text updates - ecosystems

data/artifacts.json

@@ -236,7 +236,8 @@
       "ecosystems"
     ],
     "topics": [
-      "power"
+      "power",
+      "economy"
     ],
     "url": "https://huggingface.co/blog/evijit/hf-hub-ecosystem-overview"
   },
@@ -548,10 +549,12 @@
     "type": "paper",
     "description": "Organisational AI policies in newsrooms and universities reveal practical, domain-specific approaches to managing risks like bias, privacy, and environmental impact — areas often underaddressed in top-down regulations like the EU AI Act. These bottom-up guidelines offer actionable insights on AI literacy, disclosure, and accountability that can inform more adaptive and effective global AI governance. The study argues for integrating real-world organisational practices into international regulatory frameworks to bridge implementation gaps.",
     "areas": [
-      "ecosystems"
+      "ecosystems",
+      "agency"
     ],
     "topics": [
-      "power"
+      "economy",
+      "community"
     ],
     "url": "https://arxiv.org/abs/2503.05737"
   },
@@ -652,10 +655,12 @@
     "type": "paper",
     "description": "This work introduces ELLIPS, an ethical toolkit designed to guide researchers in developing language model-based systems for inferring psychiatric conditions, ensuring alignment with clinical needs and ethical principles. It emphasizes integrating autonomy, beneficence, justice, transparency, and social responsibility into every stage of model development—from data selection to deployment—to prevent harm and enhance real-world applicability. By advocating for stakeholder-inclusive, transdiagnostic, and multilingual approaches, the framework aims to shift the field from convenience-driven research toward impactful, equitable mental health technologies.",
     "areas": [
-      "agency"
+      "agency",
+      "ecosystems"
     ],
     "topics": [
-      "personal"
+      "personal",
+      "economy"
     ],
     "url": "https://ojs.aaai.org/index.php/AIES/article/view/31720"
   },
@@ -678,9 +683,11 @@
     "type": "paper",
     "description": "Wikimedia data has been foundational for AI and NLP development, yet the relationship remains one-sided, with little reciprocal benefit to Wikimedia editors. This review calls for expanding the diversity and multilingualism of Wikimedia-derived datasets, embedding core content policies like neutrality and verifiability into benchmarks, and prioritizing open, compact models that serve the needs of the Wikimedia community.",
     "areas": [
+      "agency",
       "ecosystems"
     ],
     "topics": [
+      "community",
       "power"
     ],
     "url": "https://arxiv.org/abs/2410.08918"
@@ -865,7 +872,7 @@
       "ecosystems"
     ],
     "topics": [
-      "power"
+      "economy"
     ],
     "url": "https://shura.shu.ac.uk/33307/"
   },

js/data/areas.js

@@ -115,7 +115,7 @@ export const areasData = {
         description: {
             short: 'AI systems are embedded in economic, regulatory, and market ecosystems that shape and are shaped by their development.',
             paragraphs: [
-                'Understanding AI systems requires understanding the economic, regulatory, and market ecosystems that shape and are shaped by its development. These ecosystems  determine the effectiveness of different approaches to the development, governance, and commercialization of the technology –  and the most effective strategies to ensure positive outcomes for stakeholders both in and outside of its development settings.',
+                'AI systems are shaped by economic, regulatory, and market ecosystems – as these in turn are shaped by the technology. Understanding these interactions, including through the analysis enabled by more transparent systems and the collaborative experimentation supported by open models and datasets, fosters more effective development, governance, and commercialization of the technology to ensure more positive outcomes for stakeholders both in and outside of its development settings.',
                 'More open and transparent technology enables both a better study of the interactions between these ecosystems and the development of tools and versions of the technologies that can better avoid pitfalls of labor and economic displacement, excessive concentration of resources and market power, or of regulation under strong epistemic asymmetries between policymakers and large developers. In particular, open research and development enables more direct collaboration between diverse developer profiles, legislators, adopters, advocates, and other economic actors – with less dependence on access and information provided by large model developers.',
             ],
         },
@@ -132,42 +132,40 @@ export const areasData = {
         imageAltText: 'A simplified illustration of urban life near the sea showing groups of people, buildings and bridges, as well as polluting power plants, opencast mining, exploitative work, data centres and wind power stations on a hill. Several small icons indicate destructive processes.',
         imageSourceUrl: 'https://betterimagesofai.org/images?artist=LoneThomasky&title=DigitalSocietyBell',
         topics: {
-            economy: {
-                id: 'economy',
-                name: 'Economic and Labor Impacts of AI',
-                description: {
-                    short: 'How AI systems affect the economy and labor conditions.',
-                    paragraphs: [
-                        'AI is transforming work across industries from logistics and finance to media, software development, and customer service. As models are integrated into larger commercial systems, they increasingly shape how tasks are designed, monitored, and valued. The centralization of data and the ensuing integration of automated information processing, pattern recognition, and content generation tools is changing how digital and creative work is supported, organized, and distributed, and who benefits from it. These changes redefine labor and economic sectors not just through job displacement, but through the redesign of entire production systems.',
-                        'Understanding where AI systems can support workers and economic outcomes, how different kinds of deployments may favor or harm different parties, and how control of work data flows can either consolidate or displace value creation for adopters of AI systems all contribute to making sure that they are developed and deployed to benefit the workforce and industry sectors that leverage the technology, not just its developers.',
-                        'Openness and transparency into the models and datasets that support commercial applications of AI, particularly in “enterprise” settings and for specific domains of activities, supports more positive outcomes and broadly distributed benefits in two major ways. First, by enabling the development of new AI systems to be driven more directly by the economic actors who want to leverage them, maintaining control of their supply chain, expertise, and value propositions. Second, by enabling more scrutiny into the systems themselves, and supporting more robust and independent analysis of the labor and economic impacts of the technology – rather than relying on the framings and promises of developers – to better guide economic policy and strategy.',
-                    ],
-                },
-                color: 'bg-yellow-100 text-yellow-800',
-                gradient: 'from-yellow-50 to-yellow-100 hover:from-yellow-100 hover:to-yellow-200 border-yellow-200 hover:border-yellow-300 text-yellow-900'
-            },
             power: {
                 id: 'power',
                 name: 'De-Centralized Markets, Development, and Sovereignty',
                 description: {
                     short: 'Market concentration dynamics and technological sovereignty questions.',
                     paragraphs: [
-                        'The narratives and development of AI today are disproportionately shaped by a handful of actors who control the largest models, datasets, and compute infrastructure. This concentration of technical and financial power doesn’t just shape (and constrain) innovation – it defines which versions of the technology are given priority, who sets its norms, what values it encodes, and who benefits most from its integration into all aspects of society. As these dependencies deepen, they also raise questions of digital and technological sovereignty for nations and collectivities aiming to set their own terms for their digital infrastructures.',
-                        'A more balanced and resilient AI ecosystem depends on distributing the capacity to develop, study, and govern these systems. Maintaining this capacity to develop alternative versions of the technology beyond that of a few actors concentrating the majority of the computational and data resources requires a wide range of strategies, from checking abuses of market power to ensure broad participation in AI development remains incentivized and sustainable to lowering the technical and financial barrier to entry for all categories of actors; benefiting both small start-ups and public institutions and larger economic organizations resources who still may not want to reach the excesses of compute expenses of the largest developers.',
-                        'Fostering a broad ecosystem of open AI models, datasets, and tools and a thriving open research environment across universities and independent developers has a dual role to play in mitigating the risks of extreme concentration. First, it enables a better understanding into the tradeoffs involved in the development of the technology, including specifically in characterizing the role of different categories of resources and the risks their capture may pose to a competitive ecosystem. Second, it drastically reduces the cost of developing new AI systems, or of adapting existing AI technology to the needs of various actors; and allows them to control their data flows to ensure their longer-term welfare.',
+                        'The narratives and development of AI today are disproportionately shaped by a handful of actors who control the largest models, datasets, and compute infrastructure. This concentration of technical and financial power doesn’t just shape (and constrain) innovation – it defines which versions of the technology are given priority, who sets its norms, what values it encodes, and who benefits most from its integration into all aspects of society; raising questions of digital and technological sovereignty for nations and communities aiming to set their own terms for their digital infrastructures.',
+                        'A more resilient path forward requires cultivating a distributed ecosystem — one that checks abuses of market power and invests deliberately in open infrastructure. Fostering open AI models, datasets, tools, and a thriving research environment plays a dual role in mitigating concentration: first, by deepening understanding of the tradeoffs in development — including how resource capture threatens competition; second, by lowering barriers to entry, enabling actors of all sizes — from individuals to institutions — to adapt and secure their AI systems without replicating the computational excesses of the largest developers. Together, these efforts enable a more sustainable ecosystem, where adopters retain control over their data and the value of their work, and can drive innovation aligned with diverse needs.',
                     ],
                 },
                 color: 'bg-red-100 text-red-800',
                 gradient: 'from-red-50 to-red-100 hover:from-red-100 hover:to-red-200 border-red-200 hover:border-red-300 text-red-900'
             },
+            economy: {
+                id: 'economy',
+                name: 'Economic and Labor Impacts of AI',
+                description: {
+                    short: 'How AI systems affect the economy and labor conditions.',
+                    paragraphs: [
+                        'AI is reshaping work across industries, from logistics and finance to media and customer service, by embedding automated processing, pattern recognition, and content generation into commercial systems – along with new standards for efficiency, output, and oversight – often without input from those who rely on these systems daily. These changes reconfigure not just job roles but entire production structures, redefining how work is organized; its quality, autonomy, and safety; and where economic value accumulates.',
+                        'Open models and transparent data practices empower a broader range of economic actors — including small businesses and specialists within economic domains — to adapt AI to their operational contexts, maintain control over their supply chain, expertise, and value propositions, and conduct independent analysis of its economic and labor impacts. Because those who deploy and live with these systems daily understand their effects — including unintended consequences, workflow disruptions, and hidden costs — better than centralized developers pushing top-down adoption, this openness enables workers and participants in economic production to shape AI to their needs, rather than conforming to standardized, one-size-fits-all solutions. Transparency doesn’t just enable scrutiny; it shifts the basis of economic policy, investment strategies, and AI product development from promotional claims to observable, shared realities.',
+                    ],
+                },
+                color: 'bg-yellow-100 text-yellow-800',
+                gradient: 'from-yellow-50 to-yellow-100 hover:from-yellow-100 hover:to-yellow-200 border-yellow-200 hover:border-yellow-300 text-yellow-900'
+            },
             regulation: {
                 id: 'regulation',
                 name: 'Rights and Regulation',
                 description: {
                     short: 'How AI systems are regulated and how they affect rights and regulations.',
                     paragraphs: [
-                        'Regulation of AI systems, both through the design of new technology-specific rules of through the exploration of how existing laws apply to the new technical paradigms it introduces, has drawn increasing attention commensurate with the ubiquity and visibility of the technology. Notably, the unprecedented scales of data and automation can present unique challenges, and different interests and perspectives on the technology from different categories of stakeholder have raised important questions about which risks to prioritize in legislative actions, and how to arbitrate between different tradeoffs – including when considering how to apply proposed rules to open and open-source systems, which often receive significantly less consideration in drafting processes than diversity of development contexts and developer and researcher profiles who participate in it would warrant.',
-                        'Effective AI regulation requires a deep understanding of the technology\'s inner workings, including its inherent trade-offs and the feasibility of technical interventions. Open access to AI systems is crucial for this, as it empowers independent legal and domain experts to conduct research and assess the technology without relying solely on the interpretations of a few powerful developers. Furthermore, open research and centralized resources for legal compliance lower barriers to participation, which is vital for the often less-resourced organizations that produce much of the most publicly beneficial work. This openness extends beyond code to include open datasets, models, and transparent decision-making, enabling a broader community to help shape AI. Overall, collaboration on the technical artifacts and legal tools that shape the design and governance of artificial intelligence is essential to ensure the sustainability of regulatory efforts that serve all of the people whose lives it shapes.',
+                        'The regulation of AI increasingly grapples with how to apply existing legal frameworks – and design new ones – to systems whose scale, opacity, and dynamism present unprecedented challenges. Effective tools and processes for regulation require direct engagement with the technical characteristics of AI systems; an engagement uniquely enabled by open research and development. This in turn means that compliance requirements must account for the diverse needs of open and collaborative work across smaller institutions to avoid excluding all but the largest commercial actors from meaningfully participating in the technology’s development – undermining the very research needed to sustain meaningful governance.',
+                        'By collaborating on tools and norms that operationalize rights and regulations in open settings – allowing direct collaboration with legal and social expertise and rightsholders, providing transparency that allows questioning practices earlier, and facilitating the development of more rights-respecting versions of the technology – open developers and researchers enable continuous alignment between technological evolution and societal expectations. These practices, rooted in the open-source and open science ethos, facilitate integrating foresight and building with more regulatory security for especially small and medium developers; resulting in a system where innovation and accountability are mutually reinforcing, not separate domains.',
                     ],
                 },
                 color: 'bg-purple-100 text-purple-800',

js/pages/HomePage.js

@@ -49,7 +49,7 @@ export function renderHomePage() {
                 </p>
                 
                 <p>
-                    In the broader context of ${createInlineLink('Hugging Face', 'https://huggingface.co', 'bg-yellow-100 text-yellow-800')}'s efforts to support the open sharing and development of AI systems,
+                    In the broader context of ${createInlineLink('Hugging Face', 'https://huggingface.co', 'bg-yellow-100 text-yellow-800')} efforts to support the open sharing and development of AI systems,
                     the ${createInlineLink('Machine Learning and Society Team', 'https://huggingface.co/hfmlsoc', 'bg-blue-100 text-blue-800')} works on projects targeting topics at the boundary between technology and society.
                     These include measuring and improving the ${createInlineLink('Sustainability', '/sustainability', areasData.sustainability.color, areasData.sustainability.description.short)} of the technology,
                     characterizing and supporting the ${createInlineLink('Agency', '/agency', areasData.agency.color, areasData.agency.description.short)} of individual and communities in their interactions with AI,