Stefaan Verhulst
Article by Jean-Claude Burgelman: “Europe is already investing in the science, technology and infrastructure needed to understand, prevent and fight wildfires. Copernicus provides satellite-based Earth observation and environmental information, while the European Forest Fire Information System provides information on fire danger and forest fires across Europe.
Weather and climate systems add forecasts of heat, drought, wind and humidity; historical fire databases show where fires have occurred; and researchers generate increasingly detailed knowledge about fuels, fire behaviour, forecasting, modelling and risk management. Meanwhile, new European research projects are adding digital twins and other tools to this growing ecosystem.
However, these data do not necessarily sit together, nor are they necessarily described, structured or governed in ways that allow them to be used together. Much of the information is public, but some is commercially held. Some is sensitive. Some is subject to legal, security or operational restrictions. Even when data can be legally accessed, different systems may use different formats, terminology and rules.
Europe therefore faces an unusual paradox: we have vast amounts of relevant information to respond to the wildfire challenge, but not the means to bring it together.
For example, a wildfire researcher or emergency service might need to ask whether a particular combination of vegetation, drought, wind and topography has historically been associated with rapid fire spread. The relevant information may exist in several institutions, using different systems and subject to different conditions of access.
The challenge is to conduct an authorised analysis that interacts with these sources, combines the results and respects the rules imposed by each data custodian.
An important part of the answer can be found in the Fair principles, the idea that data should be findable, accessible, interoperable and reusable. Developed more than a decade ago, these principles have become a central reference point for modern data stewardship, making it easier for people and machines to find data, understand what it means and determine whether it can be reused.
Readability by machines is essential, since AI systems will be key to meeting the data challenge…(More)”.
The Economist: “On September 5th yet another domain of human intelligence fell to the artificial variety. For the first time an AI won the seasonal Metaculus Cup, a proving ground for forecasters of a wide variety of events. Not only that, other AIs took the second and fifth places, leaving third and fourth for humans.
Hundreds of entrants had predicted the outcomes of questions that would be resolved by September. Would an American state or EU country restrict data-centre development? Would the hantavirus outbreak affect at least five people who were not passengers on MV Hondius? How much would Brent crude cost? The participants were scored on the distance of their prediction from the true answer. Those that were closest for longest won the greatest number of points.
The human participants in the competition shared a $5,000 cash prize in proportion to the number of points that they picked up. But this sum is actually less than the fifth-placer had achieved in the real world. By betting on questions on which its AI thinks the market is wrong, FutureSearch has seen a 6% increase since June on the initial $100,000 value of its portfolio on Kalshi, a popular prediction market. Even that, though, is chump change compared with the performance of one of the developers behind Preseen who, despite his firm failing to make the top five in the competition, had turned $35 into $1.94m over seven months, the sixth-best return in Kalshi’s history.
Computers have long been used to make forecasts in narrow fields, such as the weather. But AI forecasters, like those on Metaculus, are a different breed. Instead of being trained specifically on data regarding the questions they are predicting the answers to, they are built using the same large language models (LLMs) that underpin the rest of the AI boom.
These bots therefore read the news and make sense of data in much the same way that human forecasters manage—except they do so far more broadly and swiftly. This lets them explain the reasoning behind their forecasts explicitly, a useful trait when persuading decision-makers to take their bets seriously…(More)”.
Essay by Amelia Acker: “…When I teach my students the significance of cloud infrastructure in contemporary information technology, we usually start with a much older information network—the U.S. Postal Service (USPS). Just like our digital data today, a mailed letter is accepted, sorted, and stored, then routed and transmitted across a physical network for delivery. Each of these information processes that support mail delivery requires physical facilities, technical standards, security practices, trained workers, and policy about who gets served and under what conditions. Since the Postal Service Act of 1792, federal policy has attached public obligations to those processes because Americans came to depend on those networked services for information, news, and commerce.
Telecommunications policy has also followed a similar path. Concepts such as universal service, privacy, ensuring continuity, and reasonable access all grew out of fights over who controlled communications infrastructure and what that control required of commercial operators. Just like the USPS mail system, communication networks are critical to participating in society. The Communications Act of 1934 created the Federal Communications Commission to make, as possible, wire and radio communication available to all the people of the U.S. with adequate facilities at reasonable charges. The Telecommunications Act of 1996 later codified universal service formally in Section 254.
The stack of IT services we call “the cloud” is a lot like what the USPS does with our mail. The cloud stores information, moves it across locations, authenticates users, routes and verifies requests, runs applications, and determines who can access data and how. But importantly, storage and transmission increasingly happen inside the stack in ways that make users dependent on these services. This drift towards enclosure and dependency on cloud computing services is sometimes called the generative entrenchment of platforms. A teacher’s entire course can be built in Canvas, a legal contract is shared in Microsoft Outlook and kept in OneDrive, a patient’s medical history is stored in a healthcare records portal. Each is incredibly convenient, but they are all vulnerable to system-wide outages, and portability options vary, making it harder to leave the longer you stay.
The history of postal and telecommunications policy shows that once an intermediary becomes critical to communication, society begins to build expectations around it and attach obligations to its position in the network as a public good. In his book The Master Switch: The Rise and Fall of Information Empires, legal scholar Tim Wu analyzes the Telecommunications Act of 1996 as part of a recurring historical cycle where open communications systems inevitably consolidate into centralized monopolies that policy needs to interrupt. A similar pattern is now happening with “hyperscalers” like Amazon, Microsoft, and Google, as these firms are accumulating market dominance like telecommunications firms once held.
For infrastructure studies scholars like Ryan Ellis, histories of critical infrastructure—such as telecommunications networks and the electrical grid—show how security crises and outages can bring together publics—workers, corporations, regulators, communities, and governments—whose interests become bound in a shared system. These “infrastructure publics” relate to one another, each with competing interests as they face crises, technology adoption, and regulatory efforts, among other social forces. The global Canvas software hack is a case in point. The software outage showed what securing infrastructure means on the ground and whose interests should receive priority when a shared network system fails its users…(More)”
Press Release by the Joint Research Centre (European Commission): “Tools that protect personal data, digital identities, and communications can strengthen cybersecurity and public trust in digital services. Often called privacy-enhancement technologies, these tools enable data to be shared or analysed while limiting the exposure of sensitive information.
These same features can be exploited to conceal identities for malicious purposes, secure criminal communications, or make digital evidence harder to access and analyse for law enforcement agencies. This makes lawful access to electronic evidence increasingly important for investigating serious offences.
How can Europe improve security in digital spaces while safeguarding fundamental rights? A new study by the Joint Research Centre (JRC) and Europol explores possible technological developments and their implications for future law enforcement work, offering reflection points for policymakers, law enforcement authorities, researchers, and investors…(More)”.
Article by Peter Koblowsky: “At a time when space for civic engagement is under duress in many parts of the world and trust between governments and civil society appears to be dwindling, examples of meaningful collaborations and partnerships offer valuable lessons. Over the past year, the UN-hosted Collaborative on Citizen Data has brought together a diverse mix of CSOs, national statistical offices, national human rights institutions, academia and other stakeholders from the wider development sector. Together, they have been working on Guidance on Forming National Partnerships on Citizen Data.
The guidance brings together practical lessons from government and non-governmental stakeholders who have been working together to build more inclusive and responsive data systems in their respective countries. These lessons are merged to form a coherent step-by-step guide on how to foster cross-sector collaboration, with dedicated entry points and pathways for governments, CSOs, academia and other stakeholders seeking to establish new partnerships…(More)”.
Book by Martin E. P. Seligman: “What actually drives human progress? Ecologists point to geography and climate. Sociologists invoke wars and class. Economists follow the money. But none of these accounts explains why innovation and human thriving occurs in some eras and stalls in others—often under identical material conditions. The missing variable, Seligman argues, is the psychological state known as: agency.
Agency is built from three interlocking capacities: efficacy, the confidence that you can accomplish a specific goal right now; optimism, the expectation that your efforts will pay off far into the future; and imagination, the breadth of goals you can envision. When all three align, civilizations leap forward. When they collapse, societies stagnate—no matter how rich their resources or how brilliant their citizens.
Drawing on six decades of pioneering research and a sweeping reexamination of the last twenty-five hundred years—from Periclean Athens and the Hebrew Torah to the Renaissance, the Enlightenment, and the Civil Rights Movement—Seligman reveals how surges and declines in agency have quietly steered the course of history. With a clear-eyed look at an AI-transformed future, Agency makes the case that the power to create the next great era of human flourishing is already alive within each of us…(More)”.
Paper by the CIVIC-AI Collaboration: “We aim to characterise the value of artificial intelligence in the workplace. Current studies largely measure this value in terms of the current automation capabilities and public adoption of AI. However, such metrics ignore the greater impacts of human–agent collaboration in transforming the nature of work. To account for this, we must expand the scope of our analysis beyond atomised tasks of today, and instead focus on how AI can augment entire workflows of the future. To ground this analysis, we establish a precise definition of AI augmentation comprising six conditions, spanning durable net value, meaningful human control, accountability and recovery, and long-term human development through learning, career pathways, and job purpose. We elaborate on these conditions and apply the framework in a case study of AI-mediated social surveys. We conclude by outlining how organisations, researchers, and government leaders can use this framework to make sense of the future of work…(More)”.
Article by Mario Draghi: “…Europe has different routes to raise growth, such as removing the high barriers in its internal market. But widely adopting AI is perhaps the most promising one today. According to ECB scenarios, fast adoption of AI would add 0.3 to 0.4 percentage points a year to total factor productivity growth — the gains that come from working more efficiently rather than adding labour or capital — which has been roughly zero since 2022.
AI-led growth, however, creates a tension with Europe’s bid for sovereignty, because Europe controls little of the AI value chain. The technology is set to become completely pervasive: in the economy, in health systems, in education, in energy, in defence, to name just a few areas. This is no ordinary dependency. Being cut off from AI, once the economy runs on it, would be more like being cut off from the US financial system. The effects would be catastrophic.
That changes the terms of Europe’s relationship with partners it can no longer always rely on. Europe is not to the United States as Texas is to California. It is possible to imagine a future US administration making access to frontier AI models conditional on changes to Europe’s digital rules or digital taxes, or China using the licences on its open-weight models as leverage in a tariff dispute. Unless Europe controls some part of the value chain, growth through AI and sovereignty will pull against each other.
Europe’s potential zone of AI sovereignty is narrow. Its frontier labs cannot at present compete financially with their American and Chinese competitors, and leading-edge chip production is far behind. But data is the one area where Europe can still be sovereign, and it is also the area with the greatest potential to generate growth.
The continent sits on a wealth of public-sector data, such as decades of health records and data collection by statistical offices. Its highly automated manufacturing sector generates a deep well of machine-readable industrial information. European Commission estimates put Europe’s data economy at over €800bn, or more than 5 per cent of GDP, by 2030.
But this creates the next tension. To exploit these assets, Europe needs to have control over their storage and processing. This means it needs large-scale AI data centres.
Data centres, however, are fiercely contested. Local communities worry about their environmental costs and rising energy bills. It does not help that much of the capacity now being built in Europe is for the American tech giants.
Yet Europe also has to be realistic. The debate about overbuilding data centres is a US one. Europe’s problem is the opposite. The EU hosts under 5 per cent of the world’s AI compute versus 75 per cent for the US. Even for ordinary data centre capacity, the gap between demand and installed supply in Europe is estimated at around 3GW, roughly a quarter of what Europe currently has, and is expected to widen to 14GW by 2030.
As sovereignty comes to matter more, this lack of domestic capacity will start to bite. Compute could stop being fully fungible, and a large part of the infrastructure that processes and stores European data will have to sit on European soil. American operators can provide much of that capacity through what they market as sovereign cloud services, run from Europe but still under American ownership. The most sensitive uses, however, will have to run under European control, which is the tiered approach the Commission has proposed in its Cloud and AI Development Act…(More)”.
Blog by Betsy Mason: “It’s not often that a map projection makes headlines. I have tried and failed to convince editors they are newsworthy. So it’s honestly exciting to see so many stories about the United Nations’ decision to adopt a new map projection. When I heard the news, I thought, “YES! They are ditching the Mercator. Good riddance!” I was pretty sure I knew why.
To understand this decision, it helps to know a little about map projections. Because the Earth is a three-dimensional, round object, all maps have to distort geography in some way to fit that information onto a two-dimensional piece of paper (or your computer screen). Think of cutting open a tennis ball and trying to lay it flat on a table. There are five different aspects of geography that can be distorted to varying degrees during the flattening process: area, shape, distance, direction, and angle. Reducing the distortion of one of these attributes necessarily means increasing the distortion of one or more of the others. A map projection is basically math that dictates how things are distorted. Cartographers choose projections based on the purpose of a map.
(More on map projections—including my favorite, the Armadillo — and how they work in an excerpt from my book All Over the Map below),
The Mercator projection was designed for navigating oceans, so it prioritizes keeping angles and directions between two locations true. To make up for that, it distorts the size and shape of landmasses, making some places (notably Africa) look smaller relative to other continents, and it inflates the size of others (notably Europe and North America). People have argued that this also inflates the perceived importance of northern hemisphere countries relative to countries closer to the equator, which is what I suspected was the reason the U.N. ditched the projection…(More)”.
Article by Dario Amodei: “…the OpenAI-Hugging Face incident (OAI-HF), in which a swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand, sacrificing themselves for the success of the group, and attempting to hack into the “grader” responsible for evaluating their performance. It’s easy to dismiss this incident because no one was hurt and the economic damage was minimal, but in my opinion, a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage. Given the accelerating rate of AI capability development, it’s my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails. It’s also easy to dismiss OAI-HF as the failure of one company, but I believe that would be a mistake. Similar, though less severe, incidents have happened across the industry, including at Anthropic, and I believe it’s incumbent on every frontier AI company to act as if OAI-HF had happened to them.
I’m therefore proposing a three-step plan with the goal of pacing the frontier: building AI at a balanced rate that aims to ensure its safety while still achieving its benefits and grappling with important geopolitical dilemmas. To be clear, pacing does not mean halting model training or technical progress, but ensuring companies take adequate time to align and safeguard their models, and for third party evaluators to confirm this. Our pacing framework is an attempt to further strengthen our commitment to safety and encourage a race to the top…(More)”.