The Collective Intelligence Project

Meet Fast Forward’s 2025 Accelerator cohort! This year’s crew is so cool. Seriously. They're doing amazing work across healthcare, education, climate, and more. Plus, 70% of the cohort is an AI-powered nonprofit. These teams are redefining what tech can do: Bayes Impact, Clinic+O, The Collective Intelligence Project, Materiom, Project Own, Scrutinize, trubel&co, Vector Control Innovations, Hola Soy Violetta, and Visilant. None of this happens without our partners. We are so grateful to Google.org; Bloomberg; Salesforce; Okta; Twilio; Arrow Impact; The SOLID Foundation; Endless; Splunk, a Cisco company; LEMNIS; New Relic; OpenAI; and GitHub. Learn more about the cohort and their impactful work here: https://bit.ly/ffwd2025

Excited to share this week’s Collections from Collective Intelligence Project—a curated look at public AI, global governance, and democratic tech. Our latest highlights cover everything from breakthrough open-source models and safety insights to policy debates that are shaping the future of AI. Public AI in Action Public AI is emerging as a powerful alternative to proprietary models. The Barcelona Supercomputing Center’s Alia model is a prime example of how trusted institutions can offer open, cost‐effective AI infrastructure to spur economic growth. AI Safety and Accountability As AI technologies advance, so do the challenges of safety and accountability. The International AI Safety Report, led by @Yoshua_Bengio, lays out pressing issues and potential solutions for managing AI risks on a global scale. Championing Democratic Governance In a recent opinion piece, our Senior Fellow Audrey Tang explores how democratic governance can counter tech dominance. Her insights underscore why transparency and public oversight are essential to steer AI development for the common good. Global AI Policy Shifts With AI export policies and international dynamics evolving rapidly, understanding their impact on global influence is crucial. The Carnegie Endowment’s podcast offers a deep dive into how these shifts might redefine US and global power balances. Looking Ahead to Paris The upcoming Paris AI Action Summit promises to be a pivotal event for discussing the role of open-source models, global governance, and democratic values in shaping AI’s future. Senior Fellow Audrey Tang and Executive Director Divya Siddarth will represent our commitment to collective intelligence and democratic approaches to AI governance in Paris. Launching Global Dialogues During the week of the summit, we’ll also be launching our Global Dialogues project (www.globaldialogues.ai) to foster collective conversations about the future of AI—exploring what people value and the kind of future they envision. We have more exciting updates coming in the next few weeks. Follow us here or sign up for our mailing list at cip.org to stay in the loop.

This week's Collections: Open Source & Global Governance The intersection of open-source AI and global governance is reshaping our technological landscape. Here are five key developments that caught our attention this week: 1?? Brazil's Bold Move in AI Regulation Brazil is emerging as a pioneer in balanced AI governance, setting new standards for combining innovation with ethical oversight. Their approach offers a compelling roadmap for other nations navigating these complex waters. 2?? Breaking Language Barriers in AI Recent research from Cohere on multilingual LLM benchmarking highlights a crucial mission: creating AI that truly serves global communities. We're focused on building AI that understands and respects cultural nuances. 3?? The release of DeepSeek-R1 shows that there is still a role in open-source for advancing AI capabilities. What does it mean for the economics and governance of AI when compute and scale no longer matter as much? 4?? Shifting Power Dynamics in AI Policy: Recent developments in global AI regulation reveal fascinating insights about who shapes our AI future. Big tech's influence on policy decisions is creating ripples across the international governance community. From Rest of World 5?? The Human Face of Technology A powerful reminder from this week: Behind every line of code and policy decision are real human stories. The winners of Rest of World's tech photography contest beautifully capture this reality. ?? Why This Matters The decisions we make today about open-source AI and global governance will shape tomorrow's technological landscape. What's your take? How do you think we can ensure AI development serves everyone, not just a select few? [Links below]

?? Reimagining a More Democratic Internet & AI Landscape This week’s CIP Collections explores how we govern, learn, and connect—shaped by technology but grounded in human values. Here are the highlights: 1?? Decentralizing the Internet A new foundation - freeourfeeds.com - is stepping up to independently govern the AT Protocol, the decentralized backbone of platforms like Bluesky. It’s time to give users—not corporations—control over the internet. 2?? AI and Education Can large language models transform education? Early results from a World Bank pilot in Nigeria show promise—but questions remain about how to redesign our education systems to fully harness this potential. 3?? AI Safety and Governance The UK AI Safety Institute represents a bold experiment in public-interest AI governance. Building public sector capacity to evaluate, shape, and govern AI is critical for ensuring its benefits are broadly distributed. 4?? Bridging Divides How do we foster more productive dialogue across divides? MIT’s Bridging Dictionary maps how polarized groups interpret key terms differently, offering a powerful tool for healthier conversations. 5?? Rethinking Internet Infrastructure Frank McCourt’s Project Liberty envisions an internet built around user autonomy. His bid to acquire TikTok isn’t about algorithms—it’s about creating a “public digital infrastructure” where users have agency. 6?? AI and Elections Despite fears, AI didn’t upend elections in 2024’s "Super Year." Robust safeguards, public awareness, and transparent governance played a crucial role—a hopeful case study for democracies navigating AI’s growing influence. From rethinking internet governance to exploring AI’s role in education and democracy, these stories show how human values can guide technological progress. ?? What resonated most with you? Let’s discuss!

This week, our team at CIP watched with heavy hearts as wildfires threatened our loved ones in Los Angeles. Throughout the crisis, a community-powered app called Watch Duty has been a critical lifeline. Run by a volunteer force of retired firefighters, tech workers and local experts, it transformed complex fire data and radio communications into real-time intelligence that helped keep communities safe. Our first Weekly Collections for January– exploring notable developments in collective intelligence and technology. Platform Governance at Scale: Meta's recent shift from third-party fact-checkers to Community Notes marks an intriguing experiment in content moderation. As information moves faster than traditional verification methods can handle, this new approach could offer interesting insights into platform governance that scales with modern challenges. Democratizing the AI Ecosystem: Divya Siddarth was on the Untangled podcast with Charles Johnson to speak about our mission to democratize AI development and build the infrastructure for collective intelligence. Tech Meets Culture: In the suburbs of Phoenix, Arizona, TSMC's new semiconductor fab has given rise to "Tiny Taipei," showing how major tech infrastructure investments create unexpected cultural ripples. As semiconductors become increasingly crucial to 21st-century economies, these developments are worth watching. Digital Public Squares: An intriguing new position paper from Jigsaw, of which Divya Siddarth was a co-author, examines how AI might contribute to healthier public conversations online. The research raises important questions about moderating discourse at unprecedented scales. Creativity in the AI Age: A thought-provoking piece in @Bostonreview asks a fundamental question about AI and copyright: To whom does the world belong? It's a fascinating exploration of how we might balance innovation with existing cultural frameworks. [Links to all referenced articles in comments below]

2024 in Review: Building Democratic AI Through Collective Intelligence As we reflect on 2024, a year marked by surging AI capabilities, our mission to build democratic AI futures through collective intelligence has never been more vital. Here's what we accomplished: Launched Global Dialogues: We built a repeatable and scalable process for fast, fair, and fun global input into AI development, enabling people to express what they actually want from AI systems. globaldialogues.ai Pioneered Community Models: We developed tools that experiment with plural alignment at different scales, giving people and communities direct agency in shaping AI models. cm.cip.org Expanded Global Partnerships - Our collaborations now span from Bhutan to Montenegro, from India to Brazil, demonstrating that AI development extends far beyond Silicon Valley and DC. Key Highlights: ? Published our Roadmap to Democratic AI ? Produced cutting-edge research with global partners on directing transformative technologies ? Welcomed new team members expanding our global expertise ? Developed scalable processes for global AI governance input Looking ahead to 2025, we're scaling global dialogues, expanding community models, and continuing to build tools that leverage frontier capabilities to enhance human coordination. The next few years are of outsize importance in shaping the future. While AI advances have been incredible, they must be matched by advances in collective intelligence. We believe this is possible through rapid, concerted effort, and we look forward to tackling these challenges together in 2025. Join us in shaping democratic AI futures: cip.org If you'd like to support our work: cip.org/donate

The Collective Intelligence Project is two! In 2024, we built real tools for real people, created infrastructure for global listening, and grew the team. Our Annual Report is live - check it out 2024.cip.org Most work in democratizing AI is still tinkering around the edges. If real transformation is coming, we aren’t ready for it. We’re working on putting necessary collective intelligence into place before it’s too late: building repeatable processes for fast, fair, and fun global input (thanks Audrey Tang) globaldialogues.ai Many people don’t see themselves as global citizens first and foremost—they see themselves as family members, community members, workers, and friends. We’ve built community models to experiment with plural alignment at a different range of scales. cm.cip.org We’re working against cultural homogenization by expanding our partnerships. CIP has spent the year in conversation with collaborators from Bhutan to Montenegro, from India to Brazil, and many places in between. Let’s not just get blinded by the minutiae of DC discourse on chips: the world is big and the possibilities are great. Finally—we’re growing! Read the report to meet our new team members, fellows, and advisors. And snag an invite to our post-holidays holiday party in January. Our work is powered by people who understand that the future is going to be crazy, love humanity and know that the world can be better than it is. Read the full report and join our mission in 2025: 2024.cip.org

?? Collective Intelligence Weekly Roundup ?? This week, we’re exploring the transformative power of collective systems—how shared data, tools, and technology can reshape our world. Here’s what stood out: 1. Public Diffusion: The Future Isn’t Owned, It’s Shared Proving a radical idea: you don’t need infinite data—you need curated data. Spawning has built Public Diffusion, a model trained solely on public domain images, that rivals top-tier AI art models. Publicly-owned AI is possible. Sign up for the beta: https://lnkd.in/eXmJquNp 2. Tech Lifelines in Crisis From conflict zones to natural disasters, platforms like WhatsApp are critical infrastructure. They enable resilience, connect communities, and provide collective support when it’s needed most. Full story from Rest of World: https://lnkd.in/e4WG8Jpm 3. Rethinking Connection: On the Grid The Drift dives into how systems meant to "connect" us often alienate us instead. How do we build platforms that foster genuine collaboration and belonging? https://lnkd.in/eRAvz9tv 4. Beyond the Brain The Atlantic explores how human cognition extends into tools, environments, and collective systems. How does this shape the way we think about intelligence—both human and artificial? Discover more: https://lnkd.in/eUnpJFKU 5. Humanitarian AI: Boon or Risk? WIRED explores how NGOs are navigating the promise and perils of AI. How can we design these tools to amplify human rights, not undermine them? Read more: https://lnkd.in/dvRwgsKQ ?? These stories remind us of a powerful truth: The future is built through collective action and shared systems. Which one resonates most with you? Let us know below!

Weekly Collections! We took a break last week, and now we're back. This week's theme: how do we ensure AI amplifies rather than replaces human connection and cultural diversity? 1. This piece from The New York Times - is human interaction is becoming a "luxury good? https://buff.ly/3VjaU6E 2. From Cohere: KAHANI, a new culturally-nuanced visual storytelling pipeline, showing how thoughtful AI design can preserve and amplify cultural authenticity rather than defaulting to Western perspectives. https://buff.ly/3D6Ffz3 3. This connects beautifully to INCLUDE, also from Cohere a new benchmark for evaluating multilingual LLMs with regional knowledge. At CIP, we've been thinking about how to make sure that AI incorporates diverse knowledge systems. https://buff.ly/49riH86 4. This fascinating piece on biomedical AI agents frames them as collaborators rather than replacements. Is this a promising paradigm for the future of science? https://buff.ly/41ItKrV 5. Epoch AI's new AI Benchmarking Hub just launched. The maturation and safe adoption of AI requires robust, transparent evaluation frameworks. Kudos to the team! (P.S Keep your eye out for something new from The Collective Intelligence Project coming soon). https://buff.ly/3Di83o0 6. Collective intelligence is also about connecting people in unexpected ways. This archive of iPhone uploads (2009-2012) creates a shared experience across time & space. Pure digital serendipity --> https://buff.ly/3ZKRIkw

Another installment of our weekly Collections: 1. Our own James Padolsey on "The AI Safety Paradox" - why focusing too much on making individual AI models 'safe' might actually make our systems more dangerous. https://lnkd.in/gDXfYw-D Just as we build banking systems assuming tellers might make mistakes, we need to build robust, resilient AI ecosystems that assume models will fail - not just "ethical" AI at the individual model level. 2. Is the grass greener under blue skies? You may have heard that many are joining Bluesky. Read an essay from Mike Masnick that inspired it: "Protocols, Not Platforms" which argued for shifting from centralized platforms to open protocols. https://buff.ly/2ZBL7dZ 4. Puja Ohlhaver's "Common Knowledge Machines" explores how to redesign X to surface genuine consensus rather than amplifying division. https://buff.ly/4fFwudH Her proposed "Community Posts" mechanism shows how careful system design could help bridge polarizing divides - echoing our belief that good institutional design is key to collective intelligence. 5. Fascinating work from Stanford shows how we might use AI to better understand collective behavior: their generative agents predict human responses with 85% accuracy compared to people's own consistency with themselves https://buff.ly/3OJhks5 6. What are your thoughts on these pieces? How do we better understand and design for both individual and collective behavior?