The Dim-Post

July 21, 2016

Make humanity great again

Filed under: Uncategorized — danylmc @ 8:44 am

US economist Robert Gordon makes the argument that robots aren’t going to take all our jobs. He is convincing, although I would add that robots don’t have to replace you in order for business models based on new technology to wipe out your industry. Like, there’s no robot journalists – except for Keith Ng – but the business model for the job has collapsed; now the ad revenue goes to companies like Facebook who provide the platform for over a billion users to generate content for them for free.

What with all my Russian revolution reading I keep wondering what Marx would say about this tendency for ‘workers’ to add value to companies by working for free. Mostly I think about this when I’m checking out and bagging my own groceries at the supermarket.

16 Comments »

  1. Which is why we never use the self checkout. It’s a pain sometimes if you’ve only got a couple of items, but I’d rather the staff had jobs.

    Comment by Stephen — July 21, 2016 @ 9:20 am

  2. As well as @1 – the checkout queue is a great enforced stop where I endeavour to try and just be

    Comment by rodaigh — July 21, 2016 @ 9:42 am

  3. I am always amused by the two different streams of thought.

    On the one hand, all jobs will be automated and there’ll only be “work” for the 1%, who will mostly just make enormous returns from either owning the machines, or from doing a few entertainment or sports type “jobs” based on them being beautiful or genetically talented.

    On the other hand, the health sector (in which I work) spends enormous amounts of time hand wringing about how there will never be enough workers to look after all the retired people as the number of working age people per dependent drops with demographic changes.

    It seems to me that both stories cannot be true.

    My personal view is that many elderly people are wealthy (inequality in NZ is in substantial part a generational issue, and Auckland house prices mean there are many elderly people sitting on $2-3M dollar houses). Those elderly people will need to sell those houses in order to get money to pay younger people to look after them. The elderly selling the houses will eventually drag prices down a bit (good for the young), and the elderly paying the young to look after them will result in wealth transfer from the old to the young. Which is kind of self correcting ultimately.

    Comment by PaulL — July 21, 2016 @ 10:18 am

  4. There are already robot journalists – http://www.theverge.com/2015/1/29/7939067/ap-journalism-automation-robots-financial-reporting

    Comment by James — July 21, 2016 @ 10:44 am

  5. Algorithms are doing more than you think
    “Using a mixture of algorithms and human stylists, London-based Thread works on the premise that a lot of men hate clothes shopping, however much they like stylish attire.
    techcrunch.com/2015/08/28/suits-you-sir/

    The business model works with 50,000 users per stylist employed ( currently)

    Comment by ghostwhowalksnz — July 21, 2016 @ 11:08 am

  6. The business model works with 50,000 users per stylist employed ( currently)

    So automation and algorithms work for activities where the customers treat the services as commodity and are time poor, or don’t see the need for significant human interaction.
    This works great for a lot of things i.e. banking, advertising, being a commodity “stylist” (whatever the fuck that is – sounds like just an advertising channel to me).

    What robots cant do (yet) is anything that requires complex non-mathematical problem solving / interpretation or any form of personal interaction.
    This is a fairly broad market.

    Comment by Gregor W — July 21, 2016 @ 11:28 am

  7. Danyl, I think the bitchy comment about Keith Ng is a little bit unfair. If current examples are anything to go by, robots will be built with genuine [people] personalities.

    Comment by Robert Singers — July 21, 2016 @ 12:01 pm

  8. I dunno. The Robert Singers robot seems to operate fine without one😉

    Comment by Gregor W — July 21, 2016 @ 12:19 pm

  9. Well Gregor, training AIs on a mixture of Carry On book adaptions and Sven Hassel novels will always have variable results🙂 (http://arxiv.org/pdf/1511.06349.pdf)

    Comment by Robert Singers — July 21, 2016 @ 12:35 pm

  10. Yeah I read that too. I’ll be really bummed if we don’t get the robot job-pocalypse that has been promised to us. Productivity growth is good, it’s the politics of balancing out the costs and benefits that need serious improvement.

    I always use the self check-out when I can. Every unemployed person makes the problem harder to ignore.

    Comment by James Green — July 21, 2016 @ 3:45 pm

  11. You don’t need robots/AI agents to replace all labour to see massive job losses, merely making some people massively more productive will wipe out a lot too.

    Comment by Chris Bull — July 21, 2016 @ 6:53 pm

  12. This whole discussion is a little bugbear of mine. Trained in computer science, specializing in AI, in the 1990s, and having spent about 10 years of my 20 in software development on variants of AI software, I can say that I think it’s massively oversold. AI is the unimaginative brainchild of science fiction writers, for the most part, and real AI systems are far more prosaic and way, way more expensive than they typically depict. Always people look at it from the point of view of replacing some mundane human task, little appreciating that it often is the most mundane tasks that are the hardest to automate. All for what? Such work is the worst paid anyway! It’s cheap as chips to get a car driver or a person to work on a call center.

    The IT developments that really transform the world aren’t just automations of things we can already do. Which is why it’s so hard for science fiction writers to ever really foresee any of it. They fall back on the lazy expedient of the “robot” that can do everything we do, which in some way transforms society. But we already have those robots. They’re called exploited labour. They are cheap to produce, self maintaining, self teaching. Of course there’s little that transformative about them, but then the vision of life in which everyone is an aristocrat being served hand and foot by various cheap slaves isn’t really that visionary. Yes, it’s nice to have those comforts, but it fails to notice that one of the nicest things about it was always the lording it over others, the unattainable scarcity of such a lifestyle, the entry to the aristocracy that it meant. Now, everyone’s got this incredible chariot that takes you everywhere far faster and more comfortably than anything 200 years ago could ever have done. That’s a device that’s transformed humanity, totally changing the way we arrange every aspect of our lives. But what do the sci fi visionaries look to add to it? A driver.

    In a similar way, this supposed AI explosion that’s only been waiting to happen since it was first thought of, and still never eventuated, sat there as the amazing transformation predicted throughout my childhood, and yet the simple mobile phone, which actually did transform the way we interact with the world, was seldom suggested at all.

    In one of my own projects, aimed at improving on/replacing the schedulers of milk collection trucks, that part of it was extremely expensive and I still think it’s borderline if it saved much money at all, and there was quite a lot of human cost involved in the transition. But the one thing that *really* improved their ability to pick up safely, on time, reliably, was a mere addendum, something I cranked out in a few months – the GIS mapping system that put a trip map into the hands of every driver, ever run. I went from zero to hero with that with the drivers. The management never really appreciated the enormity of it, because, being bean counters, it was hard to see the dollars saved in providing a massive tool like that. It was just an offshoot of the necessity to map every road in the region in order to build the road database that the algorithms I wrote used. There were a great many things more that could have been done with that, had I ever continued to work with them on it (their eyes did boggle at the sight of being able to visualize the data that pertained to all of their thousands of farms), but unfortunately the drive was always about saving cost rather than actually building new ideas, increasing productivity in imaginative ways.

    But I had been pretty much put off AI as the software of the future, having actually personally worked on a large project from conception to delivery. At all turns it was a solution looking for a problem.

    Similarly, every early idea for how to make a better search engine involved people trying to do AI systems that categorize human knowledge. But the search engine of the future simply didn’t work that way. The idea itself was not as good as the rival that Google conceived, of simple looking at how the pages linked in and out as the way of deciding how far up the search list they should appear.

    I don’t see driverless cars as an amazing future technology. I see them as a long slow process, with a huge number of issues to iron out, all to produce something that we already solved. We already know how to get cars around without crashing into things, and it’s not really very expensive. They’ll probably eventually come, and we’ll marvel at how what we did was transform the road system into a technologized version of the train system, and wonder why we didn’t just make more trains, that transformative technology that does actually address the traffic problem.

    Comment by Ben Wilson — July 28, 2016 @ 10:10 am

  13. Driverless cars are a utopian solution looking for a problem. People forget why Top Gear was popular, and that is despite what the online urbanistas would have you believe, outside of the worst parts of trhe commute most people actually enjoy driving.

    Comment by Sanctuary — July 28, 2016 @ 10:16 am

  14. >Driverless cars are a utopian solution looking for a problem.

    Yes, they’d be nice to have. So long as I’m the only one. The moment other people are involved, suddenly I’m not so sure. Don’t know if I want other people to be able to program a vehicle to travel the streets, actually. I feel much the same about drones flying around too.

    Comment by Ben Wilson — July 28, 2016 @ 1:28 pm

  15. I think the key here is to recognise that robots have a place but that human error can be easily rectified whereas computer error, especially if the computer is in a sole charge position, may not be easily rectified. I would like the self service checkouts at supermarkets to be reduced for that very reason. There’s only one or two members of staff monitoring six to eight checkouts, and these staff members aren’t always there. You may not have any bags left at your counter, or the machine may be being difficult, or it may have paid out the wrong change. Likewise, with ATM’s I would like to see a reduction and an increase in banks opening on the weekends and maybe at nights. I am happy that there has been an increase of some banks opening on the weekends but this still isn’t enough to reflect the customer demand for human-to-human contact in their banking life. Some banks could close at 4.30pm and reopen at 6.30pm for two or three hours; others could open at 7.00am or 8.00am instead of 9.00am.

    I happen to have worked at Sanford Limited in Christchurch for two years. I worked as a Pre Spiral Assistant which was basically mussel grading. The whole factory staff of 232 people was made redundant in April 2015 following a restructuring which began in December 2014 and which included our top man Terry being transferred to work at the Havelock plant about three months before the redundancies were announced. The company paid out more than $2million in redundancy, annual leave, and long service entitlements which I appreciated. It was the type of company where it made sense to replace the staff with the technological common-sense that existed at the Havelock plant because Christchurch was a manual plant where it experienced the sharp end of the stick of human error.

    Sickness meant that in a room which held an average of 14 staff members to grade these half-shell mussels and separate the mussel meat, and it was a very cold room which accounted for some of this sickness, that many days there were three fully trained staff members away, representing more than 20% of that room’s workforce. Staff had to be found from other rooms to make up the shortfall and these staff were not always fully trained in Pre Spiral. On the flip side of that argument is that we were very skilled, talented people, and many of us were quicker and smarter than any computer can be in terms of grading product. You had many people who had worked there for a long time (eight years, 10 years, 15 years, five years) and who were understandably devastated by the announcement. The CEO was there when the company announced that it was suggesting closure of the plant; he had a prior engagement on the day, two weeks later, when the closure was confirmed and didn’t make it to the plant that day. My opinion is that, as the three other members of the top echelon of the company were there on the day of the official announcement of closure, that the CEO should have been there also. He had an expertise and a viewpoint that was unique. He happens to be of the opinion that intellectual property will trump commodities holding and the ownership of buildings and solid assets and I suspect that that may be one of the reasons why the plant was closed.

    It was explained to the media by the company that there was a shortage of mussels coming in and in particular wild spat which the company and the Government are jointly working on to solve but that it did not happen in time for the Christchurch plant to remain open. It was also stated by company representatives that the company may reopen in two to three years once the supply of spat has been sorted out. This is a very simplified way of explaining things, and very far from the truth. The company had been holding back on catching spat for some time, so that they had not produced as much as they could have done when it came to mussels going through the Christchurch factory. Furthermore, the off-season in 2014 consisted of three weeks, with some work for staff in that time. The off-season for 2013, however, lasted eight whole weeks. So coming into 2015 and the announcement of the redundancies, we had actually seen off quite a good productive year, much better than the year before it. (Normally the off-season would last four weeks but it fluctuated wildly).

    The company bigwigs also never intended to reopen the plant, and I have been told that they won’t. This was quite a convenient decoy and relies upon public optimism and goodwill to carry it off, which was in large supply following the announcement because the staff always had much more faith in the company than the company actually deserved. The site in Riccarton is prime real estate and I suspect that it will be sold and the money will go towards improvements in technology at the other plants. To its credit, the Havelock plant expanded as much as it possibly could after the Christchurch plant redundancies. I, due to paying high amounts in Christchurch for my living expenses and due to having days off from Sanford when it rained heavily because of the effect on harvesting during rainy periods, deemed that I was not in a good enough financial position to opt for one of these jobs. I was also put off by the new fangled technology in use at the Havelock plant; the fact that I would only receive my personal annual leave entitlements and no redundancy at all, concluding that maybe they would sign me up to a new contract at the new plant and dismiss me for some frivolous reason within three months; and I was also put off by the fact that, in a small town such as Havelock, it would be locals who got the ‘first rights’ to these new jobs in the night shift and that it would probably be hard to find suitable accommodation there.

    Anyway, just an example to let people know some of the realities in these advances in technology.

    Comment by Daniel Lang — July 29, 2016 @ 10:29 am

  16. I think the key here is to recognise that robots have a place but that human error can be easily rectified whereas computer error, especially if the computer is in a sole charge position, may not be easily rectified.

    Quite the opposite is true in fact. People routinely hide their errors or attempt to blame other or “the system”. Machines don’t.
    Human behaviours are also notoriously difficult to change. You don’t need to explain or justify anything to machines.

    The great advantage though that humans have over machines is that they are cheap to buy, redeploy and as necessary retrain from a task perspective.
    Machines, not so much.

    Comment by Gregor W — July 29, 2016 @ 10:59 am


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Create a free website or blog at WordPress.com.

%d bloggers like this: