These days we are talking a lot about the effects of robotization on the labor market and the effect of drones on aviation safety. Interesting, but both are little more than a sideshow of something else that is happening underneath our eyes. What if we no longer need the people who write the codes, the algorithms? Will I rule the code? Or will the code rule me?
Skipping the code?
Until recently, I thought I had survived the digital revolution quite well and probably profited from it. Not like my little brother though, who is not like me from the IBM typewriter generation, but from the first generation writing code for an Antari. He now is a ‘software architect’ of apps for the ski slopes of New Zealand and the US. If anyone is a winner of the digital revolution, he is. But now there are signs I might come back into in the game and change roles with him – that is, if the present digital transformation does not make us all redundant first.
What I am getting at, is the likelihood that before long you do not actually have to be able to write code or know your algorithms, but can guide or ‘train’ them to do what you want with them. Let me try to give an ‘innocent’ example. If right now you want to make a trip from, say, your place in the US to southern France, but do not really know where to go or what to do, you have the choice to either ask a lot of questions on google and visit a host of websites or apps, including airbnb, or go to a real life human being at a travel agency to plan the trip for you (but who is probably never been there herself and has to resort to google and websites herself). In the near future, you can just say to your mobile device: ‘Plan a trip to southern France’. It still might come up short on the sort activities you want (though a visit to a rock festival may be part of it if you have done so before), but basically asking the question sets something in motion that is equivalent to the development of ten apps and twenty websites now. If the software does not get it, you do not rewrite the code, you coach it towards a better understanding by telling it for instance about your first summer love. After all; when was a holiday more memorable. No? Well, maybe you should not go back to any time before your honeymoon holiday.
Maybe I’m amazed
The specialists are calling this code creating code a part of the ‘deep learning’ that is now taking place. It is also connected to a grand term like ‘Artificial Intelligence’ (AI), but these are grand words that do more to hide than reveal what is now taking place.
First just a little bit more about what is happening now. Not because I pretend to be the expert, but because by writing it down - and you reading this – something can be learned that we need to be aware of, in the line of the Paul McCartney song: Maybe I’m amazed.
We used to think that we could get closer to the working of the brain by literally hardwiring a computer the way the brain works. We never got close. It takes up so much energy that you end up with a meltdown. Until recently we thought we could come closer to the brain by letting thousands upon thousands of people writing line after line of code. We would stash up the lines of codes until it covered everything. Some people warned about building a tower of Babel. They asked what the standard language should be, and then they wondered who would have final authority over the algorithms. Is 'Open Source' an alternative? Again it turns out that it does not work that way.
Shooting the gorilla’s
This amazing thing is what is happening: code is already learning how to write code; algorithms are leading to other algorithms. When engineers from Google try to find out how their search queries actually work, they ‘dive into a sea of mathematical formulas’. Most of them never are written by the human hand, they are ‘made’ by other formulas. They are the result of combinations of combinations. Though they might not yet be called ‘thinking’, the sum of it all goes way beyond our time and capacity to read it all, and in fact beyond our human comprehension. It is a new reality, a commodity, ours to use or abuse.
This was most clearly proved in an incident not long ago, when Google suddenly discovered that some ‘black people’ were tagged as ‘gorillas’. Immediately it tried to undo the damage, but the only thing they could think of was to eliminate any possibility to tag anything as ‘gorilla’. There were just too many unknowns in the code to solve a problem without creating more problems.
The awareness of awkwardness
This awkwardness is more revealing than the initial mistake. As long as we can go along with the algorithms, it all seems fine. We are doing nothing but strengthening our own preferences. It is when you interfere with this, that human awkwardness reveals itself. This is also the case with Facebook when it more or less got caught preselecting content for its news function. Very embarrassing for Mr. Zuckerberg’s quest for legitimacy, but in this case he might have been trying to do the right thing. In my blog I stated that you could not really blame Facebook for what, as a business, it was created to do, but that this is nog good enough when it comes to protecting public values. You need people and their democratic checks and balances for that.
But where does this leave me and my public values, when it feels like much of the coded world is already growing beyond human reach? How to tame what we created without making things worse in this interconnected world of code? How to tease it into something that fits our limitations and sensibilities, how to cajole it into some form of acceptable behavior, however defined? How to teach it? How to teach it to deal with our human weaknesses?
The questions all have the same answer: we can’t. We cannot expect to be better than what we create.
Where to go from here?
Now hold on. Before we land up in some kind of Monster of Frankenstein discussion, where are we, really? Just the other day I tried to get the Siri-application in my iPhone to search for a phone number on a website. I got hopping mad at the the stupid ‘girl’ and tried to teach ‘it’ by shouting abuse. It tactfully remained silent before repeated its question. I think most devices and applications will remain numb at our efforts to goad them with anything else then written code and exact ‘if A than B’ instructions. As in former transformations, we move through a world at different speeds, the old along side the new, the opportunities in sidestep with the risks. And when I have no true answers to the big things happening now, what better thing to do than challenging my awkwardness myself? How to go, as Wired intriguingly states, ‘to go from commanding my code to parenting it’?
So let me go back then to what triggered this blog: my own ability to transcend my ignorance of coding when I read that soon I might skip the coding and go for telling the software what to do. What will I do with that luxury?
Finding the trainer in me
First answer: I will challenge myself to think. To think of what I really want my software to do. Preventing myself from being distracted, is probably the key to that.
True contemplation might lead me to conclude I don’t need anything from my software, but knowing my creativity, I will probably think of something. So the next question might be: for whom am I doing this? Just for myself, or also for others. Do I give my instruction just to myself, or also for others? Is this for business or for pleasure?
You know what? I think I want to approach this whole question of training my software like the trainer that I am. I will think of what I want to transfer: knowledge, skills, insights. I will think of how I want to transfer it all: with words of course (probably too many), but also with images, storylines, questions. And I will want to make sure that what I want to convey, also comes across as such: I want to be able to check what has been learned.
Ruling the Code
Now I don’t know about you dear reader (glad you are still with me), but I think that in the coming months I am going to think about things to tell my software, while trying to prevent the need of a second honeymoon. From writing this blog, I have already deduced that it is going to take some wisdom on my part. The quality of my thinking has to be reflected in my training of the software. As always with training, you have to take your subject as seriously as you take yourself. I can only rule the code, when I rule myself.
Peter Noordhoek
A bit of literature:
The End of Code. Soon we won’t program computers. We’ll train them, Jason Tanz. Wired, June 2016.
Inside OpenAI, Elon Musk’s wild plan to set Artificial Intelligence free. Cade Metz, Wired, April 27, 2016. http://www.wired.com/2016/04/openai-elon-musk-sam-altman-plan-to-set-artificial-intelligence-free/
New Vision for Education: Fostering Social and Emotional Learning through Technology, World Economic Forum, 2016. http://www3.weforum.org/docs/WEF_New_Vision_for_Education.pdf
The End of Reflection, Teddy Wayne. New York Times. June 11, 2016 http://www.nytimes.com/2016/06/12/fashion/internet-technology-phones-introspection.html
When digital becomes human. Klantrelaties in transformatie. Steven van Bellegem. Lannoo BV, Belgium, 2014.