By Umberto Bacchi TBILISI, March 13 (Thomson Reuters Foundation) - From a new driveway to a freshly repainted home, artificial intelligence can spot early signs of gentrification, allowing cities to plan before soaring house prices push residents out of their neighbourhoods, researchers said on Wednesday.
Scientists from the University of Ottawa, Canada, said they used the technology to show in detail for the first time what the process, common to cities around the world, actually looks like. "Gentrification is reshaping our cities, but at the same time it is hard to determine where and how fast it occurs. Now we can see that," said Michael Sawada, co-author of the research published in scientific journal Plos One.
Gentrification - the conversion of working-class districts into more affluent areas - is regarded as a sign of economic development by some and of social injustice by others, as it can lead to higher rents that price out old residents. Sawada and his team used images of Ottawa's streets taken by Google for its Street View service between 2007 and 2016 to train an algorithm to spot building improvements, such as a new fence or a window replacement.
The computer was then able to recognise renovation works with a 95 percent accuracy and pin them on a map highlighting areas where they were more widespread and happening at a faster pace - a sign of gentrification. The team of researchers found at least five areas of Ottawa that were undergoing significant changes they were not previously aware of, Sawada said.
The mapping method, which is free to use and could be easily applied to other cities, improved on traditional tools, like surveys or census data, that tend to be limited in scope and slow in tracking change, he added. "By identifying where gentrification is about to happen or just starting we can start to think (about) how we can address the inequalities it might produce," he told the Thomson Reuters Foundation.
Hyun Bang Shin, a professor of Urban Studies at the London School of Economics, said mapping changes was key to inform city-wide policies to curtail the detrimental effects of gentrification. But further innovation was needed to track changes in countries like India or China, where slums and informal settlements are sometimes levelled to make way for new neighbourhoods but might not be covered by Street View, he said.
Changes to the exterior of a house alone were also not definitive proof that new, more affluent people had moved in, added Anthony Breach, an analyst with the British think tank Centre for Cities. "This data can tell you only so much," he said, explaining property prices in the area might also need to be factored in, he added.
(With inputs from agencies.)