What drives a cat owner to invent a door that automatically closes when pets try to carry prey into the house? Countless nights of waking up at 3 a.m. to euthanize badly wounded small animals his cat would carry into the house, said Ben Hamm, creator of an artificially intelligent (AI)-assisted pet door called Critterblock. The contraption consists of a wood-framed glass door that uses computer vision to detect when Hamm’s cat, Metric, tries to sneak a critter through the opening.

“It got so bad that I learned to code, learned to solder and hand label 23,000 images to stop him,” said Hamm, an Amazon employee, as he stood on the main stage during an invention demo at the recent GeekWire Summit.

The final day of the eighth annual technology conference drew more than 800 people in policy, tech, business and media to downtown Seattle’s Hyatt Regency for panels, demos and fireside chats around technology’s impact on society.

Hamm was one of four presenters at the demo, akin to a speed-dating event, where attendees later used an app to vote on their favorite invention. While the cat door didn’t place first, Hamm said the invention has mostly kept his place prey-free; it only “unfairly locks (Metric) out 1% of the time.”

Fed up with the unexpected visitors, Hamm began labeling more than 23,000 images of his cat a year ago and fed the data into Amazon’s online machine learning service, called SageMaker. The device is powered by the Amazon Web Services DeepLens camera, which uses a web of computation models to lock the cat door, text Hamm a photo of the detected prey and donate “blood money” to the non-profit environmental organization Audubon Society, said Hamm during a presentation in June. Four cats are now monitored by Hamm’s invention, which saves images to ensure the device will become more accurate in the future.

In an afternoon fireside chat, Jay Carney, Amazon’s global corporate affairs senior vice president, reiterated the company’s plan to propose facial recognition rules that will be proposed to Congress. “Like any new technology, it can be used powerfully for good and potentially for ill,” said Carney. “We’re eager to work with lawmakers and regulators to find that balance.”

He said Amazon has not received any complaints that law enforcement agencies misused Amazon’s facial recognition software. However, a 2016 report by Georgetown Law’s Center on Privacy and Technology found most law enforcement agencies had done little to ensure the accuracy of their facial recognition systems.

Carney said it’s not the tech company’s responsibility to “control how law enforcement agencies use technology,” instead placing the onus on governments and regulators.

The tech conference’s final day demonstrated the pervasiveness of technology in our governments, offices and homes. Now, if only an AI-assisted door would kick kittens out of bedrooms when they attack owners’ feet in the middle of the night.

Copyright 2019 Tribune Content Agency.

Recommended for you