. It's the buggy whip story just being retold with a new product. Not anything to get worried about.

People need to stop living in a fairy tale land where near full employment is a reality in the near future.

Based on what? The fact that you can't imagine what people are going to do if they can't do what they currently are doing? I'm old enough to predate the internet. The World Wide Web was just becoming a thing while I was in college. Apple, Microsoft, Google, Amazon, Cisco, Oracle, etc all didn't even exist when I was born. Vast swaths of our economy hadn't even been conceived of back then. 40 years from now you will see a totally new set of companies doing amazing things you never even imagined. Your argument is really just a failure of your own imagination. People have been making that same argument since the dawn of the industrial revolution and it is just as nonsensical now as it was then.

I'll be surprised if labor participation rate of 25-54 year olds is even 50% in 10 years.

Prepare to be surprised then. Your argument has no rational basis. You are extrapolating some micro-trends in your company well beyond any rational justification.

Re:Failure of imagination

[ranton]

It's the buggy whip story just being retold with a new product. Not anything to get worried about.

The buggy whip story shows that an entire species which had significant economic value for thousands of years found that technology had finally reached a point where they weren't needed. Instead of needing 20 million of them working in our economy in 1920, by 1960 there were only about 4.5 million. While they were able to take advantage of the previous technological revolutions and become even more useful because of better technology in the past, most horses could not survive the invention of the automobile.

The human intellect has enabled our species to keep up with technology for at least 100 years longer than arguably the second most productive work bearing species in our economy, but that doesn't mean we will forever. Most likely the fate of the horse in the early 1900's gives us a good view of what will happen to human workforce participation in the next 50 years. Around 9 million horses are still part of our economy today, but if horses had kept up with humans in our economy there would be more like 60 million.

Using the automobile revolution as a guide for what will happen in the AI revolution, human working age workforce participation could be at 15% in 40 years.

40 years from now you will see a totally new set of companies doing amazing things you never even imagined.

Of this we have no disagreement. But these new companies will continue to need less people to do the work. The oldest 5 companies to be added to the DJIA employ 1900 people per $1 billion in revenue. The newest 5 companies require only 1100 employees. Looking at the 10 largest companies by market cap, the companies founded in the last 50 years require 1270 employees per $1 billion in revenue. The companies founded earlier than that (all three of them founded over 100 years ago) require 1850 employees.

You may not see the people being displaced by these changes already, but we all heard them loud and clear during the Brexit and US Presidential campaigns. Unfortunately their rage is misplaced towards outsourcing when the real culprit is a technologically advanced economy that doesn't need their skills anymore. Not enough to justify living wages in their society that is.

Customer service bots and automated vehicles alone have the capability of displacing 10's of millions of workers in a very short time. Whether we look at 100 years ago or just this century so far, it is clear we won't find new industries for many if not most of these workers.

You may have a better imagination than me, but a belief in Santa Claus doesn't make him real.

Re:Failure of imagination

[fluffernutter]

Were those people able to get hired elsewhere?

Your question is complete. The correct question to ask is if these people were able to get hired elsewhere *at the same salary when adjusted for inflation*. To that, the answer is no. It hasn't been true on average since the 70's. Sure, some people will find equal or better jobs, but salaries have been steadily decreasing since the onset of technology. Given a job for less money or no job, most people will pick the job for less; and that is why we are not seeing a large change in the unemployment rate.

Re:Failure of imagination

[gtall]

There is another effect. When the buggy whip manufacturers were put out of business, there were options for people to switch to and new industries were created. However, if AI gets apply across an entire economy, there won't be options because there is unemployment in every sector. And if AI obviates the need for workers, investors in new industries will build them around bots, so no real increase in employment. That and yer basic truck driver ain't going to be learning how to program.

Re:I'm not worried

[Waffle Iron]

He's hired primarily free market people as opposed to corporatist

Free marketers don't generally campaign on a platform of protectionist trade policies and direct government intervention in job markets.

Re:Accounting isn't what you think it is

[fluffernutter]

Seriously? Quite a bit actually.

Then why don't you give at least one specific example.

There is a tremendous amount of judgement that goes into accounting and much of it is anything but rigid.

And this is where a system like Watson shines. While you may be able to fall back on personal experience to make these judgement calls, Watson can easily run thousands of simulations on each set of numbers, based on real world possibilities. You will always be making a guess based on what might happen, but Watson will come close to knowing what will happen.

See there is your problem. Every company is unique in some way.

Sure this is obvious. If every company were not unique in some way, then Watson would only need to learn from one company; thus the need for Watson to learn off of several hundred companies. The reason why the case for AI is hard to understand is because we are not able to fathom remembering every detail of every company and being able to isolate what was done differently in company 237 that allowed it to prosper versus company 938. We would have trouble even comparing two companies down to the detail that Watson would be capable of. Given 1000 companies, Watson will know what every single one did right and wrong. Watson will know where mistakes were made simply because it will be able to find another couple companies in its dataset that did better or worse in a similar circumstance.

This all said, I am far from an AI believer. I don't think AI will really be able to drive a car in the near future, at least not as dynamically as a human. However, most professions will benefit from Watson's ability to understand huge datasets down to excruciating detail and freely be able to pick out specific scenarios that worked in the past. This isn't really even AI, it is just a very organized search engine. The human mind almost does these things backwards through necessity, because we cannot process such large datasets.

Don't worry, Trump has the solution

[Jeremi]

Turns out it's rather simple, really --- just ban computers. He's going to start by replacing computers with human couriers for the secure-messaging market, and move outward from there. By 2020 we should have most of the Internet replaced by the (now greatly expanded) Post Office.

Re:Keynes failed prediction, the 15 hr work week

[avandesande]

The wealthy have been pocketing all the productivity gains.

Re:This is not news or new

[djinn6]

I think you've been drinking too much Ayn Rand Coolaid.

As people are freed from jobs that are highly repetitive, there are always more complex, less repetitive jobs out there because the consumer is always looking for the next big thing to improve their lives/increase their free time/reduce their work load.

And everyone has both the talent and the initial capital to create that next big thing?

Entire multi billion dollar industries have been created after the introduction of the ATM and will continue to be created.

Why does it matter if a dozen people made all of those billions?

If we get to a point where there are 10x more job seekers than jobs, then we can revisit the issue, but right now there are about 5.5 million job openings in the US...

And there are 7.4 million unemployed people, and that's not counting people who want full-time employment but only found part-time work.

The Obama economy was of his own making after the first 2 years due to the ACA and excessive regulation, and, like the Carter economy, it will be unleashed with the next administration.

Yeah and Bush did a real great job...