This Week in Tech: Artificial Intelligence Edition

Don’t know about you but the SHIFT offices havebeen obsessed with the Women’s World Cup … in fact we scratched out this post while eagerly awaiting the U.S. v France game. Here are some of the stories that our B2B tech teams have been talking about between matches:

AI benchmarks set  

A group of 40 technology companies, including Facebook and Google, released a set of benchmarks this week that have been designed to measure the performance of artificial intelligence tools, with a goal of helping businesses get a handle on the category. The benchmarks from MLPerf cover image recognition, object detection and voice translation, and aim to help companies find the best tools for their AI projects.

The benchmarks apply to a wide array of platforms, from mobile devices to data center servers and network infrastructure. Davide Schubmehl, an AI analyst at IDC, is quoted as saying the metrics will help decision-making at organizations that have been hesitant to adopt AI. “It’s coming at a useful time, as we’re seeing more organizations move from experimentation to production,” he told The New York Times.

ID by heartbeat detection

MIT Technology Review reported on a U.S. Special Forces project that uses a long-range infrared laser to identify individuals by their heartbeats. Cardiac signatures are apparently as unique as your iris or fingerprint, and are a more reliable way to identify someone than facial recognition.  Known as Jetson, the device uses a method called laser vibrometry to detect surface movement caused by the heartbeat – at up to 220 yards away, and through light clothing like a shirt and jacket.

Jetson extends similar infrared sensors used in medical settings by using a standard device used to monitor vibration from afar in structures such as wind turbines. It works by focusing a quarter-size laser beam on a target for 30 seconds to get a reliable reading, so it currently only works when the subject is standing or sitting still.

According to the team that developed it, Jetson is more than 95 percent accurate, and is expected to be used alongside facial recognition for identification purposes. Longer term, the tech could find other uses, such as scanning for irregular heartbeats and other conditions remotely.

Ethics group says no to facial recognition 

Police body cam maker Axon (formerly known as Taser), on Thursday announced that it would not be adding facial recognition technology to its devices based on recommendations from an independent research board.

According to Techcrunch, the board’s guidance was to not use facial recognition  now, or maybe never. The reasons the 11-member board cited included:

• Facial recognition algorithms are still too faulty to be used ethically
• The technology should only be used with consent of those it will affect
• Real-world political and ethical considerations should be considered before police use of the technology

This is just the latest in a growing resistance to government use of facial recognition – on top of San Francisco’s recent ban, the Massachusetts town of Somerville banned city use of facial recognition this week.

(Un)related pro tip

Benchmarks and metrics can help define your PR strategy. Hit up the SHIFT blog for a primer on The Metrics Execs Want to See.