by Senior Futurist Richard Worzel, C.F.A.
Reality is messy.
You know how, when the office photocopier gets jammed, you always call that guy in your office who seems to be the only one who knows how to unjam it? That’s the kind of reality that robots and automation can’t cope with.
Reality is unexpected.
Physicists at the beginning of the 20th Century were convinced that they had all the answers, being Newtonian physics, except for one or two minor glitches that remained to clean up. Einstein grabbed onto those minor glitches, and proceeded to start the unraveling of the entirety of known physics, leading to relativity theory and quantum physics, remaking the field into something completely new and unexpected. That’s the kind of problem solving that computers don’t do.
Reality is unpredictable.
While populations can be predictable, individuals aren’t. So dealing with an irate customer, or a patient that needs support, or a grieving husband, are not things that can be learned by rote, then passed off to computer or a robot.
Forty years ago, when robotics was still very much in its infancy, one of the biggest problems was teaching robots how to interpret the information fed to them by a camera, radar, or a sonar signal. Robots didn’t know who to interact with reality. They could only respond to specific situations for which they had a program. As Garry Kasparov, the former human chess champion who was defeated by IBM’s Deep Blue computer in 1997, said that Deep Blue was actually about “as intelligent as an alarm clock”[1]
Computers have traditionally worked by algorithms, which are step-by-step recipes for doing something, or heuristics, which are rules of thumb for doing something that’s more difficult to define. Both can take a long time to create in order to allow computers (or robots and automation, which are physical extensions of computers, just as hands are physical extensions of brains) to come to terms with the real world.
But that’s changing in a couple of important ways. One of the changes involves evolutionary algorithms, neural networks, and other ways of coming up with new ideas. These fall into the broad, and rather messily defined field of Artificial Intelligence (or AI), which we’ll talk about another day. The other is the emergence of co-bots (which, for simplicity, I’ll spell as “cobots”), which are robots that work in collaboration with people, hence the name.
Industrial robots have tended to be massive machines that work in restricted areas to perform heavy tasks, such as automobile assembly. They have to be kept separate from humans because they work according to programming that didn’t take account of the harm they might do to humans. While very useful, computer scientists and experts in robotics have come to see that segregating robots from humans has vastly reduced the flexibility and value of the robots.
Cobots have developed that are lighter, moved more deliberately, and have sensors that cause them to avoid hitting humans, or, if they do, not to hurt them. They are, in other words, designed to work alongside humans, rather than in cages. And this relatively simple change is about to significantly accelerate both automation in general, and robotics in particular.
Where Robots Go
The first change this brings about is it means that cobots don’t need special working spaces, but can be used almost anywhere it’s safe for humans. This means that most workspaces, including offices, restaurants, low-tech assembly lines, stores, hospitals, and many more locations, may now be automated if it makes sense to do so.
Next, because cobots aren’t massive machines, they tend to be cheaper than industrial robots, and hence more attractive to potential purchasers.
Baxter, built by Rethink Robotics, was one of the first cobots on the market, and currently retails for about US$25,000 (plus optional extras). This is typically approaching what a minimum wage worker would earn in straight wages. If you add on any benefits, such as medical insurance, pension benefits, or unemployment insurance, Baxter may actually be less expensive to buy than it would cost to pay a minimum wage worker for one year.
And Baxter has an advertised useful life of three years, which means that the 2nd and 3rd years would be gravy from a business owner’s point of view. Indeed, a recent article in the IEEE[2] Spectrum publication estimated that Baxter would cost the equivalent of about $4/hour – far less than a minimum wage worker, even without including benefits. And cobots don’t get tired, sick, or have personal problems that employers must cope with. All of this makes cobots financially interesting to business owners and managers.
Better with Humans
But beyond simple cost, cobots also increase the productivity of the humans they work alongside, largely because humans and cobots complement each other’s strengths. Robots are good at doing the same things over and over again, and doing them precisely. They can also be used in risky environments we don’t want to expose people to. It is, after all, easier to cope with the aftermath of a dead robot than a dead human.
On the other hand, humans aren’t particularly precise, get bored with doing the same things repeatedly, but are good at stepping into situations that require initiative, an ability to examine and diagnose problems, and then find a way to fix them, even if it’s only unjamming a photocopier.
Going back to photocopiers, they’ve become remarkable pieces of automation, capable of scanning, adjusting color balance according to set parameters, photocopying on both sides of a page, collating, and stapling or binding the results, delivering the finished product in neat, stacked, little piles, all done quickly, and with a minimum of fuss. What photocopiers are not good at is opening new boxes and packages of paper, riffling it so it separates slightly and feeds more easily, replacing toner cartridges, clearing paper jams and clogged ink jets or printer drums, and figuring out why the photocopier isn’t working the way it should.
It’s not that a robot couldn’t be built and programmed to do such things. It’s just that these are the awkward, messy parts of the reality of the photocopying process, and are more efficiently and effectively done by humans, whose evolution over millions of years have led to them being superb problem solvers, artificers, and manipulators of oddly shaped things.
So, why not have humans do the awkward, difficult-to-define parts of the process?
Likewise, why not have a robot (or some other form of computer automation) perform the parts that require precision and repetition, or involve risks that would be unacceptable for a human?
As one cobot client in a promotional video puts it, “[Sawyer the cobot] is really there to do the hard labor jobs that we really don’t want the people to do. We want them to use their brains a lot more and [do] a lot less of the physical labor out there on the floor.”[3]
https://www.youtube.com/watch?v=VerSi4uMXD0
The result of this collaboration is that both humans and robots perform better, and the hybrid result is better than either one going it alone.
And that isn’t just nice theory. It’s been verified in the real world:
“In a human-machine study conducted by MIT researchers at a BMW factory, it was shown that teams made of humans and robots collaborating efficiently can be more productive than teams made of either humans or robots alone. Also, the cooperative process reduced human idle time by 85 percent.”[4]
Why Does This Matter?
Or, to put it another way: So what?
Well, there are several “so-whats” about the emergence of cobots.
First, their cheapness means that the market for them will expand much more rapidly than earlier kinds of robots. And the fact that cobots don’t need a special, risk-free environment means they can be put to work almost anywhere that’s clean and dry. (And I’m sure special purpose cobots will evolve that are suitable to more challenging environments.)
They do relatively simple tasks, such as moving things from A to B, or checking manufactured items for defects, then removing those that aren’t acceptable. And because cobots perform relatively simple work, they’re relatively easy to program, often by having a person physically move a cobot through the motions you want it to make – showing it what to do, in effect.
Rethink Robotics, the makers of Baxter & Sawyer, quotes one of their clients as saying their robots are “stupid simple” to program, just by moving them through their paces. And because they can be designed to work at almost any scale, they can be used in fine, precision processes, such as manufacturing involving electronics components, where they can employ small, or even microscopic waldoes (manipulators).
All of this means that the first “so what” is that such robots can, in the right circumstances, move very quickly into a work environment and become productive. Since they’re not intended to replace people, but to complement them, they will have humans overseeing their work, and can rapidly get up to speed. And the economics and their precision mean that they can pay for themselves rapidly as well.
The Acceleration of Automation
Because of the first so-what, automation will spread much more quickly than even devotees expect. As more and more organizations, both profit and non-profit, see examples of cobots being used, and run into people who have worked with them or used them, the idea of using them will spread. And as the economics and improved results of using cobots become more widely known, the interest in cobots will spread as well.
So the 2nd so-what is that cobots will advance the entire field of automation much more rapidly than might happen otherwise. And because humans often have to see a success before they will consider experimenting with a new idea, cobots will spread the idea that robots in particular, but automation more generally, can be worthwhile, can be non-threatening, and may be worth investigating.
This will also mean that that the speed with which robotics and automation spread will catch people by surprise. Within 10 years’ time, robots in the workplace won’t even rate an eye-twitch, let alone an exclamation.
The Widening Gap Between the 1% and the 99%
A bigger, broader change will be that productivity will increase, and that’s generally a good thing. Higher productivity growth is perhaps the single biggest driver of prosperity. But, as has always been the case, this prosperity won’t be evenly distributed.
For the human coworkers of cobots, increased productivity may mean they can do more interesting, more intellectually stimulating, and potentially better paying work. It will probably also mean that organizations will either be able to get more results with the same number of people, or the same results with fewer people. In the first instance, everyone benefits, especially the owners. In the second case, the workers that remain may have better working conditions and possibly better pay, the owners will benefit from better results, but those workers who are displaced will lose their livelihoods and be far worse off.
Over a longer period of time, this will exacerbate the divide between owners and workers, or, in the popular cliché, between the 1% and the 99%. This has been going on for some time, but the acceleration of computer intelligences, robots, and automation will likewise accelerate the widening of this gap.
I’ve written about this before, going back more than 20 years, and this will eventually lead to a society of aristocrats and peasants – unless we choose to do something about it. That’s a subject I’ve written about before, and will return to again, but isn’t directly germane to this discussion.
Death by Inattention
There is a largely unnoticed, certainly little discussed downside to having robots (or computers) manage important processes under human supervision, and it is best illustrated by self-driving cars.
There’s a categorization for different levels of self-driving cars (also called autonomous vehicles, or AVs), ranging from Level 0 (no automated features) through Level 5 (fully automatic, and capable of dealing with any highway situation). In my mind, the most dangerous levels are Levels 3&4, where the car is fully capable of driving on its own, but where the driver should be alert, and prepared to take over and deal with ambiguous, unusual, or emergency situations.
The problem is that in such situations, drivers are unlikely to be paying full attention. If you are on a long-distance journey, and allowing your AV to drive itself while you are theoretically monitoring what’s going on but not actually doing anything, then the odds are you’re going to be bored. Worse, you probably won’t be paying attention, certainly not the concentrated attention you would be giving to the task if you were driving the car.
And since emergencies tend to arise quite suddenly, and usually without warning, the odds that you could go from being bored and inattentive, to fully alert, completely aware of what’s going on, and capable of taking decisive action in a complex situation, are, in my opinion, vanishingly small.
Likewise, a human overseeing a robot in a dangerous or critical situation is likely to be lulled into complacency by the vast amount of time spent doing routine things. As a result, when an immediate or instantaneous need to intervene occurs, it’s unlikely that the human will be able to shift gears fast enough, and respond quickly enough, to the change from boredom to full engagement.
This is a risk with automation or robotics that I haven’t seen discussed.
“Death to Robots!”
A more speculative so-what is whether there’s a backlash against automation in general, and robots in particular. Certainly there have been people who have been vocal in their dislike for things like self-check-out counters in supermarkets, but that’s not the same thing as, say, people boycotting supermarkets that have them.
Part of the reason for this is that automation has been slowly creeping into our lives for as long as any of us can remember. The difference in future is going to be the speed with which automation accelerates, and how pervasive it becomes. And, of course, robots are much more noticeable, and just plain flashier than a self-check-out counter.
And there is historical precedence. The Luddites of 19th Century England were textile workers who smashed automatic weaving machinery that they believed were threatening their livelihoods.
I expect, but am not certain, that there will be an anti-robot, anti-automation backlash, a kind of “We don’t buy from machines” movement. Where or when this will emerge is beyond my abilities of seeing, but I would be surprised if something like this doesn’t happen, even though I don’t know how or when it might.
So What Happens Next? And What Should We Do About It?
Everyday robots, led by cobots, will start appearing in businesses, hospitals health care clinics, and other places where you will see them, slowly at first, then with increasing frequency. They won’t have the speedy take-up of, say, smartphones after the introduction of the BlackBerry or iPhone, but they make too much economic sense for people who do things to ignore them.
So the first thing to do is to watch how quickly cobots start to appear in your life, or in the businesses and organizations you encounter. They won’t be front-of-the-house at first, but if you keep your eyes open, you will probably start noticing them in the next 3-5 years.
Next, think about how cobots in particular, but robots, AI (such as IBM’s Watson), and automation in general might be of use to what you do, the organizations you’re involved with, and to your work specifically. As I’ve said before, if you can’t beat ’em, co-opt them.
In particular, if you are someone who does repetitive work, even if it requires specialized skills, start thinking about what you could do and would do if, suddenly, there was a robot working with you that could do the routine things. This would free you to, as the video above of a Sawyer cobot user said, “use your brains a lot more.”
What could you accomplish for your organization if you had more time to think about what was going on? How could you improve the way your organization ran? What tasks would you like to have automated so you don’t have to do them?
Or, to focus this further, imagine that your boss came to you and said, “We’re getting a cobot in here to do the routine stuff you’ve been doing. Take a couple of weeks and write yourself a new job description, then come back to us and let us know what role you’d like to have in this organization.”
What would you say to them? What would you propose?
The One Thing You Can’t Do
The one thing you can’t do is turn back the clock, and pretend that this isn’t happening, or that it won’t happen to you.
If you own or run a business, or a hospital or health-care clinic, or a non-profit organization, start asking yourself how you could make your people more valuable by relieving them of routine tasks, and how you could improve your operations by using cobots.
If you’re an employee, start thinking about what a cobot could do for you and with you that would make you more valuable to your organization.
Because, like it or not, there are cobots in your future. And they’re coming faster than you expect.
© Copyright, IF Research, May 2017.
[1] Doherty, Mike, “Garry Kasparov’s next move: teaming up with machines”, Toronto Star, 8 May 2017, https://www.thestar.com/entertainment/books/2017/05/08/garry-kasparovs-next-move-teaming-up-with-machines.html
[2] The IEEE started in 1884 as a group of electrical engineers, and now is widely considered to be the world’s leading society of technologists.
[3] “Sawyer at Work: Customer Success Story – Tuthill”, Rethink Robots website, viewed 10 May 2017, http://www.rethinkrobotics.com/sawyer/
[4] Tobe, Frank, “Why Co-Bots Will Be a Huge Innovation and Growth Driver for Robotics Industry”, IEEE Spectrum website, 30 December 2015, http://spectrum.ieee.org/automaton/robotics/industrial-robots/collaborative-robots-innovation-growth-driver
