Erik Van Heeswijk

🚀 How to boost innovation with generative AI?  Get the latest insights in Pioneering Tech 👉 https://edu.nl/6wevh Alessandro Bozzon Mondai | House of AI Nederlandse AI Coalitie | NL AIC #ailabs #generativeai #kunstmatigeintelligentie #scientificresearch #government #pioneeringtech #innovation #impact #technology #tudelft

1

At Moxie (Academic Insight Lab), we recently started an advice column called "𝑫𝒆𝒂𝒓 𝑴𝒐𝒙𝒊𝒆," where we answer real questions from our members and subscribers. 🎓 𝐎𝐮𝐫 𝐥𝐚𝐭𝐞𝐬𝐭 𝐢𝐧𝐬𝐭𝐚𝐥𝐥𝐦𝐞𝐧𝐭 𝐟𝐞𝐚𝐭𝐮𝐫𝐞𝐬 𝐚 𝐭𝐡𝐨𝐮𝐠𝐡𝐭𝐟𝐮𝐥 𝐢𝐧𝐪𝐮𝐢𝐫𝐲 𝐟𝐫𝐨𝐦 𝐚 𝐏𝐡𝐃 𝐬𝐭𝐮𝐝𝐞𝐧𝐭: "I'm concerned that by using AI-driven tools, my thesis might feel like it's being written by a 'third party.' 𝑨𝑰 𝒊𝒔 𝒊𝒏𝒄𝒓𝒆𝒅𝒊𝒃𝒍𝒚 𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒕, 𝒃𝒖𝒕 𝒂𝒔 𝑰 𝒄𝒐𝒏𝒔𝒊𝒅𝒆𝒓 𝒊𝒏𝒕𝒆𝒈𝒓𝒂𝒕𝒊𝒏𝒈 𝒕𝒉𝒆𝒔𝒆 𝒕𝒐𝒐𝒍𝒔, 𝑰'𝒎 𝒘𝒐𝒓𝒓𝒊𝒆𝒅 𝒂𝒃𝒐𝒖𝒕 𝒍𝒐𝒔𝒊𝒏𝒈 𝒄𝒐𝒏𝒕𝒓𝒐𝒍 𝒐𝒗𝒆𝒓 𝒎𝒚 𝑷𝒉𝑫. It seems too easy, too quick. What about the traditional academic journey filled with its unique challenges and deep engagement with the subject? How do I maintain the depth of analysis and personal touch in my thesis?" 💫 When writing our response, we created this useful graphic to help graduate students and researchers determine when it might be appropriate to use AI-generated feedback. 🔗 https://lnkd.in/e5u47bBM

20

1 commentaar

The Humans Have Entered the Chat - panel discussion Very excited to be heading to International Documentary Film Festival Amsterdam (IDFA) DocLab later this week, to present the research with MIT Open Documentary Lab on XR embodiment & AI we released recently and share plans for this year's research focus with Sarah Wolozin and Scarlett Kim (more on this to come). I'll also be moderating a panel discussion on AI, 'The Humans Have Entered the Chat', with filmmakers and artists, description below. If you'll be at IDFA, come along! Saturday 16 Nov, 3-5pm, ITA: Studio 1. The Humans Have Entered the Chat What do we see when we look at an AI creation? So often it seems we see something in the shape of ourselves. Why are we driven to anthropomorphize these machinal outputs—are we looking for self-recognition, connection with others, or something else entirely? In the maelstrom that is the current discourse around AI, it can be hard to find direction and certainty. Audiences want to understand where the (real) human is in the loop, and who the trusted guides are that can help us navigate this rapidly shifting space. Documentary makers and artists can play such a guiding role—they after all are experimenting with many new AI-based tools, or using traditional screen practices to challenge our ideas of the human. They can also point us towards some of the potential creative benefits of the tools that move beyond our self-image, into forms of collective and collaborative intelligences. Our actual, live human panelists include the makers of Oryza: Healing Ground (IDFA DocLab Competition for Digital Storytelling); AI & Me (IDFA DocLab Spotlight); and Replica (IDFA Forum). Join us as we dig into the human, the nonhuman, the artificial and the real. Panelists include Tamara Shogaolu, mots (Daniela Medovescu + Octavian Mot) and Chouwa Liang.

17

My presentation proposal 'Metadata madness? The power and complexity of combining article and journal metadata from 30+ data sources in a data warehouse' has been accepted for the NISO Plus 2024 Global/Online conference. To support negotiations, contract management and national open access strategies with high quality data, the Dutch library consortium UKB started building a data warehouse in 2020. In the past four years this datahub grew both in scope and impact. It contains metadata from almost half a million Dutch research articles from 2018 till present, obtained from 30+ sources. More than 200 institutions, 700 publishers and 21.000 journals are covered. Challenges in the quality and availability of metadata are addressed by combining multiple sources and the usage of business rules. The presentation will give insight in the set-up of the datahub, the data quality challenges and the importance of PIDs, metadata standards and open, interoperable data sources. Besides these challenges, the presentation also addresses the power of the datahub by showing several concrete and implemented use cases that benefit not only the Dutch consortium and institutions but also researchers, funders and publishers. The presentation is scheduled on Sept. 17th 5:00pm CEST https://sched.co/1eNBL If you want to know more about the UKB datahub but can't attend the NISO conference, send me a PM.

31

How is mobile technology shaping the dynamics of family life? 📱✨ I’m thrilled to share the findings of my latest study, together with dr. Anneleen Meeus and Prof. dr. Kathleen Beullens, which dives deep into the interplay between mobile devices, parent–child relationships, and well-being. 📊👨‍👩‍👧 Using survey data from 390 parent–child dyads, we uncovered new insights about the role of mobile technology in fostering (or hindering) connection within families in today’s digital age 💡 Here’s what we uncovered: - Parents reported significantly higher levels of child technoference (i.e., interpersonal interactions being interfered with the use of technology) than children themselves reported.  - From children's perspective, both parent and child technoference was negative for the parent–child relationship, whereas family-oriented mobile technology use when physically together (e.g., gaming, watching a movie …) was beneficial for the parent-child relationship and the child's well-being. - From parents’ perspective, only child technoference was negatively related to the parent–child relationship and parent's well-being. The key takeaway? 📱 Mobile technology use can both strengthen and strain family relationships, depending on the family member and the type of technology use. 👉 Want to know more? Check out the full study (published in Mobile Media & Communication):

71

4 commentaren

dutch

1

Key question asked to the many Key Opinion Leaders visiting the Lab of the Future in Amsterdam today October 1st: What is the biggest challenge we face in making data AI ready? See the answer in the first comment. #loft2024 Lab of the Future #loftdebate.

33

1 commentaar

Eric's talk offers valuable insights for anyone interested in applying AI responsibly, especially when using citizen data. It’s a unique opportunity to learn about the complexities of working with real-world data and how ABMs can provide practical, actionable insights. Excited to attend!

6

Excited by this very well argued paper on technosolutionism! ⚡ In my world, we are currently working with the notion of 'quantum for good' - as ELSA scholars we need to scrutinise this closely, and intervene ex ante. For instance, what do we mean by 'quantum use cases' - let's think critically about what might be called a 'hammer and nail' approach. And what is the role of public policy in promoting technosolutionism - with quantum framed as a key solution in addressing technological sovereignty and strategic interests? Lots of juicy stuff to get stuck into! 🤓

26

2 commentaren

🚀 Tomorrow (Tuesday, May 14th) Luke Zettlemoyer, Professor at the University of Washington and Research Director at Meta, will be speaking in-person here in #Amsterdam. You are welcome to attend the talk! 📍 Amsterdam Science Park LAB42 UvA (Auditorium L1.02) ⏰ 11:00 AM 🔗 For online participation use: https://lnkd.in/ee7mBtQd Branch-Train-Merge: Embarrassingly Parallel Training of Expert Language Models Abstract: Existing language model (#LM) training regimes entangle compute, data, and parameters, requiring expensive synchronous communication with massive supercomputers. This talk introduces a new algorithm called Branch-Train-Merge (BTM) that asynchronously trains LMs that are fundamentally modular. In BTM, components (or experts) of the LM are specialized to distinct domains in the training corpus, and experts are conditionally updated based on the domain of the incoming document. We show how BTM enables LMs that are rapidly customizable (with the ability to mix, add, or remove experts after training), embarrassingly parallel (requiring no communication between experts), and sparse (needing only a few experts active at a time for inference). Key to our proposal is exploring what constitutes the domains to which experts specialize, as well as reflecting on the data sources used to train LMs. Our new techniques chart a path towards collaborative and iterative LM development, where anyone can contribute and maintain experts at modest computational cost.This talk describes work done at the University of Washington and Meta, primarily led by Suchin Gururangan and Margaret Li, along with many other collaborators. ------------ Bio: Luke Zettlemoyer is a Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington, and a Research Director at Meta. His research focuses on empirical methods for natural language semantics, and involves designing machine learning algorithms, introducing new tasks and datasets, and, most recently, studying how to best develop self supervision signals for pre-training. His honors include being named an ACL Fellow as well as winning a PECASE award, an Allen Distinguished Investigator award, and multiple best paper awards. Luke received his PhD from MIT and was a postdoc at the University of Edinburgh.

2

Are you thinking about doing more science communication, but not really sure where to start? Maybe you are already communicating, but would like to explore a wider range of channels? Or you just want an A-to-Z overview of everything #scicomm related? Jean Paul Bertemes , Serge Haan and Dirk Hans have written the handy, short book '50 Essentials on Science Communication' - and that title tells you everything you need to know. The best part? It's a free e-book! Link in the first comment. Thanks to Hans and Toon from The Floor is Yours for the tip (in their great newsletter - another insiders' tip)!

58

3 commentaren

Utrecht University and its spin-off, DialogueTrainer, have secured funding from Click.nl to elevate their virtual conversation trainer to the next level. By leveraging artificial intelligence, the platform aims to more closely mimic reality with avatars that respond to speech, intonation, emotion, facial expressions, and even body language. The research, conducted under the banner of the AI & Education Lab, will span four years. “In the AI & Education Lab, scientists collaborate with tech companies that develop tools and solutions to improve education,” says Sergey Sosnovsky, coordinator of the AI & Education Lab. “In this time of AI breakthroughs, it is particularly important for these companies to future-proof their products and services. Our collaboration and knowledge exchange will contribute to the design, development, evaluation, and deployment of next-generation AI-enabled dialogue technologies for the domain of education.” AI Labs (Utrecht University) Read more on this research: https://lnkd.in/eUPys3nK #uuscience #generativeAI #AI #conversationmodels

42

New in CODATA Data Science Journal: Elias Herman Kruithof, Christophe Vanroelen, Laura Van den Borre Title: Five Suggestions Towards User-Centred Data Repositories in the Social Sciences https://lnkd.in/eF3nMqpZ #codata #FAIRdata #OpenScience #datascience #opendata #datasciencej

1

To support providers of General Purpose AI models in complying with the EU AI Act, the European Commission's AI Office has established four working groups tasked with drafting a Code of Practice for General Purpose AI. On behalf of Conclusion and Conclusion AI360, Friso Spinhoven had the privilege of participating in two of these groups—one focused on identifying and assessing system risks and the other on AI governance for high-risk models. The working groups kicked off this week, and while the first draft (available here: https://lnkd.in/esWZA7Nd) looks promising, there’s still a significant amount of work to be done to provide truly practical guidance for providers. With around 300 participants per working group, the diverse perspectives make finding compromises a real balancing act. However, there is broad agreement that more detail is needed. The clock is ticking—less than six months remain to finalize the code, which is due by early May. We are excited to see how this evolves but it is important that we all keep investing in Responsible AI! WOMEN IN TECH ® Global Women in AI Governance™ (WiAIG) Women in Tech®Netherlands #W4EAI

8

2 commentaren

"AI is All About the Data!" AI applications are only as good as the data they rely on. Expecting Large Language Models (LLMs) to deliver accurate, complete, up-to-date and actionable information is a misconception. That’s why dm2find’s AI chatbot for Destination Marketing Organizations (DMO's) comes with access to a rich, curated, and ever-growing database of restaurants, museums, markets, and events. Want more control? You can even connect your own data! Why build when you can rent this solution as a SaaS and get started tomorrow? Let dm2find help you to boost tourism with reliable data at its core. Watch https://lnkd.in/e8EWDvMC Visit https://dm2find.ai/dmo #aitourism #chatbot #dmo #cmo #tourismmarketing

4

1 commentaar

Will we never learn? "By 2027, the global AI sector could annually consume as much electricity as the Netherlands, according to one recent estimate. And a new study in Nature Computational Science identifies another concern: AI’s outsize contribution to the world’s mounting heap of electronic waste. The study found that generative AI applications alone could add 1.2 million to five million metric tons of this hazardous trash to the planet by 2030, depending on how quickly the industry grows." Meanwhile ~every company is scrambling to find more nails to smash with their new AI hammers. Maybe we could lay off using enormously wasteful AI products to generate banal and meaningless social media posts, or generating the first two useless lines of dialog in a chat session before the user asks "can I just talk to a f***ing human?!"

2

Good news: the journal Zygon has just published a special issue that we (Rachel Schulder Abrams Pear, Hans Van Eyghen, Gijsbert van den Brink, and yours truly) edited on replication in the humanities, replication in historiography in particular and even more specifically replication in the history of science and religion. We replicated a chapter of John Hedley Brooke's famous book on science and religion. A few years ago, I published a series of articles with Lex Bouter arguing that replication is possible and can, under certain conditions, be valuable in the humanities as well. Some people agreed, others (fiercely) disagreed. So we decided to put the matter to the test by actually carrying out four replication studies. Two of them are presented in this special issue. I've immensely enjoyed the journey with this wonderful group of people (authors, original authors, advisory board members, etc.) and learned much from it. More convinced now than ever that under specific conditions replication can indeed be useful in the humanities. Yet, much is still unclear and replication definitely shouldn't become a conditio sine qua non for research in the humanities. I hope that others will, therefore, join us in this adventure and explore for themselves what can be learned by actually carrying out replication studies in the humanities. Here's the set-up (all articles fully open access): 1. Intro: https://lnkd.in/emjvyYmj 2. Direct replication, also considering Puritan sources: https://lnkd.in/e8ytxNHx 3. Conceptual replication, also considering Jewish sources: https://lnkd.in/eD8DREK6 4. Our reflections: https://lnkd.in/eaKM75w2 5. Response by the original author John Hedley Brooke: https://lnkd.in/etnf8VaT 6. And finally some fascinating reflections from Jeremy Brown: https://lnkd.in/eW_GPwhS Maartje Stols-Witlox (she/her) 🟥, Charlotte Rulkens, Pim Huijnen, Sarah de Rijcke, Bart Penders, J. Britt Holbrook.

42

4 commentaren