Using Google Street View to Predict Voting Patterns
Led by Stanford’s computer vision scientist Timnit Gebru, a team of researchers were able to perform image analysis using Google Street View to accurately predict neighborhood voting patterns.
Using Artificial Intelligence and the 50 million images of street scenes and location data, they were able to find out the make and model of roughly 22 million cars (8% of all in the country), and cross-reference the data with election voting records and Census data to predict various neighborhoods’ income, education, race and voting trends.
With AI algorithms, the team’s software was able to classify the cars in 50 million images in just two weeks, which they suspect would have taken a human 15 years to complete. At this rate, they recommend it will be wise for researchers and policymakers to work together to make data collection efficient while protecting individuals’ privacy, so that AI can continue to be used for such projects.
For comments, questions or concerns, please contact Daniella Soloway
Connect CRE News
Your source for daily news covering CRE transactions and trends. Stay informed on national, regional and property sector news that matters to your business.
Connect Events
Whether digital or in-person, Connect Events set the stage to bring together relevant content with CRE’s most active players to engage, influence and inform.
Connect CREative
A full-service marketing agency dedicated to CRE clients. Combining our CRE background with our team’s business, marketing, communications, technology, to develop and execute comprehensive strategies to create, build, and and grow successful brands.