Zooniverse Wins 2025 HAVI Award for AI in Humanities

🔬 GOOGLE'S GEMINI 3.0 PRO SOLVES 500-YEAR-OLD MYSTERY IN NUREMBERG CHRONICLE In a remarkable demonstration of multimodal AI reasoning, Gemini helped decode handwritten annotations that scholars couldn't decipher for centuries. What happened: - Researchers fed high-resolution images of a 1493 printed book into Gemini 3.0 Pro - The AI identified mysterious "roundels" as biblical chronology calculations - Gemini parsed abbreviated Latin, interpreted Roman numerals, and linked them to other passages - The model determined someone was reconciling dates from different biblical traditions Why this matters: This isn't pattern recognition. Gemini combined vision, language, and historical knowledge to solve an applied reasoning problem without human assistance. This points to growing roles for AI in digital humanities, archival research, and historical analysis where vast visual and textual material remains unexplored. Sources: - https://lnkd.in/eqW7wan3 #GoogleAI #Gemini #DigitalHumanities #AIResearch #MachineLearning

Distilling Stanford’s CME295: When Mathematics Meets Ontology Self-studying Stanford’s CME295 (Large Language Models) has been an intellectually rewarding journey. While the mathematical formulas governing the fundamental movements within these structures can be dizzying, the "language" they speak is undeniably fascinating. I hold the highest respect for anyone who can intuitively and perfectly grasp equations like Softmax or Cross-Entropy Loss. Moving beyond the surface, I’ve been dissecting the mechanics of SFT, the LIMA hypothesis, and the structural and technical principles behind QLoRA—specifically how NF4 and Double Quantization achieve efficiency without sacrificing the "intelligence" stored in high-precision adapters. However, the true "aha" moment came from a conceptual shift: I realized that the Transformer’s Attention mechanism is effectively a computational realization of Ontology. In philosophy, Ontology explores the nature of being through relationships; in a Transformer, a token’s identity is defined entirely by its relational context to others. Seeing this philosophical pillar manifest through mathematical weights was a moment of pure intellectual thrill. My previous work with review data curation served as the perfect mental laboratory to ground these abstract theories, allowing me to visualize how raw information transforms into structured knowledge. The bridge between hardware-level optimization and the essence of linguistic meaning now feels solid. It is fascinating to see how deep theory and practical efficiency converge in the current AI landscape. #AI #MachineLearning #StanfordCME295 #Transformers #Ontology #DataScience #QLoRA #SelfStudy #DeepLearning #TechInsight

We are delighted to announce that Adler Zooniverse has been selected for a 2025 Humanities and AI Virtual Institute (HAVI) award from Schmidt Sciences! ✨ HAVI supports groundbreaking work at the intersection of AI and the humanities. The 23 awarded projects—based across multiple continents and disciplines—will use AI to open new windows into human history, culture, language, and creativity. In our project, “Communities in the Loop: AI for Cultures & Contexts in Multimodal Archives,” Adler Zooniverse Co-PI Dr. Samantha Blickhan, Ph.D. Blickhan will collaborate with an award-winning interdisciplinary team to throw open the pages of the early African American press through the power of community and AI. ⭐ Read more about the project here: https://lnkd.in/gHQs3C9H #AdlerPlanetarium #Zooniverse #SchmidtSciences #CitizenScience #ResearchGrant #AIResearch #HumanitiesResearch

New Publication — Aletheia Institute for Integrative Theoretical System Studies The Aletheia Institute announces the publication of a new monograph: The Use of AI as a Discursive Instrument and Research Assistant in Scholarly Inquiry Epistemology, Methodology, and Integrity in Humanities and Social Science Research This work examines the use of artificial intelligence in humanities and social science research from an epistemological and methodological perspective. Rather than treating AI as an epistemic agent, the book develops a human-centered framework in which AI functions strictly as a discursive instrument within researcher-directed inquiry. Key themes include: the limits of AI cognition and authorship, non-dual epistemology and phenomenological verification, dialogical research methodology, and integrity and responsibility in AI-assisted knowledge production. The book is intended for researchers, philosophers, and independent scholars seeking a rigorous, non-ideological account of AI’s proper role in scholarly inquiry. 📘 Available via open access on Zenodo: https://lnkd.in/eciA6Rcd

Here's the biggest AI advancement that I've observed in 2025 (read till the end): #LinkedInNewsEurope It was 2009 when my late university supervisor, Professor Paolo Augusto Masullo, encouraged me to write my dissertation in Philosophical Anthropology on the Singularity, Artificial Intelligence, and Intelligence Amplification, tracing a path from Nietzsche and Deleuze to futurists such as Ray Kurzweil. At the time, Europe felt very different. Within continental philosophy, these subjects were rarely considered “serious” enough for academic work. Meanwhile, across the Channel and across the Atlantic, most major funding flowed toward Metaphysics. I was grateful for the chance to explore such innovative questions, but it was clear that the wider philosophical gaze, both continental and Atlantic, still regarded these topics as peripheral, even speculative. Fifteen years later, in 2025, AI is no longer relegated to fiction or niche fascination. Philosophers everywhere, from Europe to the English-speaking world, are finally giving these themes the attention they deserve. Yet, I can’t help but wonder: if they hadn’t been overlooked for so long, would the gap between exponential technological growth and the much slower (but urgently needed) development of philosophical and ethical frameworks be quite so vast? As the age of AGI approaches, this gap matters. The values we choose will shape the meaning we carry with us into a post-human, super-human future. So, the biggest AI advancement of 2025: A model? A benchmark? No. It’s the bittersweet realisation that we spent decades treating the most important philosophical question of our time as science-fiction fandom.

Assoc Research Prof Tom Lippincott, has been selected for a 2025 Humanities and AI Virtual Institute award from Schmidt Sciences. The $11 million program is supporting 23 teams worldwide that are developing new ways to connect artificial intelligence with humanistic research. His project, "An ML Toolkit to Find Hierarchical Structure in Multimodal and Multilingual Data," produce a Python library for humanities scholars to study structures and patterns within large-scale oeuvres of poetry, music, text and other material. Co-PIs are Prof Meredith Martin (Princeton), Prof John Hale (Hopkins) and Asst Prof Robert Lieck (Durham). Lippincott’s work focuses on how machine learning can support and extend traditional scholarship in the humanities, with an emphasis on interpretable, unsupervised models of cultural production and interpretation. He holds roles across the university, including director of the Center for Digital Humanities, affiliate of the Center for Language and Speech Processing, and member of the Alexander Grass Humanities Institute and the Data Science and AI Institute.

❄️Stop 20: University of Ljubljana After a chilly walk along the canals in the Slovenian capital, Researchy steps into the warmth of the University of Ljubljana, ready to learn about how AI can strengthen SSAH. 🐧✨ The Horizon Europe Widening-funded centre of excellence Artificial Intelligence for Digital Humanities (AI4DH) aims to empower digital humanities and social science researchers to uncover previously hidden patterns, deepen their analyses and gain new insights at a much faster pace by using AI technologies. Researchy has regained the warmth in his wings and new faith in interdisciplinarity between technology and social sciences. Future discoveries are just around the corner! 💫 🌐 Learn more: 🔗 https://ai4dh.eu/ Click here to follow the journey ➡️ https://lnkd.in/e-d3xgpe Antoine Doucet- Marko Robnik-Šikonja - Špela Arhar Holdt - Slavko Žitnik - Polona Tratnik - Simon Krek

Only 21 days are left until 2026. Today, we are having research associate Martin Kutz (Technische Universität Dresden) on the #ScaDSAICountdown. Martin Kutz is working in the research area "Responsible AI" at ScaDS.AI Dresden/Leipzig and at the Chair of Religious Education (Protestant) of our principal investigator Prof. Birte Platow. In his dissertation project, he reviews use cases of conversational agents in religious contexts and wants to establish the conditions for a successful interaction between human and machine in such contexts. But what inspires Martin Kutz in his work? "A few years ago, the so-called Richter Window was completed in Cologne Cathedral. Many tourists make a pilgrimage to the cathedral just to see it, sitting in the soft, warm light that bathes the grey, old stone in a sea of colour. Actually, it is nothing special. Unlike the other large windows in the church building, it does not show detailed images of saints. It simply consists of over 11,000 glass squares in 72 colours, arranged at random. It was created by Dresden-born artist Gerhard Richter. The only connection to the other windows is that a selection of their colours has been incorporated. What fascinates me about the window – apart from its effect – is the interplay of technology (random generator) and craftsmanship (mouth-blown glass). The artist combines both to create an impressive work of art that is inspiring and perhaps even invites to reflect on religious questions. It inspires me in my research on the social impact of AI to ensure that humans are not displaced or replaced by technology, but rather to look at where humans and technology can create something new and meaningful together. This is happening, for example, in my project with the University of Siegen: we have set up a robot in a classroom. The young pupils recognised it in its role as an expert and did not see it as a replacement for the teacher. In addition, it helped the children in religious education to reflect on interacting with an intelligent social robot in a responsible manner." - Martin Kutz

This move by Schmidt Sciences is bold and forward-looking. If you are seeking insights into how AI works and does not work in real world scenarios, you should keep an eye on these projects. Several of the projects apply AI to study of the past while others focus on critical engagement with the technology itself. Each will provide important insights for anyone who is interested in how to apply AI most effectively, and learn where the weaknesses and potential lie. I believe the results of these projects will revolutionize applications of AI, architectures of AI, and our relationship to the technology because they are driven by humanities research questions. And the findings will change how we use AI in the private sector, in government, and in non-profit organizations. The mechanical aspects of vision and language production are reflected in the design of current AI architectures, but the how and why of human knowledge production are not. Each one of these Schmidt Sciences funded projects brings nuance to the AI conversation by considering the historical, geographic, cultural, and epistemological context of the information used to train and test the models. The questions pursued in these projects are not abstract or generalized but culturally specific and richly complex. Some explore cultural production of the past and others look to the future. One of many projects that focuses on how knowledge evolves looks at how scientists communicate ideas through images. Another uses vision-language models to trace patterns in iconography from medieval manuscripts. Two of the projects investigate AI and the law: one looking at how AI influences judges and lawyers who use large language models in their work, and another looking back at medieval law and with an emphasis on the distortions that arise in translation and over time. The project that I am involved in is, in my contribution, influenced by 20 years at Stanford Libraries working with researchers who wanted computational systems that allow information to evolve and grow as new discoveries are made. Our team, led by Giovanna Ceserani and Sebastian Ahnert, with Michele Mauri (Politecnico di Milano), Allen Romano, Ph.D., Yang Diyi, Ali Yaycioglu, and Richard Roberts, will create an AI architecture that is modeled on scholarship and the library and archival practices that support scholarship. We believe that this approach will preserve future generations’ ability to access and understand knowledge in its full complexity.