Five Great Government AI Projects

It’s at the peak of the famous technology hype cycle, but don’t let that make you cynical. Artificial intelligence is already making a difference in public service delivery. And unlike other technologies - it’s designed to be a quick learner, so it doesn’t rely on humans to figure everything out.

The tool has potential in a huge number of areas, and will change how many of us live and work. Here are five ways that public servants can already use AI to make a difference.

1. Immigration

 Long queues for security scans may soon be a thing of the past. Train stations in Washington DC and Los Angeles, and Denver International Airport will soon trial a system that scans automatically for weapons.

As with traditional body scanners, the tech uses radio frequencies to look a person over, but this alternative promises none of the privacy issues or sensitivities that we feel in a human-operated system.

Only the machine will see us naked, and is trained to spot weapons and explosives. In fact, “we never build an image that would enable anyone to see anatomical details, so there’s no naked peepshow in the first place,” Michael Ellenbogen - CEO of Evolv, the startup responsible - told the Guardian.

“None of the raw data is stored and none of the data we do keep is traceable to an individual.” Will there be a need for immigration officers in airports of the future? Many countries already have automated gates to let citizens enter and exit the country. Now, they could have AI frisking you on your way to the plane.

2. Tackling Littering

In Japan, some local governments are using a new machine learning platform to track littering on the streets, measuring the impact of anti-littering campaigns and finding out where these occurrences are most likely to take place.

The tool, created by Pirika, uses crowdsourced pictures to help understand what’s happening at ground level. The software then combs through images to create heat maps and report area-to-area comparisons.

This has led to targeted clean-up efforts, enforced anti-litter patrols, and better layout for trash bin locations. However, it is early days for the software, which launched in 2011.

The system cannot yet differentiate between strewn rubbish and garbage that has been disposed of properly. It still requires the “judgement of trained staff,” according to the website.

Regardless of these foibles, imagine the potential of this system when combined with robotics. Singapore is already exploring the use of robots to clean its waterways, sweep the streets, and work in its airport.

Both technologies are at their early stages, but show how quickly human labour could be replaced by an intelligent, automated equivalent.

3. Education

AI has great potential to help students personalise their learning. For example, researchers envision ‘intelligent textbooks’ which can provide students with “immersive learning experiences” such as 3D virtual laboratories that are “adaptive to student’s current state of learning”, according to a research paper called AI Grand Challenges for Education that was published in AI Magazine. Machine learning can also help improve education systems themselves.

“Greater use of data can show where problems occur in certain subjects in certain districts, allowing for early intervention and efficient management of a schooling system”, wrote Vivek Puthucode, General Manager for Public Sector in Microsoft Asia Pacific.

The company has built an automated translator for its Skype messaging service. Once honed, these systems could allow teachers in the developed world to beam into classrooms in rural nations, sharing their expertise and boosting global teaching standards.

4. Customer Service

There is plenty of hype around chatbots, but one system has an edge - it uses facial tracking to understand body language and monitor a citizens’ emotions.

“There are a lot of chatbots out there that use social talk, but they use it randomly, and that’s not useful”, Justine Cassell, director at Carnegie Mellon University told MIT Technology Review.

Cassell has developed Sara, a chatbot that has helped out in this year’s World Economic Forum to help attendees network with other people of interest.

Sara takes note of a person’s tone and choice of words, studies a person’s facial expression and body language. Data is then fed into a programme that churns out a suitable response to build rapport.

This technology could help chatbots be more sympathetic as they assist people, perhaps soothing a frantic citizen who is stressed about their tax deadline. Will humans have the edge when it comes to empathy?

5. Defence

Singapore’s Armed Forces (SAF) is looking into artificially intelligent unmanned systems to fight alongside its soldiers, it was reported this weekend, pre-empting its manpower shortage.

“The SAF is expected to face a reduction in the number of national servicemen, given Singapore’s low birth rate”, said Ng Chee Khern, Permanent Secretary of Defence Development, Ministry of Defence.

The government is also looking to improve command, control and communication systems with sensors and cameras, beefing up agency coordination in counter-terrorism operations.

The United States Department of Defense is doing the same, spending billions of dollars to develop autonomous weapons. The team is exploring intelligent fighter jets that can identify enemies on the ground, and has tested smart missiles and autonomous ships that can hunt for enemy submarines.

AI technology is far from perfect, but the fact that it’s already being trialled shows its potential impact on government. If you’re interested to find out more, watch this AI pioneer from Google’s Deep Mind discuss how quickly AI learns, and its potential applications in the world.

Mustafa Suleyman, Co-founder of Google DeepMind at FutureFest 2016 from Nesta UK on Vimeo. Changi airport by Nan-Cheng Tsai, licensed under CC BY-ND 2.0 Japan street by Dick Thomas Johnson, licensed under CC BY 2.0 Students learning by Thomas Galvez, licensed under CC BY 2.0 Missiles by Naval Surface Warriors, licensed under CC BY-SA 2.0