Automation – all the way down

Machines struggle with the things that people find easy. Talking, seeing and walking in particular.

  • Robots already dominate in factories, where the work is repetitive and comes to the robot, and safety issues can be easily contained.
  • The best of these robots (in car factories, say) can retool themselves to complete several routine tasks in the same place.
  • Most of these robots can’t see, and rely on perfect timing for everything to work.

In offices, shops and even warehouses, where the work is not quite routine enough for blind robots, they can’t compete with humans.

But they are getting there, and if you expect Moore’s law to continue (which some people don’t) then they will quickly get better.


Every year, Amazon runs a competition for a warehouse robot, based around picking boxes from shelves.

  • This is not an easy task for a robot.
  • There are shelves filled with boxes of different shapes, sizes and colours, though lots of them are standard brown.

Packed tightly, these brown boxes make edge recognition difficult.

  • Stacked loosely, the robot has to work out how to take a box from the bottom of the pile without destabilising it.

People find understanding this kind of arrangement easy, but robots don’t.

  • It requires enormous numbers of calculations, and the more computing power you can throw at it, the better.

This year’s competition was won by a suction arm powered by AI.

  • All of the robots in the contest were worse (slower and less accurate, more likely to drop items) than humans.

But they were better than last year’s robots – almost three times better.

  • And industry experts expect the machines to eventually be better than humans – perhaps six times better using the current design.

Once a machine can outperform a human – or perhaps even simply approach human performance, if the machine is cheap enough or the human expensive enough – then the calculations begin.

  • Robots don’t need meal breaks or comfort breaks.
  • They don’t even need sleep – they work continuously.
  • They don’t get injured (though they can break down) which means they don’t file compensation claims.

Eventually, the robot works out cheaper to run than the human, and the human job disappears.


The service sector will be the key battleground in Western economies – that’s where most of the jobs are.

The battles already won include ATMs replacing bank-tellers, and self-service in all its forms, particular supermarket check-outs.

  • The more we do for ourselves, the fewer people are needed to do the same work.

Food preparation is an interesting area.

  • We’ve had a burger robot for a couple of years – made by Momentum Machines – but only recently was it announced that an unmanned restaurant based around the machine would open in California.
  • The total labour cost for burger making in the US is $9bn, and McDonalds employs 1.8M people – this is a big market if the robot gets things right.
  • The recent increases in the minimum wage in the UK and US are only likely to speed the adoption of machines like this.

As well as restaurants, the machines could end up in food trucks, or as stand-alone vending machines.

  • They could also support “gourmet” burger restaurants by using high quality ingredients, and moving the money saved in preparation towards front-of-house.

Japan has a chain of 262 sushi restaurants (Kura) where robots make the food and conveyor belts (long familiar in the UK through Moshi Moshi and Yo sushi) deliver it.

  • Orders are through touch panels (also being trailled by McDonalds).
  • There’s also an automated hotel being trialled in Japan.

The automation means that Kura undercuts the opposition on price, but other dimensions of competition are possible: speed, convenience, hygiene are all candidates.

There’s also the possibility to hybridise the consistency of chain restaurants with the personalisation of standalones.

  • Once one branch understands your idiosyncrasies, this information can be transmitted to all the others.

The second major employment area, and one where many low-skill, low-paid jobs are still being created, is retail.

The three trends here are online (ie. fulfillment direct from warehouses), kiosks and vending machines, and probably less significantly, automation within retail outlets.

The Amazon story is well-known.

  • You can buy almost anything, at competitive prices, and have it delivered for free the next day (with Amazon Prime, which costs £70 a year in the UK, and includes free video on demand.
  • Most people can wait a day – or in the case of big cities like London, where I live, two hours if you pay a delivery charge.
  • For those who can’t, delivery by drones or terrestrial robots is probably just around the corner.

Amazon in theory moves retail jobs into the warehouse, but in practice, warehouses are much more efficient, and easier to automate.


Amazon famously owns Kiva Systems, which makes the warehouse robots that move giant stacks of pallets around.

  • These systems were originally used by a wide range of retailers, but I understand that the newer models are now restricted to Amazon alone.

It’s also worth noting that Amazon (and other online “retailers” like Netflix) are a rare example of where automation in a service industry is being used directly to benefit the end customer, rather than just the service provider.

  • Buying from Amazon is much quicker and more convenient than going to a shop, especially for someone like me, who works from home.
  • I click on a button, and the next day someone knocks on my front door and hands me whatever I’ve ordered.

Contrast this with self-service kiosks in supermarkets, and online travel (hotels and airlines).

  • Here I hope I’m saving money against the way things used to be, but I’m definitely trading this off against extra work for me.

Growth in kiosks and vending machines is the second automating force in retail.

  • We already have consumer electronics (eg. iPad) dispensers in airports and fancy hotels.

The aim here is to combine the advantages of “online” ordering with instant delivery.

  • There will be jobs created in maintenance, restocking, and repair to partially offset the loss of sales jobs, but nowhere near enough.

The third factor is automation in the store itself.

  • The robots aren’t quite ready for prime-time here, but perhaps they will be stacking shelves in a supermarket near you quite soon.

But phones will be used more – to scan bar codes, to research products and to pay.

  • Online virtual assistance will replace in-store staff.

Eventually stores will become one giant-sized vending machine, perhaps a warehouse with an attached showroom of product samples.


We’ve seen today that robots are about to come out of the factory and hit the service sector, particularly in restaurants and shops.

The other big areas for automation will be agriculture, education and health.

We’ll consider each of these in separate later posts.

Until next time.


 

Don’t forget to share this via , , Google+, Pinterest, LinkedIn, Buffer, , Tumblr, Reddit, StumbleUpon and Delicious.

Mike Rawson

Mike Rawson has recently re-awoken a long-standing interest in robots and our automated future. He lives in London with a single android - a temperamental vacuum cleaner - but is looking forward to getting more cyborgs soon.

Leave a Reply

Your email address will not be published. Required fields are marked *

Automation – all the way down

by Mike Rawson time to read: 5 min
Hi there - can I help you with anything?
[Subscribe here]
 
More in Economics, Editorial
Jobs
Where did all the jobs go?

It's part of the human condition to assume that the future will be like the past. Such thinking has its...

Close