Applications of AI and ML range from frivolous to serious, and a lot of applications these last two weeks lean toward the more serious, but let’s start with something somewhat frivolous.
The world’s weather encapsulated in Google’s terrarium
Imagine there’s a little terrarium in your living room that can mimic the current conditions of any place around the world. Just ask it to show you what the weather is like in Dhaka, and there it is, recreated in front of your very eyes. But . . . you don’t have to imagine it. Google just created a terrarium that does just that. Watch it at work here:
Is that six-pack plastic netting around the bird’s neck?
The World Economic Forum said the Pacific Ocean’s garbage patch is three times the size of France, and it contains 1.8 trillion pieces of plastic. One man—who ten years ago was horrified by how much floating plastic trash he saw in remote parts of the ocean—recently started using AI-powered drones to take aerial photos of trash on beaches to build the world’s largest plastic-trash dataset. The short game? Help with local clean-up efforts. The long game? Build a system that can chronicle the spread of plastics and potentially trace the large-scale dumping of plastics back to an original source. Watch the video here:
That’s an altercation if I’ve ever seen one
Beyond identifying trash, drones are now being trained to spot violent behavior in public places. The dramatic music notwithstanding, this video below shows how drones can identify violence from the sky. So far the researchers working on this application have been able to identify, with 88 percent accuracy, violent behavior in videos. Watch the video here:
Researchers use Wikipedia to identify the beginnings of online arguments
Law enforcement combats human trafficking with facial recognition
United States’ DHS moves ahead with facial recognition tech at border
Researchers use AI to preserve Australia’s indigenous languages
MIT researchers employ AI to “see” body movements through walls
Uber’s head of AI and Harvard team up to identify wildlife in camera trap photos
ML identifies specific gene marker that confirms asthma
Two students use ML to predict potential wildfires
Uber’s patent shows how it aims to identify drunk people before they get in an Uber
US border “wall” may end up being an AI-powered sensor system
Are you trying to build an AI/ML product? Here’s an O’Reilly article detailing lessons from people who’ve done it.
Not entirely clear on what a neural net is? Here’s a series of pretty good videos that explain the tech underpinnings.
If you like what you learned in our newsletter, share the love with others. Send them here to subscribe.