• Black Facebook Icon

©2019 Ixio

  • Ekow Duker

Man vs Machine



In last week’s blog, we looked at the EU regulations that have been introduced to govern algorithmic decision making. Clearly, the draft of regulations were prompted by feelings of disquiet over the possible negative ramifications of new technology. That led us to ask whether this is a new phenomenon or a sentiment that pre-dates the work of the EU legislators in Brussels.

It turns out that such feelings of disquiet have persisted for years. The idea that science might reduce human free will to something that is… well… not human at all, is understandably the stuff of nightmares. 

Fyodor Dostoevsky, writing more than 150 years ago in his ‘Letters from the Underworld”, itself a provocative title, had this to say:

“All human acts will then be mathematically computed according to nature’s laws, and entered into tables of logarithms which extend to about the 108,000th degree, and can be combined into a calendar… in a flash all possible questions will come to an end, for the reason that to all possible questions, there will have been compiled a store of all possible answers… man will become, not a human being at all, but an organ-handle, or something of the kind.”

The notion of machines subverting human freedoms is well established in popular culture. From the quaint suburban trials portrayed in The Jetsons, to the more desperate battles of man vs machine in The Matrix, humankind, armed with nothing more than hope, courage and goodness, has always won out in the end. What makes the current debate so fraught is that we suspect there will be no happy ending this time. We can’t help believing that the machines will win. Why? Because they're smarter than us, more relentless than us and devoid of any redeeming human qualities. The Financial Times talks about the soaring capabilities of machine learning giving rise to  “Frankenstein-like fears about whether developers can control their creations.”

Yet there is another side to this story and it’s a powerfully uplifting one. The proverbial man on the street seeking to run his small business more efficiently, can now make use of advanced analytics capabilities that only a few years ago, might have been the sole preserve of heavily classified government research laboratories. Abby and Tait Larson who run the wedding blog site StyleMePretty, trawl through thousands of wedding photos a week aided by a neural network that categorizes the photos for use on their website. The technology is not perfect but it allows StyleMePretty to scale up by categorizing large volumes of photos at a much faster rate and at a much lower cost than they could possibly achieve by employing a human. 

In our own work in Financial Inclusion, we design and deploy machine learning algorithms that help make financial services more relevant to poor communities and thus in some small way, transform people’s lives. Likewise, machine learning has positive applications in fields as diverse as oceanography, archaeology, medical diagnoses and equity trading where it propels human investigation and insight to hitherto unimaginable levels.

Because the one progresses much more slowly than the other, legislation tends to play catch up to technology. We laugh now to recall how the very first motor vehicles had to have a man walk in front of them and waving a flag to warn citizens of the passage of a possibly dangerous, new fangled contraption that few people understood. A self-driving car in that era would likely have been seized and burned in a bonfire to the righteous cheers of the townspeople.

So in the absence of legislation, as is the case in many of the jurisdictions we work in, can the developers (read data scientists) be trusted to control their creations? Like any other sphere of human endeavor, there will inevitably be ‘good guys’ in white hats and ‘bad guys’ in black hats and quite a few greys in-between. As Bill Hoggarth, a big data thought leader said in response to our blog last week, ‘consider what you would do if someone else were using your data in the manner you are planning’. Now that’s sensible advice.