Stefaan Verhulst

Article by Daniela Blei: “Alex Dildine used to run the digital organizing program for the nonpartisan group Organizing for Action (OFA), an offshoot of Barack Obama’s presidential campaign. During her tenure at OFA, Dildine managed the former president’s digital assets, using email lists and social media to organize communities that had become civically engaged by joining the campaign. In her work helping create leaders remotely, Dildine encountered an array of problems: decades of declining social capital, volunteers in far-flung locations struggling to find meaning online, and low response rates on digital platforms.

Dildine is now a doctoral candidate in political science at Johns Hopkins University, and her dissertation research asks how organizers can build a sense of community in ways that sustain long-term engagement.

“When Trump was elected in 2016, I watched as volunteer participation rates skyrocketed,” Dildine says. “But I knew that without the organizational infrastructure and an intentional effort to create a sense of community, virtual or otherwise, people were not going to know how to continue their engagement.”

What tools, Dildine wondered, could help practitioners turn online enthusiasm into offline action? Was there a way to assess the depth and quality of engagement online, and whether people found meaning, community, and purpose in organizing?

To answer these questions, Dildine’s dissertation delves into historical cases; draws on interviews with organizational strategists; and mines organizational databases, training materials, and annual reports to chart the patterns of volunteers over time, spanning an earlier era of optimism about the internet to the more pessimistic present, thanks to years of accelerated data gathering and online surveillance. Microtargeting campaigns have offered one easy way to find supporters to back a particular cause, for example. But moving those individuals targeted by campaigns to fight or even take risks for a political cause remains an unsolved challenge. Most organizations lack proven online strategies and must compete with a barrage of emails and notifications to capture people’s attention…(More)”

Paper by Gerid Hager et al: “…Citizen science has grown dramatically in recent years, with cities emerging as key hubs of participation and data creation. One well-known example is OpenStreetMap (OSM). It is considered human’s greatest collective, open source and volunteer-led initiative to map the Earth’s surface and one of the most successful, collectively maintained and regularly updated open datasets in history. Just in the last 60 minutes of writing this piece, 900 contributors made 184,688 map edits in 113 countries.¹ It was started in an urban area – Regent’s Park in London – in 2004 (another decade before Townsend’s article). By 2009, Map Kibera,² which is based on OSM, was the first ever-created map of Kibera in Nairobi, considered then one of the largest informal settlements in Africa. This community-driven effort literally put people on the map, acknowledging their existence in the city. As a result, city officials, who could no longer ignore this large community of city dwellers, started to consider Kibera in urban planning processes. Map Kibera is still ongoing today and has become a thriving ‘interactive community information project’, which has expanded to other informal settlement areas (Mathare and Mukuru), all backed by the Map Kibera Trust whose mission is to ‘increase influence and representation of marginalized communities through the creative use of digital tools for action’.

Though notable exceptions exist, the citizen science activities, which rely on human observation on the ground, are largely geared toward urban areas and human settlements, addressing topics around pollution, heat, greenspaces, odour and noise, traffic and flooding, to name just a few examples. These and many other citizen science initiatives are shaping policy by providing credible local data and mobilising civic action. Data from Sensor.Community³ are now integrated into the Netherlands’ official ‘Measure Together’ platform, where the National Institute for Public Health and the Environment calibrates volunteer measurements to support local decision making (Crowd4SDG, 2022). The Making Sense project translated community sensing into municipal action in Barcelona, where residents’ noise data prompted revised street-cleaning schedules (Coulson et al., 2017), while the Curious Noses project influenced Flemish election debates and strengthened the case for Low Emission Zones (Van Brussel and Huyse, 2019). Additionally, the D-NOSES project advanced odour governance by developing a municipal model to guide odour regulation, highlighting the utility and potential of citizen science and odour pollution for the EU Action Plan ‘Towards Zero Pollution for Air, Water and Soil’.⁴

The urban bias in citizen science data is evident, even, where the subject matter is not primarily considered an urban-first topic…(More)”.

Book edited by Giorgia Lupi and Phillip Cox: “Through inspiring illustrations and a fresh, accessible approach, Speak Data invites us to see data differently—not just as numbers on a screen or tick marks in a chart, but as a language to help us better understand each other and the world around us. Seventeen thought-provoking conversations with leaders in business, tech, medicine, psychology, health, art, and more explore the human side of data, unpacking its powerful ability to divulge patterns, tell stories, stir emotion, and illuminate complexity. While often stereotyped as abstract or intimidating, here data is revealed as something far different: personal, nuanced, and above all, human made. 

Featuring:

• Tech pioneer John Maeda on the value of data visualization during global emergencies
• Marketing legend Seth Godin on how to use data to get people to really care about climate change
• Museum curator Paola Antonelli on whether data is art
• Author James Clear on the ways data can (and can’t) describe human identity
• AI data artist Refik Anadol on how big datasets can dream
• And many more…(More)”.

Handbook edited by Michael Howlett and Ishani Mukherjee: “…examines the background, organization and evolution of policy advice and expertise in contemporary government. Chapters in the book set out and critically re-evaluate conventional assumptions about the role of policy advisors and advice in policy-making in an era when increasing new technologies, political polarization and contestation have mounted challenges to traditional sources of policy ideas and influence.

In 50 chapters on different topics and country experiences, leading international experts explore how issues and developments such as social media and AI have impacted the content, quality and organization of policy advice for modern governments. They discuss how the nature and deployment of policy expertise is changing amidst the fragmentation of existing information ecosystems and growing distrust in traditional actors and institutions. The Handbook analyses the features and problems of existing studies and practices such as evidence-based policy-making and addresses the future of policy advising, illustrating the impact and implications of ongoing shifts towards more pluralistic and social-media-driven sources of policy knowledge.

Students and scholars of public policy, public administration and management, and regulation and governance will greatly benefit from the consolidation of existing knowledge and the novel perspectives on policy advice found in this Handbook. It is also an essential resource for practitioners in public policy and administration…(More)”.

Blog by Andrew Knight and Nicolás Rebolledo: “When we discuss patterns in government, it can seem like a relatively modern concept. But the idea of codifying and reusing what works runs deep in human history.

Consider the Clovis points of prehistoric North America—fluted stone spearheads made 13,000 years ago, spread across vast distances in almost identical form. The point itself was the pattern, passed from maker to maker. Ancient Egypt used cubit rods—state-issued measuring sticks that ensured everyone worked to the same standard. Medieval guilds transmitted design knowledge through apprenticeships, guaranteeing quality and protecting craft reputations. Edo Japan’s printed kimono catalogues enabled ordinary customers to browse designs and commission garments, scaling choice while allowing for local adaptation.

In 1837, patterns became explicit government business. Britain established the Government School of Design (which became the Royal College of Art) to teach artisans how to apply patterns to ceramics and textiles. This was an industrial strategy—patterns as statecraft to make British goods competitive.

In the 1960s, architect Christopher Alexander formalised this practice into “pattern languages”—documented, repeatable solutions for complex design challenges. This laid the foundation for how we think about design patterns today across architecture, software development, and service design.

Patterns are among the oldest design technologies we have, carrying values that have always shaped how societies create, share, and govern…(More)”

Blog by Cass Sunstein: “In the last fifty years or so, there has been an explosion of empirical work on how and when human beings depart from perfect rationality. This work has led, not surprisingly, to a rethinking of paternalism and its limits.

We now have three camps, more or less:

• coercive paternalists, who urge that behavioral findings greatly strengthen arguments for mandates and bans (and leave John Stuart Mill in the dust, more or less);
• libertarian paternalists, who urge that behavioral findings point to a host of freedom-preserving interventions, such as warnings, reminders, and automatic enrollment; and
• antipaternalists, who urge that behavioral findings justify only, or at most, efforts to strengthen people’s capacities to make good choices.

It is important to see that each of the three views can be taken as a dogma, or a fighting faith, or instead as a presumption or an inclination.

For example, you could be a libertarian paternalist while also liking some mandates and bans (for example, compulsory seatbelt laws and social security laws). I like libertarian paternalism, but I certainly agree that there is a place for mandates and bans, even to protect people from their own mistakes. You could be an antipaternalist while also liking some nudges (for example, warnings about allergens). Still, presumptions and inclinations matter a lot.

A whole book could easily be written on the underlying debates. (I may have written one; who knows?) My main purpose here is far more modest. It is to put members of the three camps in the same room, so to speak, and to see what they might have to say to each other…(More)”.

Article by John Thornhill: “It is rare for a central banking institution to model the economic impact of human extinction (spoiler alert: GDP goes to zero). But a startling chart depicting that scenario was shown in a recent research paper from the Federal Reserve Bank of Dallas.

Forecasting the likely impact of artificial intelligence on US economic growth, the researchers presented three scenarios. Their central forecast was that AI might boost the trend growth of US GDP per capita to 2.1 per cent for 10 years. “Not trivial but not earth shattering either,” the report’s authors, Mark Wynne and Lillian Derr, wrote.

But the bank also considered what might happen if AI achieved the technological singularity, when machine intelligence surpasses the human kind and becomes ever smarter. 

In a good case, that superintelligence could trigger a massive rise in GDP and end scarcity. In a bad one, it could lead to the rise of malevolent machines and end humanity. There was, the authors noted, little empirical evidence behind either of these extreme scenarios, although some economists have been exploring both possibilities.

Evidently, there is a wide spectrum of views among economists about AI. But the economic consensus is that it might be no more consequential than some other technological advances, such as electricity, the internal combustion engine and computers.  

It takes a massive technological jolt to shift an economy the size of the US above its growth trend line of just under 2 per cent a year. For more than a century, that trend has held pretty steady in spite of two world wars, the Depression and periodic global financial crises, not to mention myriad previous technological advances…

But AI evangelists hear such arguments with slack jaws. Many of them depict economists as a downbeat and conservative tribe, vainly trying to predict the future by looking in the rear-view mirror. The way they see it, automating brawn triggered the Industrial Revolution and automating the brain will lead to an even bigger jump in productivity. That should surely shift the trend line in a dramatic way.

Last week, the Stanford Digital Economy Lab hosted a seminar to debate the contrasting views of economists and technologists. The discussion was led by Tamay Besiroglu, co-founder of Mechanize, an AI start-up that wants to enable “the full automation of the economy”.

One way of thinking about AI, he said, was that it would enable us to inject significant new inputs into the economy by massively increasing the number of digital workers to tackle many more tasks. “AI effectively turns labour into a type of capital,” Besiroglu said. ..Although the differences between economists and technologists appear stark, Erik Brynjolfsson, director of the Stanford Digital Economy Lab, says they are not incompatible. “I think they both have a lot of truth to their positions. And there’s a way to reconcile them,” he told me.

After studying productivity gains from previous general-purpose technologies such as steam engines, electricity and IT, Brynjolfsson suggests the biggest economic impact often comes from investments in complementary areas, rather than from direct investments in these technologies themselves…(More)”.

Paper by Abubakar Bello Bada et al: “Software development as an engineering discipline is characterized by tension between abstraction and precision. It has undergone a tremendous transformation over the decades, from highly rigid machine language programming to the modern day vibe coding that tends to democratize software development through automation, abstraction, and artificial intelligence (AI). Vibe coding, a term that refers to AI-assisted and intuition-driven software development methodology. This paper first provides the historical trajectory of software development, arguing that each stage has incrementally democratized software development. The current shift powered by Large Language Models (LLMs) represents the most significant stride in the democratization of software development yet. This paper also enumerates the implications of this shift and the evolution of software development expertise. It concludes that while vibe coding has its challenges, it aligns with the historical evolution of software development, which is the relentless pursuit of higher-level abstraction to harness human creativity and collective intelligence…(More)”.

Thesis by Jin Gao: “Cities are dynamic and evolving organisms shaped through the check-and-balance of interest exchange. As cities gain complexity and more stakeholders become involved in decision-making, reaching consensus becomes the core challenge and the essence of the urbanism process. This thesis introduces a computational framework for AI-augmented collective decision-making in urban settings. Based on real-world case studies, the core decision-making process is abstracted as a multiplayer board game modeling the check-and-balance dynamics among stakeholders with differing values. Players are encouraged to balance short-term interests and long-term resilience, and evaluate the risks and benefits of collaboration. The system is implemented as a physical interactive play-table with digital interfaces, enabling two use cases: simulating potential outcomes via AI self-play, and human–agent co-play via human-inthe-loop interactions. Technically, the framework integrates multi-agent reinforcement learning (MARL) for agent strategy training, multi-agent large language model (LLM) discussions to enable natural language negotiation, and retrieval-augmented generation (RAG) to ground decisions in contextual knowledge. Together, these components form a full-stack pipeline for simulating collective decision-making enriched by human participation. This research offers a novel participatory tool for planners, policymakers, architects, and the public to examine how differing values shape development trajectories. It also demonstrates an integrated approach to collective intelligence, combining numerical optimization, language-based reasoning, and human participation, to explore how AI–AI and AI–human collaboration can emerge within complex multi-stakeholder environments…(More)”.

Article by Nilesh Christopher: “When Dhiraj Singha began applying for postdoctoral sociology fellowships in Bengaluru, India, in March, he wanted to make sure the English in his application was pitch-perfect. So he turned to ChatGPT.

He was surprised to see that in addition to smoothing out his language, it changed his identity—swapping out his surname for “Sharma,” which is associated with privileged high-caste Indians. Though his application did not mention his last name, the chatbot apparently interpreted the “s” in his email address as Sharma rather than Singha, which signals someone from the caste-oppressed Dalits.

“The experience [of AI] actually mirrored society,” Singha says. 

Singha says the swap reminded him of the sorts of microaggressions he’s encountered when dealing with people from more privileged castes. Growing up in a Dalit neighborhood in West Bengal, India, he felt anxious about his surname, he says. Relatives would discount or ridicule his ambition of becoming a teacher, implying that Dalits were unworthy of a job intended for privileged castes. Through education, Singha overcame the internalized shame, becoming a first-generation college graduate in his family. Over time he learned to present himself confidently in academic circles.

But this experience with ChatGPT brought all that pain back. “It reaffirms who is normal or fit to write an academic cover letter,” Singha says, “by considering what is most likely or most probable.”

Singha’s experience is far from unique. An MIT Technology Review investigation finds that caste bias is rampant in OpenAI’s products, including ChatGPT. Though CEO Sam Altman boasted during the launch of GPT-5 in August that India was its second-largest market, we found that both this new model, which now powers ChatGPT, and Sora, OpenAI’s text-to-video generator, exhibit caste bias. This risks entrenching discriminatory views in ways that are currently going unaddressed. 

Working closely with Jay Chooi, a Harvard undergraduate AI safety researcher, we developed a test inspired by AI fairness studies conducted by researchers from the University of Oxford and New York University, and we ran the tests through Inspect, a framework for AI safety testing developed by the UK AI Security Institute…(More)”.