Wednesday, August 30, 2017

Humans and machines should co-evolve

When I was asked recently to speak on the subject of whether robots are taking our jobs, I said that’s a given.  If you’re wondering if robots will take our jobs, the answer is ‘yes.’  They already are.

Planes have flown with autopilot for over 100 years, and we’re rapidly extending the list of professions under threat from accountants and lawyers, to medicine and engineering.

If you’re wondering if robots will take your particular job, the answer is ‘it depends’ on how routine your job is (The Economist, 2016).  Hint: If your job is routine, no matter how interesting and intelligent you are, you’re at risk of being replaced with automation and/or machine intelligence. 

In short, if you think a computer can do your job, it probably can.

We love our machines
Back in 1982, we were living in Santa Fe, New Mexico, in a little casita next to our landlord’s house.  Our landlord was a professional editor and author, who had purchased one of the first generation word processors (a Kaypro, I believe).  And, I remember telling my writer friend, ‘John, I’ll be the last generation that won’t use a computer.’  And, I meant it. 

Five (5) years later, I bought my first Macintosh (a Mac Plus) when I started working on my PhD and that computer became more than just a machine. It shared its secrets with me as I learned basic programming.  It was there for me when I felt I couldn’t go on. It watched over my work while I was sleeping.  It holds the place of highest honour on my office bookshelf.  Machines in general, and computers in particular, are a part of our family. An extension of ourselves.

So, this is a story not about slaying mechanical dragons. It’s about the quest to be better humans, who happen to work with machines.

Let’s face it, computers are simply better than we are (or will ever be) at certain things.  They land planes more consistently, they are better at facial recognition, they are more accurate at medical diagnoses, and they can of course beat us at games like chess and go.

Worrying about the potential downsides of machine intelligence and automation can blind us to the incredible possibilities and opportunities that such technologies may offer us in the future. We buy and sell, hire and fire, promote and demote with a huge degree of human bias (error). The good news is that machine intelligence is going to be able to help us out.

Rather than worrying about what machines can do, we should worry about what machines can’t do, in part because we still need to build better machines, and in part because that gives us a hint as to what we humans must continue to do well for our part in this co-evolution.

What are humans good for?

Walking down stairs
Even though robots can do amazing things, what they can’t do is equally astounding, like walking down a stairway. (See Nicolas Carr’s Automation and Us). So, let’s celebrate the agility we have as humans, and take care of our human bodies. 

Ever struggle with astrophysics?  Me too. Most of us find math difficult because our cognitive brains are less developed than our social brains. 

As one neuroscientist puts it, our comparatively large brains were developed to solve complex problems, but they were not the problems of thermal dynamics, they were social problems, like who was in charge in the tribe, how to deal with tribal politics, how to build inter-tribal relationships and so on.

      We are social all the way down: Collaboration is a particularly human skill and our social needs are driving us to use technology—maybe too much. 

      Social mediation may be worse than automation: While many people see automation as the worse-case scenario for humans in relation to machines, media over-use and over-dependence is just as bad as or worse than machine intelligence. 

      We used to worry about how much TV people were watching, but nowadays screen time has become far more pervasive than TV ever was.

      It’s not just that we look at screens so much, but that software and devices mediate—that is go between—us and the world around us. 

As comedian Aziz Ansari says in his book, Modern Romance, when observing on-line dating behaviour, he found that too many people spend too much time working on their profiles and not enough time dating! 

At some point, we have to look up from our screens, if we want to live life to the full.

Cultural intelligence
In a world of diversity and difference, humans are uniquely able to seek common connections between ourselves, even though we differ from one another. In a world of mindless tribalism and nationalism, it is important that humans continue to seek connections with others who are different from us. 

IBM’s Watson can make a highly accurate diagnosis of an illness, but only the attending physician can sense the patient’s will to live.

            Health ‘care’ means just that, ‘caring’ for people, even while using intelligent systems to         diagnose, and telemedicine (sensors and tablets) to gather data. We must preserve the human qaulity of kindness and compassion.  

Sadly, when we humans interact, we act more like machines than humans. For example, when I rent a car, the rental company’s representative spends the first half of their time with me gathering routine data, and the second half trying to on-sell insurance and other add-ons.  Why can’t better apps do these things and let the car renting human serve as a host or guide, welcoming me to their city and sending me off toward my final destination with style and grace?  Instead, we have turned ourselves into robots.  As Nicholas Carr observes:

“Industrialisation didn’t turn us into machines, and automation isn’t going to turn us into automatons. We’re not that simple. But automation’s spread is making our lives more programmatic. We have fewer opportunities to demonstrate our own resourcefulness and ingenuity, to display the self-reliance that was once considered the mainstay of character. Unless we start having second thoughts about where we’re heading, that trend will only accelerate.” (Carr, pp. 198-199).

The problem is most organisational work doesn’t involve inquiry or critical thinking. As Josh Bersin says in his Deloitte report,
“The future of work is not simply about using technology to replace people. The real “future of work” issue is all about making jobs “more human”—redesigning jobs, redesigning work, and redesigning organizations so that the “people side” of work has even more importance and focus than ever.”
Computers are good at providing answers, but humans are better at asking questions.  
Coupled with the rapid and radical advances in computation, we need to not just preserve, but continue to grow and develop the human character traits of curiosity and courage, coupled with compassion. These are the principals that underpin programs like Outward Bound.  Along with programming skills, we need education to focus on the attributes that makes us more resilient, more resourceful and more ready to make the hard calls when analysis has reached its limits.  Human character, however, is not just about stoicism or tenacity, it's about being both resourceful and interesting
Here's a test: Your friends would never think of taking a long trip without their smartphone, but if they wouldn't take you on a long trip, why not?  
What can we learn from robots?  What can we teach them?

Lifting the bar:  Don’t expect less from machines, expect more from them!

As Tom Peters says, when software doesn’t deliver, it’s not your fault.  For those of us who were not digital natives, we believe that we are idiots when a computer doesn’t do what we want it to do. Things are getting easier for the app generation, but computers still need to get better.  If my healthcare is going to be delivered remotely via a sensing, analytical nurse-bot, then it needs to be pretty darn good.  I don’t want it to be 80% right, or even 90% accurate (even though I would accept that from a human doctor).  Strangely, and unfairly to machines, we need our machines to be better than we are in order to trust them. And, that requires us to be smart and fussy consumers and ‘employers’ of machine intelligence.

What could possibly go wrong?

Things will change, possibly slower than many predictions, but more rapidly than we can prepare for. Headlines in the past year or so have emphasized the link between robots taking our jobs (technological change) and the economic ramifications, in particular the growing income gap in most developed countries.  This has triggered related discussions on universal basic income, and so on.  All this is good, but social changes are unlikely to evolve as fast as technological ones, unless there is a social revolution that accompanies this technical revolution.  If that happens, all bets are off.

Things won’t change – the tendency toward ‘winner-take-all’ economics in technology might see 1 or 2 companies dominating the robotics sector and progress slows; in NZ we understand how duopolies work, and globally in the 1990s we saw how one company’s dominance of the PC world meant that software development slows, or actually gets worse, when there are no rival options.  So, we need smart and fussy consumers of machine intelligence and competition among providers, or we risk having less smart robots.

Shit will happen
The notion of ‘normal accidents’ introduced by sociologist Charles Perrow suggests that when you have complex systems, there is not just a chance, but a high probability that something can and will go wrong (read, for instance, Flash Boys by Michael Lewis, on high frequency trading).  Machines make mistakes too!

A happy ending

One way to think to about it is not so much that machines and humans are in a zero-sum game, where a job is either held by a human or lost to a machine.  Perhaps a more salient metaphor is that of co-evolution, wherein humans and machines are both evolving together, like some believe dogs and humans have evolved in relation to one another. 

As Garry Kasparov says, the fact that machines are getting better and better offers a chance for humans to get better and better.  “Machines have calculations. We have understanding. Machines have instructions.  We have purpose. Machines have objectivity. We have passion.” 

 As Peter Drucker has said, “We don’t know the future, yet we create it.”  When it comes to automation and machine intelligence, humans and technology will co-evolve, but whether humans continue to develop our unique qualities is totally up to us.

You can listen to this talk given at the University of Auckland 'Raising the Bar' event on 29 August 2017. Podcast.

Or, on my Nine to Noon interview with Kathryn Ryan, 5 September 2017.

Bersin, Josh, of Bersin Deloitte (2017) “The Future of Work: It’s Already Here... And Not As Scary As You Think,”, September 21, 2016,

Brynjolfsson, Erik and McAfee, Andrew (2014). The second machine age: Work, progress and prosperity in a time of brilliant technologies. New York: Norton.

Carr, Nicholas (2014). The glass cage: Automation and us. New York: Norton.

Colvin, Geoff (2015). Humans are underrated: What high achievers know that brilliant machines never will. New York: Penguin.

No comments: