Can you imagine what it is like living in Raqqa, Syria? Only a few years ago, it was filled with impressive buildings both old and new, and was home to more than 200,000 people. Now, around 80 per cent of the city has been left uninhabitable, according to the UN. A combination of civil war and the battle against Islamic State has damaged it beyond recognition.
But however horrific the situation, it can be too easy to put problems in other countries out of your mind. Such destruction is hard to imagine and so can lead to fewer people contributing to fundraising campaigns. To change this, UNICEF has teamed up with the Massachusetts Institute of Technology Media Lab to use artificial intelligence to bring things a little closer to home.
Using a technique called style transfer, the team created pictures of cities around the world as if they had gone through the war in Syria. Their algorithm uses artificial intelligence to learn what destruction of Syrian towns looks like from different images and then transfers it to pictures of London, New York or elsewhere. All the images are collected .
Advertisement
鈥淢aking things visual, salient and immediate are all powerful influences on thinking, emotions and behaviour,鈥� says at University College London, who isn鈥檛 involved with the project. 鈥淚 can see that this is likely to have the desired effect.鈥�
Simulated destruction
The results are pretty striking. Piles of rubble amass in the streets and windows are blasted out. But will the images raise any more money?
Before the website was launched, the team ran an experiment on the crowdsourcing platform Amazon Mechanical Turk. This involved 1000 paid participants looking at either real before-and-after images of a Syrian neighbourhood or simulated before-and-after images of a war-wrecked US neighbourhood.
They were asked to imagine what it would be like for such a disaster to happen in their neighbourhood, and then asked if they would donate some of their payment to charities providing humanitarian aid in Syria.
The simulated images managed to raise 96 per cent as much in donations as the real images. Though this is slightly less, the team hope that refinements will vastly increase overall donations.
鈥淚t means that we are doing almost as good. We will continue to test how to get the biggest effect,鈥� says at MIT.
In the future, the team will also expand the website to provide images relating to floods, wildfires and other disasters. 鈥淲ith things like climate change, we want to teleport people to the future so they realise that we should do something now,鈥� says at MIT.
