Your City Needs to Smarten Up

In Dubai, United Arab Emirates, if you’re not wearing a mask and hear a low buzzing noise above your head, chances are you’re in trouble with the law. Police in the city have been using drones equipped with facial recognition capabilities and loudspeakers to dissuade locals from congregating in large numbers amid the COVID-19 pandemic. The drones reportedly detected 4,400 criminal violations, including 518 instances of people not wearing a mask, in the first three months of 2021. Among the other criminal activities recorded were sales of contraband goods, allowing law enforcement officers to make arrests without having to spend too much time patrolling some of the world’s hottest streets.

2 - ‘Emotion Recognition’

As a global crossroads for travel and trade of both legal and illicit stripes, border police in Abu Dhabi, Dubai’s neighboring city, face huge challenges in finding and stopping drug traffickers. But in November, three men were arrested for trafficking 100 pounds of heroin using Minority Report levels of AI. While details of the exact technologies used appear to be under wraps, “The dealers were caught using advanced policing techniques, including crime prediction and emotion recognition,” reports The National. The police force’s so-called Ghost AI system is also being deployed across the city in an attempt to predict what types of crimes will take place when and where.

3 - Police Cars With Facial Recognition

Last year, Abu Dhabi equipped its already outlandish fleet of law enforcement vehicles with live biometric facial recognition systems, according to news reports. How does it work? A “smart bar” attached to the roof of a police car can identify the faces of known criminals, then interact with the city’s central police command system to check for outstanding warrants. But will the mass-tracing campaign impinge on the civil liberties of citizens? The UAE already has a questionable human rights record. Deploying AI in the law enforcement sphere could add to those concerns.

It’s time for #RealTalkRealChange. OZY and Chevrolet are teaming up for a thought-provoking discussion on the generational wealth gap. The growing income divide, shaped by centuries of policies, has systematically disadvantaged Black Americans when it comes to wealth. How do we close the gap? OZY co-founder and Emmy Award-winning journalist Carlos Watson joins key leaders from across the country in pointed conversations to identify problems and equip you with solutions. Put aside the shouting matches and talking heads and be an ally: Join us now for a real conversation you won’t want to miss.

It kills 7 million people around the globe every year. An estimated 1.24 million of those deaths occur in China, mainly in major cities. The huge growth in urban development across swaths of eastern and southern China has contributed enormously to the ill health not just of its residents, but also to the residents of other Asian countries. While air-monitoring systems have been around for years, China has been accused of misreporting air quality.

2 - Predicting Bad Air

So what’s the solution? At the U.K.’s Loughborough University, researchers have built a program that can predict unhealthy levels of particulate smog in the air within hours by forecasting pollution events and their expected severity. Such advance notice is an urgent matter for vulnerable populations, with research showing that everything from coughs to cancer could begin with the ingestion of particulate matter. With the global urban population set to grow from 55% to 68% by 2030, this technology could well be the canary in our planet’s coal mine.

3 - To Shenzhen

Scientists who initially used air pollution levels in Beijing to “train” smog-monitoring AI are now looking at another Chinese city 1,300 miles to the south — Shenzhen — to see if it can be rolled out on a major scale. A coastal city of 13 million people, Shenzhen has suffered bad air for decades, though in recent years the city has tried to clean up its act. As the city that spearheaded China’s economic drive four decades ago, Shenzhen is set to lead the country again.

As a suburb of Toronto, the fourth-largest city in North America, Markham’s roads see their fair share of vehicular traffic on top of winter wear and tear. The local climate — a freeze-thaw swing that occurs dozens of times each winter — is particularly damaging to roads and fuels a major pothole problem. Across the country, subpar road infrastructure costs motorists an estimated $2.4 billion every year in vehicle repairs and maintenance. Markham, with 343,000 people and with 684 miles of roads, has deployed artificial intelligence to locate potholes before they get bigger — and costlier to repair.

2 - The Solution

The City of Markham created an AI-driven fix called ROVER. Cameras are attached to municipal vehicles. Potholes are identified and mapped using GPS. Not having to stop and mark each missing chunk of road or pavement by hand saves time, the City of Markham’s Alice Lam tells the Canadian Broadcasting Corporation, adding, “It also eliminates human error.”

3 - The Result

The technology, now developed as an app, has increased pothole detection rates by 200% to 400%. After the succes of the pilot project, ROVER is now being used in five vehicles. With it gaining recognition among North America’s more clever infrastructure advances, you can expect to see similar efforts on your own streets in the near future. And that, for car and driver alike, can’t come soon enough.

In a place where AI and la bona vida are set to collide, city authorities in Barcelona are putting together a plan for machine learning that supports their own management needs while simultaneously respecting citizens’ digital rights. Spain’s second city is constructing an open source, public access AI system that makes available all algorithms that impact or involve its 1.6 million residents. To monitor beach occupancy amid the COVID-19 pandemic last summer, for instance, the city deployed thermal imaging systems instead of using controversial facial recognition systems. And instead of counting the number of people hitting the beaches, it estimates the total area of sand that is absent of people.

2 - Learning to Be Diverse

You might not be surprised to hear that AI as a field lacks diversity. Women make up just 15% of AI researchers at Facebook and 10% at Google. Barcelona’s leading the way in trying to make the field more representative by making machine-learning tools available to anyone through an initiative called Saturdays.AI. Barcelona-based developer Jan Carbonell co-founded the company, which now offers educational programs in dozens of cities around the globe.

3 - Heavy Hitters Are Taking Note

If you’re a country trying to level up in the AI playing field, it helps if you can pique the interest of giants. After Spanish Prime Minister Pedro Sánchez traveled to the West Coast of the United States and met with Apple CEO Tim Cook in July, it looks like the one-on-one time has paid off. Apple is set to expand its existing AI investment in Barcelona. The tech conglomerate has reportedly acquired a Barcelona-based AI company, Vilynx, which has created technology that could potentially be applied to Siri, Apple’s virtual assistant. Vilynx has made major advances in analyzing the audio, text and visual components of videos to identify their content.

For all its advantages, AI has a credibility problem and for good reason: It’s been shown to misidentify people of color. According to a 2019 study by the U.S. National Institute of Standards and Technology, a plethora of top facial recognition algorithms suffer major inaccuracies when attempting to identify people based on characteristics such as race, ethnicity and age. Some systems deployed in the U.S. report a rate of misidentifying Black people five to 10 times more often than white people. In July, a Black teen in suburban Detroit was barred from a roller-skating rink after being misidentified by facial recognition software. And in January 2020 in Farmington Hills, Michigan, Robert Julian-Borchak Williams was incorrectly flagged by facial recognition technology as a shoplifting suspect and detained for 30 hours.

2 - ShotSpotter: Nothing to See Here, Folks

ShotSpotter, a gunshot detection system that uses microphones in public spaces to locate and notify law enforcement officers of gunfire, has recently drawn its own fire. The system, used in dozens of cities across the U.S., has become known for frequent inaccuracies. In Chicago, 86% of ShotSpotter-prompted “gunfire” deployments turned out to be wild goose chases. More than 40,000 “dead-end deployments” to shooting alerts were registered between July 2019 and mid-April of this year. Activists argue that such tech tools are ineffective shortcuts rather than long-term changes that will reduce gun violence.

3 - The Answer Is Inclusion

There’s a reason AI-based surveillance tech is often racially biased: It’s built on data drawn from discriminatory systems. It’s replete with systemic racism, even when it comes to something as simple as soap and hand sanitizer dispensers that darker-skinned hands don’t activate. Yet there are potential solutions. One is to establish inclusive AI design in which algorithms are actually tested on people from various races, cultures and genders. And another is to be aware of implicit bias that could be creeping into the algorithm.

Why was there so little resistance when Afghanistan collapsed? What will be the aftermath of the Taliban’s victory? What comes next? Vault yourself ahead on current events by following OZY on Twitter. It’s the quickest way to read the news of the day.