Henrik Roonemaa: when computers rate art

Photo: Erakogu

Late Wednesday night, a new fascinating service was announced by Google’s bosses in San Francisco, in front of 6,000 onlookers and over a million Youtube viewers. Many of us have probably been faced with a situation, for instance, when, returning from a trip, all the thousand pictures shot should be reviewed, sorted out, the bad ones (and repeats) deleted and a few dozen kept – if lucky. Most of us lack the time and the will for doing that. Thus, Google is volunteering to do the bothersome job for us. Load the pictures into the Google server and the computer decides for you, which are the better ones.

Quite unusual, to have computers make such qualitative decisions – on ordinary user level, mind you. Google might then, as well, be shown black squares by Malevitš and Lapin – asking the computer, which artist has done a more creative job. It feels like an impossibility, for how will a computer know which square I like most… or, coming back to where we begun – which travel pictures I would count worthy to be kept. So: the mere fact that Google is trying to make a computer guess your preferences, is remarkable.

Even more remarkable is the way they do it. Among other things, they look to see what is happening on the pictures. Do they feature any famous objects? A well-known scene? Are there people on the pictures? Are these people, to the knowledge of the social network Google+, relatives of yours? Is the picture pretty, aesthetically? Indeed, hundreds of people have, for the sake of Google’s self-educating algorithms, evaluated various photos and thereby the computers have gotten a hang on what kinds of pictures people like.

This may sound appalling. But, actually, algorithms of this sort already do a lot of controlling in our lives. They drive planes; they decide which way we go according to the GPS; which books we purchase over Amazon; what music we listen to over Spotify; whose news are shown to us in Facebook.  On the stock markets, algorithms do most of the work for a long time, already. Slowly but surely humanity yields decision making to computers – this being comfortable. Computers are fast, they cannot be emotionally manipulated. Nor bribed. And, should something go wrong, we may boldly blame them, for they cannot defend themselves. Decision making takes concentration, digging deep. Which takes time. And… we don’t have time, because we are so busy keeping track with who, of our friends in Facebook, have added new pictures, or discovered a new fun café.

That same night, the Google chief Larry Page said that rules and regulation often hinder the useful changes. In his opinion, there ought to be a place or two on the Earth where all changes could be tested, free from rules. Considering that three fourths of Estonian budgetary spending is determined by law anyhow, and that we are faced with big problems like population decline, loss of healthy years, the gender wage divide, and low added value of working hours – all easily expressed in numbers – then why not become the first country in the world where, instead of the Parliament, an algorithm makes the decisions? I’m convinced this is a rules-free idea enough – for Larry Page & Co to jump on board and help us work it out.

Author is editor-in-chief if the journal Digi