Ai can find the politics of a place by analyzing cars on the GuGe street view view


Ai can find the politics of a place by analyzing cars on the GuGe street view view

The Google street view is full of cars. It’s a simple and pedestrian fact that artificial intelligence researchers have used the fact to do amazing things. By analyzing models, they can predict the demographics of the cities they study.

For example, the team from Stanford university analyzed whether they saw more trucks or cars in a city. As the number of pickup trucks increased, the probability of a republican vote in the city was 82 percent, while 88 percent of the more cars voted for the Democratic Party.

The artificial intelligence system, when dealing with a large amount of data, will shine, and then predict the contents. In this case, the data came from GuGe Street View in more than 50 million images in 200 cities. From there, the researchers used object recognition technology to pick out cars from other objects in the image. They then had to classify the vehicles, of which about 22 million vehicles, or 8 percent of the total, were classified by model, model and year. To do so, they trained an AI tool called a neural network to identify them. (specifically, they used a convolutional neural network, which is good at processing images.)

The neural network has passed 50 million images in just two weeks. According to a study published in the journal PNAS, this could be unfortunate for people around age 15.

The authors of the study must also find out that car types are associated with factors such as political orientation in the region and other demographic information. To do so, they used a regression analysis (a mathematical and statistical tool) to look at the relationship between vehicle types and the information they get from voting data and census data.

The study’s lead author, is a former researchers at the Stanford artificial intelligence laboratory, Tim nita, bloomington (Timnit Gebru) said that eventually they learned was “surprising” accurately. Their system, for example, predicts that caspar, Wyoming, is a republican. This was supported by the results of the 2008 presidential election, which used the results as an indicator of the real world.

However, she cautioned that their system was not accurate enough that it could replace the actual census – although it could add one. Or, in resource-poor countries, such a method might help to collect demographic information without a comprehensive census.

But the big picture is more than just the car’s image and voting history. Bloom says, this strategy represents a new kind of tools, social scientists can by a large amount of data (e.g., Google street view images) on artificial intelligence (AI) techniques become loose, to take advantage of this tool. Of course, it doesn’t need to focus on cars and politics. Instead, researchers can look at trees and public health, says Gebru. It’s not just street images: it can be filtered through satellite photos.

At the end of the day, having an AI system is an order of magnitude more efficient than just using the human eye


Please enter your comment!
Please enter your name here