Educait

Could this be the ImageNet moment for scientific AI? Today, as part of a large international collaboration, we're releasing two massive datasets that span dozens of fields - from bacterial growth to supernova! We want this to enable multi-disciplinary foundation model research. You might ask: why would such diverse training data help AI? Well, as we've seen over the past few years, breadth of training can lead to stronger performance. We want AI to exploit common phenomena across sciences – take for example, waves, which appear in many systems! The first dataset, "The Well", contains curated simulations from 16 scientific domains, each capturing fundamental equations that appear throughout nature: - Fluid dynamics & turbulence - Supernova explosions - Biological pattern formation - Acoustic wave propagation - Magnetohydrodynamics All validated and/or generated by domain experts. The second dataset, "Multimodal Universe", brings together astronomical data from major observatories and surveys: - Hundreds of millions of observations - Multiple modalities, object types, wavelengths - Data from JWST, HST, Gaia, and several other major surveys All unified in a single, ML-ready format. For ML researchers, this is: - 115TB of validated, well-understood scientific data - Clean train/test splits - Benchmark tasks ready - Available now on HuggingFace! We are thankful to have both of these papers accepted to NeurIPS 2024, in the datasets & benchmarks track. Our code is open-sourced here, with easy-to-use APIs: - https://lnkd.in/eCz8BmqN - https://lnkd.in/e6Vv82P7 I'm excited about this both for ML research, and for the problems it will enable us to solve in science! This was a collaboration with a 'dream team' of researchers around the world, including: - Multimodal Universe: Eirini Angeloudi, Jeroen Audenaert, Micah Bowles, Ben Boyd, David Chemaly, Brian Cherinka, Ioana Ciucă, Aaron Do, Matt Grayling, Erin Hayes, Tom Hehir, Shirley Ho, Marc Huertas-Company, Kartheik Iyer, Maja Jabłońska, Francois Lanusse, Henry Leung, Kaisey Mandel, Juan Rafael Martínez Galarza, Peter Melchior, Lucas Meyer, Liam Parker, Helen Qu, Jeff Shen, Mike Smith, Connor Stone, Mike Walmsley, John F. Wu, PhD - The Well: Ruben Ohana, Michael McCabe, Lucas Meyer, Rudy Morel, Fruzsina Agocs, Miguel Beneitez, Marsha Berger, Blakesley Burkhart, Stuart Dalziel, Drummond Fielding, Daniel Fortunato, Jared Goldberg, Keiya Hirashima, Yan-Fei Jiang, Rich Kerswell, Suryanarayana Maddu, Jonah Miller, Payel Mukhopadhyay, Stefan Nixon, Jeff Shen, Romain Watteux, Bruno Régaldo-Saint Blancard, François Rozet, Liam Parker, Shirley Ho. I'm very proud of the team and we are all excited to see what the community can do with these amazing datasets!

I was recently awarded the privileged of participating in the very first African Girls Can Code cohort in Zimbabwe, which ran from the 11th to the 15th of November. This empowering event was made possible by UN Women , Elevate Trust Zw and OmniContact International . It was an engaging and eye opening experience for girls of different ages and backgrounds, and I am beyond grateful that I was granted such an amaizing opportunity. On the first day, one of your lovely mentors, Henrica Kadira , kicked off the event by telling us a bit about her background and how she got to where she is today. She gave us pointers on how to improve and what steps to take to put ourselves on the global market in order to gain more opportunities within the tech industry. Shortly after that, we were taken through Design Thinkining by Jabulani Simplisio Chibaya , who was also a mentor at the bootcamp. He took us through the steps of design thinking : Define, Ideate, Prototype and Test ,as well as taught us the importance that documentation of projects holds. Soon after Jabu, we had Juliet Masvaure come in and educate us on digital literacy, why it is important, as well as the skills that are required to be considered a digitally literate individual. Our fourth mentor for day one was Kudzai Nyakwima , who took us through Digital Marketing. He explained how digital marketing was a game changer and why we should implement it in our skills. He also took us through the various types of digital marketing before handing us back to Jabu for our final lesson of the day, which was tech tools. He described them as a developer's skeleton. He took us through various tools including IDEs, Text Editors, Debugging tools and more. Day two drove in with a new strategy at hand. We were finally going to be hands on. Day two was all about Data. Data collection, data cleaning as well as the importance of Data. Juliet took us through data cleaning using Python's Pandas library before we were graced by Dr Panashe from Educait who took us through a basic introduction of Artificial Intelligence. Day three was dedicated to Front End development. Kudzi took us through front end design using HTML and CSS, then Michael Nyamande jumped in and took us through JavaScript. By the end of the day, we were able build a very basic Web page. Day four threw us into Back End development. Mike took us through how AI has impacted backend development and showed us various tools that developers were currently using to develop different projects. As the day drew to a close, we were taken through some cyber security basics as well as how to code for a safer world by members from CyberAlliance

We’re excited to be recognised as runners up for Zimbabwe’s Best AI Initiative award! 🏆🏆 Our very own Dr. Panashe Chiurunge was honoured at this year’s ICT Excellence Awards. Thank you Ministry of ICT Postal and Courier Services Zimbabwe for nomination. The Educait team is excited to continue our mission of democratising AI in education across Africa. #Educait #AIinEducation

Countries that are first to capitalize on periods of rapid technological change often reshape the global balance of power. BCG's new AI Maturity Matrix assessed 73 economies on GenAI and broader AI. Five—the US, China, the UK, Singapore, and Canada—are currently categorized as AI pioneers. Comparatively, they are well-positioned on investment, infrastructure, innovation, talent, AI regulation and ethics. Evolution is rapid, with contenders like India, France, Saudi Arabia, the UAE, and others narrowing the gap in areas. BCG's report includes high-level recommendations for countries to drive AI adoption with ambition, skills, policy, investment, research, and the right ecosystem. For enterprises, leveraging their global context in a world of comparative advantages and interconnected supply chains is key to gaining an edge. P.S. Link to AI Maturity Matrix and report in comments. Credits to BCG. --- ♻️ Repost to help your network level up! 📌 Want more insights like this? Get on the list: https://lnkd.in/eJ6QQeru

The overlooked GenAI use case: cleaning, processing, and analyzing data. https://lnkd.in/gcVN_psf Job post data tell us what companies plan to do with GenAI. The most common use case is data analytics projects. Examples: - AstraZeneca: using LLMs on freeform documents to structure results from their Extractables & Leachables testing (https://lnkd.in/gGA_9mjC) - Trafigura: The Document AI team is using LLMs to extract data from a corpus of commodity trading documents to generate credit reports (https://lnkd.in/gRvntqHi) The startup ecosystem is overlooking this use case, instead focusing on other areas such as customer support, sales & marketing and code gen.

Because generative AI is changing so quickly, I've created a research assistant agent that reads through preprints each morning and identifies the latest research on the impact of generative AI on teaching and learning (and, occasionally, instructional design). I've created a Substack where I'm sharing the papers I think I are worth knowing about (it's pretty low volume as my threshold is pretty high). Each post summarizes an article in a single sentence, includes relevant quotes, and provides the reference. I'm sharing here in case you might find it useful and want to follow along. (You can get these as a free email newsletter by subscribing at the bottom of any post.) As an example, check out this RCT comparing in-class active learning to using LLMs to study: "We find that students learn more than twice as much in less time when using an AI tutor, compared with the active learning class."

Day 2 of Programming, Innovating, and Mastering Data Mastering Prompt Engineering with Henrica Kadira: We learned the essentials of prompt engineering, focusing on data extraction, preparation, exploration, and visualization. We also enhanced our skills in presenting insights with impactful tools like pie charts, line graphs, and bar charts. Insights from Dr Panashe Chiurunge: We had the honor of hosting Dr. Panashe, who shared insights on Educait, an innovative learning platform, inspiring us with his extensive knowledge in #AI and #financialengineering. Data Dive with Juliet Masvaure: Juliet took us through data cleaning using Data Colab, where we learned to organize and refine messy datasets—a challenging yet rewarding experience! Exploring Data Tools with Jabulani Simplisio Chibaya: He introduced us to various data tools and Python Software Foundation libraries, showcasing effective techniques for data organization and analysis. Michael Nyamande 's Magic: Today was a highlight with Michael as we explored AI using a Teachable Machine, trained models, and even experimented with #Scratch coding. He wrapped up with music generated by AI, leaving us eager to explore more! #UNWomen #ElevateTrust #OmniContact #Stemfem #AGCC #Day2 #AI #MachineLearning #DataScience #DataAnalysis #TechEducation #WomenInTech #STEM #Python #DataVisualization #Coding #TechTrends #FutureOfTech #Programming #Innovation

New research from Harvard shows that AI tutors double the rate of student learning, reduce time to competency & increase student engagement & motivation compared with in-person classes. The study (currently still under peer review) compared the learning gains, time to competency & levels of engagement and motivation of two groups of students: one group led by an AI tutor, and the other group led by a human professor who delivered in-person active learning classes. The Results: ➔ Those taught by the AI tutor showed on average 2x learning gain Vs those taught in the classroom. ➔ Those taught by the AI tutor took less time to reach the intended outcome (49 mins vs 60 mins). ➔ Those taught by the AI tutor reported higher rates of engagement (4.1/5 vs 3.6/5) and increased motivation (3.4/5 vs 3.1/5). In this week's blog post, I summarise the research & ask: ➔ What does this data mean for how we design & deliver education in the immediate future? ➔ Can these results scale beyond high-achieving Harvard students? ➔ Are we thinking too small? Do we focus too much on optimising traditional tutoring models when AI could enable entirely new, more effective & inclusive learning approaches? Read my full analysis in my latest post (link in comments). Thanks Gianluca Mauro for sharing the research & sparking these thoughts! Happy innovating! Phil 👋 #AIinEducation #EdTech #FutureofLearning #EducationInnovation #TeachingandLearning

📰 Thrilled to be featured in today's Irish Mail on Sunday discussing AI in education! 🏫🤖 My key points : • 🎒 Students are already using AI extensively - teachers need to catch up! 🏃♂️💨 • 📊 AI can assist with admin tasks and personalize learning experiences • 📝 Clear guidelines for AI use in Irish schools are urgently needed • 👩🏫👨🏫 All teachers should be trained on how AI works 🚨 The ethical use of AI by students is a huge issue we must address. Teachers need guidance to use AI effectively and teach students responsible usage. 🗓️ This Wednesday at 8 PM, I'm diving deeper into AI tools for school leaders in my upcoming course. Learn how to leverage AI to reclaim your time and improve work-life balance. Can't make it? Don't worry - a recording will be available! Register here: https://lnkd.in/eXFf5qhQ Human Intelligence Movement #AIinEducation #EdTech #SchoolLeadership Adam Pogrund

❗ California Assembly Bill 2876 (AB 2876) aims to integrate artificial intelligence (AI) literacy into the K-12 curriculum across the state. The bill mandates that the Instructional Quality Commission incorporate AI literacy into the math, science, and history-social science frameworks when these curricula are revised after January 2025. The bill defines AI literacy as understanding how AI works, including its principles, concepts, and applications, along with recognizing its limitations, implications, and ethical considerations. This effort is intended to prepare students for an AI-driven future, equipping them with the skills needed to navigate and responsibly use AI technology. AB 2876 was signed into law by Governor Newsom on September 29, 2024, as part of California's broader initiative to address the growing role of AI in society and ensure that students are prepared for the evolving job market and the ethical challenges posed by AI. Very exciting and important news. We are continuing to learn that AI literacy is not providing schools access to an app or giving them a laundry list of applications that write lesson plans or create worksheets. AI literacy is about teaching teachers, leaders, and students the skills they need to be able to use AI as a teammate by practicing integrity, ethics, critical thinking, empathy, trust and effective communication. At Designing Schools we aim to bridge the gap between education and enterprise in our work, in accordance with what this bill is aimed to do, with our California partners now and in the future. Irvine Unified School District Santa Ana Unified School District Butte County Office of Education Solano County Office of Education Colton Joint Unified School District San Bernardino County Orange County Department of Education Desert Sands Unified School District Ocean View School District Dr. Sabba Quidwai