Yet this assumes that you can get hold of that training data, says Kautz. He and his colleagues at Nvidia have come up with a different way to expose private data, including images of faces and other objects, medical data, and more, that does not require access to training data at all.
Instead, they developed an algorithm that can re-create the data that a trained model has been exposed to by reversing the steps that the model goes through when processing that data. Take a trained image-recognition network: to identify what’s in an image, the network passes it through a series of layers of artificial neurons. Each layer extracts different levels of information, from edges to shapes to more recognizable features.
Kautz’s team found that they could interrupt a model in the middle of these steps and reverse its direction, re-creating the input image from the internal data of the model. They tested the technique on a variety of common image-recognition models and GANs. In one test, they showed that they could accurately re-create images from ImageNet, one of the best known image recognition data sets.
As in Webster’s work, the re-created images closely resemble the real ones. “We were surprised by the final quality,” says Kautz.
The researchers argue that this kind of attack is not simply hypothetical. Smartphones and other small devices are starting to use more AI. Because of battery and memory constraints, models are sometimes only half-processed on the device itself and sent to the cloud for the final computing crunch, an approach known as split computing. Most researchers assume that split computing won’t reveal any private data from a person’s phone because only the model is shared, says Kautz. But his attack shows that this isn’t the case.
Kautz and his colleagues are now working to come up with ways to prevent models from leaking private data. We wanted to understand the risks so we can minimize vulnerabilities, he says.
Even though they use very different techniques, he thinks that his work and Webster’s complement each other well. Webster’s team showed that private data could be found in the output of a model; Kautz’s team showed that private data could be revealed by going in reverse, re-creating the input. “Exploring both directions is important to come up with a better understanding of how to prevent attacks,” says Kautz.
A pro-China online influence campaign is targeting the rare-earths industry
China has come to dominate the market in recent years, and by 2017 the country produced over 80% of the world’s supply. Beijing achieved this by pouring resources into the study and mining of rare-earth elements for decades, building up six big state-owned firms and relaxing environmental regulations to enable low-cost and high-pollution methods. The country then rapidly increased rare-earth exports in the 1990s, a sudden rush that bankrupted international rivals. Further development of rare-earth industries is a strategic goal under Beijing’s Made in China 2025 strategy.
The country has demonstrated its dominance several times, most notably by stopping all shipments of the resources to Japan in 2010 during a maritime dispute. State media have warned that China could do the same to the United States.
The US and other Western nations have seen this monopoly as a critical weakness for their side. As a result, they have spent billions in recent years to get better at finding, mining, and processing the minerals.
In early June 2022, the Canadian mining company Appia announced it had found new resources in Saskatchewan. Within weeks, the American firm USA Rare Earth announced a new processing facility in Oklahoma.
Dragonbridge engaged in similar activity in 2021, soon after the American military signed an agreement with the Australian mining firm Lynas, the largest rare-earths company outside China, to build a processing plant in Texas.
The U.S. only has 60,000 charging stations for EVs. Here’s where they all are.
The infrastructure bill that passed in November 2021 earmarked $7.5 billion for President Biden’s goal of having 500,000 chargers (individual plugs, not stations) around the nation. In the best case, Michalek envisions a public-private collaboration to build a robust national charging network. The Biden administration has pledged to install plugs throughout rural areas, while companies constructing charging stations across America will have a strong incentive to fill in the country’s biggest cities and most popular thoroughfares. After all, companies like Electrify America, EVgo, and ChargePoint charge customers per kilowatt-hour of energy they use, much like utilities.
Most new electric vehicles promise at least 250 miles on a full charge, and that number should keep ticking up. The farther cars can go without charging, the fewer anxious drivers will be stuck in lines waiting for a charging space to open. But make no mistake, Michalek says: an electric-car country needs a plethora of plugs, and soon.
We need smarter cities, not “smart cities”
The term “smart cities” originated as a marketing strategy for large IT vendors. It has now become synonymous with urban uses of technology, particularly advanced and emerging technologies. But cities are more than 5G, big data, driverless vehicles, and AI. They are crucial drivers of opportunity, prosperity, and progress. They support those displaced by war and crisis and generate 80% of global GDP. More than 68% of the world’s population will live in cities by 2050—2.5 billion more people than do now. And with over 90% of urban areas located on coasts, cities are on the front lines of climate change.
A focus on building “smart cities” risks turning cities into technology projects. We talk about “users” rather than people. Monthly and “daily active” numbers instead of residents. Stakeholders and subscribers instead of citizens. This also risks a transactional—and limiting—approach to city improvement, focusing on immediate returns on investment or achievements that can be distilled into KPIs.
Truly smart cities recognize the ambiguity of lives and livelihoods, and they are driven by outcomes beyond the implementation of “solutions.” They are defined by their residents’ talents, relationships, and sense of ownership—not by the technology that is deployed there.
This more expansive concept of what a smart city is encompasses a wide range of urban innovations. Singapore, which is exploring high-tech approaches such as drone deliveries and virtual-reality modeling, is one type of smart city. Curitiba, Brazil—a pioneer of the bus rapid transit system—is another. Harare, the capital of Zimbabwe, with its passively cooled shopping center designed in 1996, is a smart city, as are the “sponge cities” across China that use nature-based solutions to manage rainfall and floodwater.
Where technology can play a role, it must be applied thoughtfully and holistically—taking into account the needs, realities, and aspirations of city residents. Guatemala City, in collaboration with our country office team at the UN Development Programme, is using this approach to improve how city infrastructure—including parks and lighting—is managed. The city is standardizing materials and designs to reduce costs and labor, and streamlining approval and allocation processes to increase the speed and quality of repairs and maintenance. Everything is driven by the needs of its citizens. Elsewhere in Latin America, cities are going beyond quantitative variables to take into account well-being and other nuanced outcomes.
In her 1961 book The Death and Life of Great American Cities, Jane Jacobs, the pioneering American urbanist, discussed the importance of sidewalks. In the context of the city, they are conduits for adventure, social interaction, and unexpected encounters—what Jacobs termed the “sidewalk ballet.” Just as literal sidewalks are crucial to the urban experience, so is the larger idea of connection between elements.
Truly smart cities recognize the ambiguity of lives and livelihoods, and they are driven by outcomes beyond the implementation of “solutions.”
However, too often we see “smart cities” focus on discrete deployments of technology rather than this connective tissue. We end up with cities defined by “use cases” or “platforms.” Practically speaking, the vision of a tech-centric city is conceptually, financially, and logistically out of reach for many places. This can lead officials and innovators to dismiss the city’s real and substantial potential to reduce poverty while enhancing inclusion and sustainability.
In our work at the UN Development Programme, we focus on the interplay between different components of a truly smart city—the community, the local government, and the private sector. We also explore the different assets made available by this broader definition: high-tech innovations, yes, but also low-cost, low-tech innovations and nature-based solutions. Big data, but also the qualitative, richer detail behind the data points. The connections and “sidewalks”—not just the use cases or pilot programs. We see our work as an attempt to start redefining smart cities and increasing the size, scope, and usefulness of our urban development tool kit.
We continue to explore how digital technology might enhance cities—for example, we are collaborating with major e-commerce platforms across Africa that are transforming urban service delivery. But we are also shaping this broader tool kit to tackle the urban impacts of climate change, biodiversity loss, and pollution.
The UrbanShift initiative, led by the UN Environment Programme in partnership with UNDP and many others, is working with cities to promote nature-based solutions, low-carbon public transport, low-emission zones, integrated waste management, and more. This approach focuses not just on implementation, but also on policies and guiderails. The UNDP Smart Urban Innovations Handbook aims to help policymakers and urban innovators explore how they might embed “smartness” in any city.
Our work at the United Nations is driven by the Sustainable Development Goals: 17 essential, ambitious, and urgent global targets that aim to shape a better world by 2030. Truly smart cities would play a role in meeting all 17 SDGs, from tackling poverty and inequality to protecting and improving biodiversity.
Coordinating and implementing the complex efforts required to reach these goals is far more difficult than deploying the latest app or installing another piece of smart street furniture. But we must move beyond the sales pitches and explore how our cities can be true platforms—not just technological ones—for inclusive and sustainable development. The well-being of the billions who call the world’s cities home depends on it.
Riad Meddeb is interim director of the UNDP Global Centre for Technology, Innovation, and Sustainable Development. Calum Handforth is an advisor for digitalization, digital health, and smart cities at the UNDP Global Centre.