One way to keep track of the pervasiveness of this new species of intelligence is to pay attention to the new devices and software arriving weekly, not those from the mega-labs at Alphabet or Tesla, but those being launched by start-ups.
A few weeks ago, Techcrunch reported on a wall-mounted device that will assist those charged with workspace facility management. The hardware for the system is pretty simple. You mount the VergeSense device on the wall with a bracket or adhesive and plug it in, or it can operate on batteries with a life of up to three years.All images courtesy of VergeSense
One properly placed sensor will cover about 1,000 square feet. The data collected includes people count, desk utilization, heat mapping, and sitting vs. standing. It does this with numerous sensors: PIR (passive infrared) motion sensing, as well as conventional optical, acoustic, temperature, and ambient-light collection. The sensors transmit the information to your network or the cloud wirelessly through Wi-Fi or cellular networks. But these sensors aren’t just portals for raw data sent out for analysis.
Dan Ryan, a cofounder of VergeSense, explained to Techcrunch’s Natasha Lomas that the key to the system is in the pretraining of the analytical models that run on the [sensor] device. “The whole concept around what we’re doing is we’re using machine learning in pre-trained AI modules to do all the processing on the device itself.”
The routines written for the “early phase” deployment of the VergeSense system are designed for commercial office buildings (several Fortune 500 companies in the San Francisco area), but Ryan and his codevelopers assume other capacities because the sensors can learn to process and analyze new information.
Today, VergeSense helps with space planning, room and desk utilization, and even emergency information such as an exact count of every person in the building during an emergency event. The information includes those standing or sitting, and even those who’ve fallen down. Tomorrow, you might want to turn the sensors’ attention to business machines around the building, traffic flows in and out of particular areas on the weekend, or wherever your needs take you.
DEEP LEARNING SENSOR:
Each sensor creates a sphere of intelligence and the more data they collect, the smarter they get. As more powerful machine-learning modules become available sensors automatically learn to detect new types of events.
Those interviewed for the Techcrunch review were careful to point out that the system doesn’t tie its data to individuals. Ryan explained, “We’re not getting any personally identifiable information about anybody. It’s all anonymous counts—i.e., ‘I saw a person’ or ‘I detected an object here or there.’” Facial recognition isn’t one of VergeSense’s smart resources.
AS THE INVENTORY GROWS
A smart system like VergeSense has room to grow, and likely that growth will depend on other developers contributing with new routines for the sensors. That’s been true of many other AI-enabled products. Consider the meteoric rise of app store offerings for Amazon’s speaker/personal assistant Alexa.
In September 2016, the number of apps you could download for your Alexa device jumped from 1,000 to 3,000, according to the Techcrunch blog. Five months later, in February 2017, Wired reported that because Amazon opened up the device to other developers with the ASK (Alexa Skills Kit), “Today there are 10,000 skills available on Alexa.” And growth is still accelerating. Five months later, in a July 3, 2017, posting that revisited the number of apps, Techcrunch announced, “Amazon’s Alexa passes 15,000 skills.” It’s worth noting both of these publications chose to describe the new apps as skills.
Alexa is a speaker. Not too long ago a speaker was something we passively listened to—but not usually for advice. Now, Alexa, Google’s Home, and even Siri for that matter can track and report on our investment portfolios or FitBit health stats, offer advice about traffic patterns, and even test us in a short game of Jeopardy. Now we talk to our speakers and listen to what they have to say as well as what they might play from any of our music libraries.
The success shared by the speaker/assistant devices is an indicator of how dramatically AI might impact the future development of our digital age, and it explains how a technology like VergeSense can be born with an ability to provide its own future generations, simply because it can learn.