Deep Learning Conference Panelists Discuss How AI Can Make Your City Smarter and Safer

by Kristin Uchiyama

Artificial intelligence is already reshaping our cities, according to a panel of experts gathered Thursday at the Washington, D.C., edition of the GPU Technology Conference — and more changes are coming, fast.

“Our cities are increasingly challenged to bring intelligence and safety to their citizens’ lives,” said moderator Milind Naphade, NVIDIA’s chief technology officer for AI Cities, to a room of entrepreneurs, investors and technologists attending GTC DC.

This kicked off a discussion of how citizens stand to benefit, and what kinds of policies and investments are necessary to bring these benefits about as soon as possible.

Video: One of the Most Powerful Sensors

Cities abound in data and cyber-physical environments with amazing potential for value creation for its inhabitants. Sensors and actuators are all around us: video cameras, environment sensing, traffic sensing, smart meters, vehicles, as well as the mobile phones we all carry. This is big and fast data.

“This is an exciting time because there’s an explosion of enabling technologies. Video is one of the most powerful sensors, but it takes so much bandwidth to move that video around. With NVIDIA you can process data the moment it’s created,” said Naveen Lamba, director of analytics at Grant Thornton.

GTC DC smart cities panel
Milind Naphade, chief technology officer of AI Cities at NVIDIA, left, with John Garofolo, senior advisor at the National Institute of Standards and Technology; Richard Kidd, deputy assistant secretary, U.S. Army; Naveen Lambda, director of analytics at Grant Thornton; and Russell Brooks, director of smart cities at Transport For America.

The Camera Tipping Point

There are 500 million cameras worldwide today and that number is expected to grow to 1 billion by 2020. But where’s the inflection point where cameras are not just viewed as a forensic, after-the-fact investigative enabler but more as a real-time, decision-making source of data?

“I think we’re there — the first decade of the century was the big data decade and now we’re squarely in the big thinking decade,” said John Garofolo, senior advisor of Information Access Programs for the U.S. National Institute of Standards and Technology. “The next decade is going to be big context. That means bringing a lot of data together with communications and computing that are tightly integrated.”

Be All You Can Be: AI Can Help the Army See

One of the most unique perspectives on how cities can leverage AI came from Richard Kidd, deputy secretary assistant for the U.S. Army, which has 3 million people living in bases around the world.

“There are traffic jams every day when people try to get into the post and have to stop their vehicles to flash their ID cards,” Kidd said. “We’d really like to use AI to detect the person and be able to drive right through.”

“We’re also very concerned about what types of aircrafts are flying into our airspace and what they’re doing there,” Kidd added. “It would be good to know when it’s just a teen trying to get pictures or when it’s an intelligence organization trying to understand what’s going on in our wires.”

What a Disaster! AI to the Rescue

AI also helps deal with city-scale challenges like emergency and disasters. There’s a lot that can done in terms of planning and simulations.

“There is tremendous potential for simulation and AI in a learning environment,” Kidd said. “The Army makes great use of war games and is in the process of developing a serious game to train our garrison commanders where they’re hit with various events and have to respond. And this can be done on a much larger scale. I think AI has a role to play in training all the players at the federal level from these big natural disasters.”

AI on the Move

To be sure, challenges remain. Panelist Russell Brooks, director of smart cities at Transport for America, told the crowd that he was envious of the U.S. Army because it faced less layers of approval for projects and seemed to be able to scale and replicate easily.

“One of our struggles is that these cities operate in silos and start from scratch,” Brooks said. “There are 100 cities looking at autonomous vehicles and they’re largely doing it by themselves — that’s a challenge in terms of  time, effort and money. We’re working with them to determine how to implement these programs and how, together, we can use AI to get people to and from doctor’s offices, schools and child care.”

“I’ve never heard anyone say that they were jealous of the federal government before, so that’s exciting,” Kidd quipped back.

Interoperability and the Role of Standards

NIST has a long history of developing performance evaluation metrics and tools for applied AI technologies ranging from speech recognition to fingerprint and face biometrics to video analytics.

The challenge is to look inside the AI engines and understand why one is better than another when applied to the same problem. NIST is in the process of rolling out one of five initiatives that focuses on understanding and measuring AI, how bias relates to AI as well as the security, privacy and ethics issues around this.

“This needs to be our focus moving into the future. One of the things we’re trying to do is to leverage technologies that have been created across domains to lift everyone up,” Garofolo said.

The Promise of AI for Smart Cities

New technologies using AI are promising to change the way cities operate, “We’re clearly in an age of AI renaissance. AI is disrupting industries and verticals. It’s all about tapping into the big data around us and extracting actionable insights where AI shines.”