Paper by Harini Suresh et al: “Growing interest and investment in the capabilities of foundation models has positioned such systems to impact a wide array of services, from banking to healthcare. Alongside these opportunities is the risk that these systems reify existing power imbalances and cause disproportionate harm to historically marginalized groups. The larger scale and domain-agnostic manner in which these models operate further heightens the stakes: any errors or harms are liable to reoccur across use cases. In AI & ML more broadly, participatory approaches hold promise to lend agency and decision-making power to marginalized stakeholders, leading to systems that better benefit justice through equitable and distributed governance. But existing approaches in participatory AI/ML are typically grounded in a specific application and set of relevant stakeholders, and it is not straightforward how to apply these lessons to the context of foundation models. Our paper aims to fill this gap.
First, we examine existing attempts at incorporating participation into foundation models. We highlight the tension between participation and scale, demonstrating that it is intractable for impacted communities to meaningfully shape a foundation model that is intended to be universally applicable. In response, we develop a blueprint for participatory foundation models that identifies more
local, application-oriented opportunities for meaningful participation. In addition to the “foundation” layer, our framework proposes the “subfloor” layer, in which stakeholders develop shared technical infrastructure, norms and governance for a grounded domain such as clinical care, journalism, or finance, and the “surface” (or application) layer, in which affected communities shape the use of a foundation model for a specific downstream task. The intermediate “subfloor” layer scopes the range of potential harms to consider, and affords communities more concrete avenues for deliberation and intervention. At the same time, it avoids duplicative effort by scaling input across relevant use cases. Through three case studies in clinical care, financial services, and journalism, we illustrate how this multi-layer model can create more meaningful opportunities for participation than solely intervening at the foundation layer…(More)”.
How this mental health care app is using generative AI to improve its chatbot
Interview by Daniela Dib: “Andrea Campos struggled with depression for years before founding Yana, a mental health care app, in 2017. The app’s chatbot provides users emotional companionship in Spanish. Although she was reluctant at first, Campos began using generative artificial intelligence for the Yana chatbot after ChatGPT launched in 2022. Yana, which recently launched its English-language version, has 15 million users, and is available in Latin America and the U.S.
This interview has been edited for clarity and brevity.
How has your product evolved since you introduced generative AI to it?
At first, we didn’t use generative AI because we believed it was far from ready for mental health support. We designed and guardrailed our chatbot’s responses with decision trees. But when ChatGPT launched and we saw what it could do, it wasn’t a question of whether to use generative AI or not, but how soon — we’d fall behind otherwise. It’s been a challenge because everyone quickly began developing with generative AI, but our advantage was that, having operated our chatbot for a while, we had gathered over 2 billion data points that have been invaluable for our app’s fine-tuning. One thing is clear: It’s crucial to have a model tailored to the specific needs of our product…(More)”.
Little Communes Everywhere
Review by Jay Caspian Kang: “…I was thinking about all this while I read “The Commune Form: The Transformation of Everyday Life,” a forthcoming book by the comparative-literature professor Kristin Ross. Ross—who has previously written about the Paris Commune of 1871 and France’s student uprising of May, 1968—focusses particularly on the ZAD de Notre-Dame-des-Landes, a thousand-acre commune created by French farmers and their allies in the late two-thousands, in an effort to block the construction of a new airport, which would have kicked many people off their own land. (The French government had designated the land a zone d’aménagement différé, or a “deferred development area”; the farmers kept the acronym but used it to mean zone à défendre, or “zone to defend.”) For a commune to work, Ross argues, one must have both a physical space to defend against an antagonist and an articulated vision for an alternative organization of human relationships and economy. The “commune form,” as she defines it, is a “political movement that is also the collective elaboration of a desired way of life—the means becoming the end.” Theory, in other words, needs to be put into practice, in an intimate and earnest setting, so that people can test out their ideas about living within the context of an actual place among actual people.
Ross identifies one of the motivating forces behind the creation of the ZAD as alienation, which was “less the loss of some human essence than it was the loss of possibilities: the sense of blockages and impasses brought on by the destruction and fragmentation of the social tissue by capitalism.” Drawing upon the work of the French philosopher Henri Lefebvre, Ross refers to “the colonization of everyday life,” each part of our day becoming dominated by economic reasoning. This, she writes, dispossesses us of “our dignity, our social life, our time, the sense of mastery over our lives, the beauty and health of our lived environment, and of the very possibility of working together to invent our future collectively.” Under such conditions, the commune becomes the only alternative…
Physical spaces, whether pools or parks, can be reclaimed through collective action, in much the way that admissions policies at exclusive magnet schools can be protected by a small group of dedicated parents. Small, everyday victories are the only real cure for alienation. What else would work?…(More)”
Collective Intelligence in Open Policymaking
Book by Rafał Olszowski: “This book examines the nexus of collective intelligence (CI), a feature of online projects in which various types of communities solve problems intelligently, and open policymaking, as a trend of large groups of people shaping public policies.
While doing so, it presents the current state of theoretical knowledge for these concepts, many practical examples of successful and unsuccessful projects, as well as additional research and laboratory experiments. The book develops an analytical framework based on qualitative research, which is applied to the analysis of different projects in selected case studies, such as Decide Madrid; Better Reykjavik; Loomio; Deliberatorium; Civic Budget of the City of Kraków.
The book is structured into four chapters, addressing essential questions in the field: (1) Opening Policymaking; (2) Beyond the Individual: Understanding the Evolution of Collective Intelligence; (3) A Review of the Projects Using Collective Intelligence in Policymaking; (4) Online Public Debate. How Can We Make it More Intelligent?…(More)”.
21st Century technology can boost Africa’s contribution to global biodiversity data
Article by Wiida Fourie-Basson: “In spring in the Southern hemisphere, the natural world is on full throttle: “Flowers are blooming, insects are emerging, birds are singing, and reptiles are coming out of their winter hibernation,” wrote Pete Crowcroft, known as @possumpete on the citizen science app, iNaturalist.
Yet, despite this annual bursting forth of life, a 2023 preprint puts the continent’s contribution to the Global Biodiversity Information Facility at a dismal 2.69%, with huge disparities between African countries…
Since its formation in 2008 as part of a graduate project at the University of California, the iNaturalist platform has evolved into one of the world’s most popular biodiversity observation platforms. Anyone, anywhere in the world, with a smartphone can download the app and start posting images and descriptions of their observations, and a large community of identifiers helps to confirm the species’ observation and label it as “research grade”.
Rebelo says iNaturalist is now used on a massive scale: “During the 2023 City Nature Challenge almost 67,000 people made nearly two million observations over four days – that is, five observations each second. Another 22,000 specialists identified 60 thousand species of animals, plants, and fungi. Few citizen science platforms are as powerful and efficient.”..
Andra Waagmeester, data scientist at Micelio in Belgium and a Wikimentor, believes the dearth of biodiversity data from Africa can be solved by combining the iNaturalist and Wikipedia communities: “They are independent communities, but there is substantial overlap between them. By overlaying the two data sets and leveraging the semantic web, we have the means to deal with the challenge.”
The need for biodiversity-related knowledge from Africa was first acknowledged by the Wiki-community during the 2018 Wikimania conference in Cape Town. The Wiki Biodiversity Project has since grown into an active global community that leverages crowd-sourced knowledge from platforms like iNaturalist…(More)”.
More Questions Than Flags: Reality Check on DSA’s Trusted Flaggers
Article by Ramsha Jahangir, Elodie Vialle and Dylan Moses: “It’s been 100 days since the Digital Services Act (DSA) came into effect, and many of us are still wondering how the Trusted Flagger mechanism is taking shape, particularly for civil society organizations (CSOs) that could be potential applicants.
With an emphasis on accountability and transparency, the DSA requires national coordinators to appoint Trusted Flaggers, who are designated entities whose requests to flag illegal content must be prioritized. “Notices submitted by Trusted Flaggers acting within their designated area of expertise . . . are given priority and are processed and decided upon without undue delay,” according to the DSA. Trusted flaggers can include non-governmental organizations, industry associations, private or semi-public bodies, and law enforcement agencies. For instance, a private company that focuses on finding CSAM or terrorist-type content, or tracking groups that traffic in that content, could be eligible for Trusted Flagger status under the DSA. To be appointed, entities need to meet certain criteria, including being independent, accurate, and objective.
Trusted escalation channels are a key mechanism for civil society organizations (CSOs) supporting vulnerable users, such as human rights defenders and journalists targeted by online attacks on social media, particularly in electoral contexts. However, existing channels could be much more efficient. The DSA is a unique opportunity to redesign these mechanisms for reporting illegal or harmful content at scale. They need to be rethought for CSOs that hope to become Trusted Flaggers. Platforms often require, for instance, content to be translated into English and context to be understood by English-speaking audiences (due mainly to the fact that the key decision-makers are based in the US), which creates an added burden for CSOs that are resource-strapped. The lack of transparency in the reporting process can be distressing for the victims for whom those CSOs advocate. The lack of timely response can lead to dramatic consequences for human rights defenders and information integrity. Several CSOs we spoke with were not even aware of these escalation channels – and platforms are not incentivized to promote mechanisms given the inability to vet, prioritize and resolve all potential issues sent to them….(More)”.
The citizen’s panel on AI issues its report
Belgian presidency of the European Union: “Randomly select 60 citizens from all four corners of Belgium. Give them an exciting topic to explore. Add a few local players. Season with participation experts. Bake for three weekends at the Egmont Palace conference centre. And you’ll end up with the rich and ambitious views of citizens on the future of artificial intelligence (AI) in the European Union.
This is the recipe that has been in progress since February 2024, led by the Belgian presidency of the European Union, with the ambition of involving citizens in this strategic field and enriching the debate on AI, which has been particularly lively in recent months as part of the drafting of the AI Act recently adopted by the European Parliament.
And the initiative really cut the mustard, as the 60 citizens worked enthusiastically, overcoming their apprehensions about a subject as complex as AI. In a spirit of collective intelligence, they dove right into the subject, listening to speakers from academia, government, civil society and the private sector, and sharing their experiences and knowledge. Some of them were just discovering AI, while others were already using it. They turned this diversity into a richness, enabling them to write a report on citizens’ views that reflects the various aspirations of the Belgian population.
At the end of the three weekends, the citizens almost unanimously adopted a precise and ambitious report containing nine key messages focusing on the need for a responsible, ambitious and beneficial approach to AI, ensuring that it serves the interests of all and leaves no one behind…(More)”
Sorting the Self
Article by Christopher Yates: “We are unknown to ourselves, we knowers…and there is good reason for this. We have never looked for ourselves—so how are we ever supposed to find ourselves?” Much has changed since the late nineteenth century, when Nietzsche wrote those words. We now look obsessively for ourselves, and we find ourselves in myriad ways. Then we find more ways of finding ourselves. One involves a tool, around which grew a science, from which bloomed a faith, and from which fell the fruits of dogma. That tool is the questionnaire. The science is psychometrics. And the faith is a devotion to self-codification, of which the revelation of personality is the fruit.
Perhaps, whether on account of psychological evaluation and therapy, compulsory corporate assessments, spiritual direction endeavors, or just a sporting interest, you have had some experience of this phenomenon. Perhaps it has served you well. Or maybe you have puzzled over the strange avidity with which we enable standardized tests and the technicians or portals that administer them to gauge the meaning of our very being. Maybe you have been relieved to discover that, according to the 16 Personality Types assessments, you are an ISFP; or, according to the Enneagram, you are a 3 with a 2 or 4 wing. Or maybe you have been somewhat troubled by how this peculiar term personality, derived as it is from the Latin persona (meaning the masks once worn by players on stage), has become a repository of so many adjectives—one that violates Aristotle’s cardinal metaphysical rule against reducing a substance to its properties.
Either way, the self has never been more securely an object of classification than it is today, thanks to the century-long ascendence of behavioral analysis and scientific psychology, sociometry, taxonomic personology, and personality theory. Add to these the assorted psychodiagnostic instruments drawing on refinements of multiple regression analysis, and multivariate and circumplex modeling, trait determination and battery-based assessments, and the ebbs and flows of psychoanalytic theory. Not to be overlooked, of course, is the popularizing power of evidence-based objective and predictive personality profiling inside and outside the laboratory and therapy chambers since Katherine Briggs began envisioning what would become the fabled person-sorting Myers-Briggs Type Indicator (MBTI) in 1919. A handful of phone calls, psychological referrals, job applications, and free or modestly priced hyperlinked platforms will place before you (and the eighty million or more other Americans who take these tests annually) more than two thousand personality assessments promising to crack your code. Their efficacy has become an object of our collective speculation. And by many accounts, their revelations make us not only known but also more empowered to live healthy and fulfilling lives. Nietzsche had many things, but he did not have PersonalityMax.com or PersonalityAssessor.com…(More)”.
Dynamic Collective Action and the Power of Large Numbers
Paper by Marco Battaglini & Thomas R. Palfrey: “Collective action is a dynamic process where individuals in a group assess over time the benefits and costs of participating toward the success of a collective goal. Early participation improves the expectation of success and thus stimulates the subsequent participation of other individuals who might otherwise be unwilling to engage. On the other hand, a slow start can depress expectations and lead to failure for the group. Individuals have an incentive to procrastinate, not only in the hope of free riding, but also in order to observe the flow of participation by others, which allows them to better gauge whether their own participation will be useful or simply wasted. How do these phenomena affect the probability of success for a group? As the size of the group increases, will a “power of large numbers” prevail producing successful outcomes, or will a “curse of large numbers” lead to failure? In this paper, we address these questions by studying a dynamic collective action problem in which n individuals can achieve a collective goal if a share of them takes a costly action (e.g., participate in a protest, join a picket line, or sign an environmental agreement). Individuals have privately known participation costs and decide over time if and when to participate. We characterize the equilibria of this game and show that under general conditions the eventual success of collective action is necessarily probabilistic. The process starts for sure, and hence there is always a positive probability of success; however, the process “gets stuck” with positive probability, in the sense that participation stops short of the goal. Equilibrium outcomes have a simple characterization in large populations: welfare converges to either full efficiency or zero as n→∞ depending on a precise condition on the rate at which the share required for success converges to zero. Whether success is achievable or not, delays are always irrelevant: in the limit, success is achieved either instantly or never…(More)”
Empowered Mini-Publics: A Shortcut or Democratically Legitimate?
Paper by Shao Ming Lee: “Contemporary mini-publics involve randomly selected citizens deliberating and eventually tackling thorny issues. Yet, the usage of mini-publics in creating public policy has come under criticism, of which a more persuasive strand is elucidated by eminent philosopher Cristina Lafont, who argues that mini-publics with binding decision-making powers (or ‘empowered mini-publics’) are an undemocratic ‘shortcut’ and deliberative democrats thus cannot use empowered mini-publics for shaping public policies. This paper aims to serve as a nuanced defense of empowered mini-publics against Lafont’s claims. I argue against her claims by explicating how participants of an empowered mini-public remain ordinary, accountable, and therefore connected to the broader public in a democratically legitimate manner. I further critique Lafont’s own proposals for non-empowered mini-publics and judicial review as failing to satisfy her own criteria for democratic legitimacy in a self-defeating manner and relying on a double standard. In doing so, I show how empowered mini-publics are not only democratic but can thus serve to expand democratic deliberation—a goal Lafont shares but relegates to non-empowered mini-publics…(More)”.