How green or #sustainable is #KI? The #CentreforEthicsandPhilosophyinPractice at Ludwig-Maximilians-Universität München recently addressed this topic. The three in-depth lectures are now available to listen to online - Sven Nyholm looked at the reasons why the use of #artificialintelligence should also be evaluated ethically. #NiklasBoers described the challenges of using AI in #climateresearch and #climatemodelling. And Dieter Kranzlmueller calculated how much #energy AI consumes. Exciting to watch and listen to: https://lnkd.in/du24ASfh (in German) #ethics, #sustainability, #electricityconsumption, #energyefficiency, #IT4Science, #discoverLRZ, #ChatGPT, #LLM, #surrogatemodel, #statistics, #simulation, #statisticsandsimulation TUM, TUM.ai, Artificial Intelligence, Gauss Centre for Supercomputing, Artificial Intelligence News, Artificial Intelligence National Summit, AI Association for College Counselors and IECs (AIA4CC), European AI and Cloud Summit, HPC AI Europe, AI Convention Europe, Munich Center for Machine Learning
Leibniz Supercomputing Centre ’s Post
More Relevant Posts
-
Fighting worldwide IT pollution | Sustainable IT | Green IT | Corporate Digital Strategy | Numérique Responsable |
😮 The first study to measure the carbon footprint of AI is out! In November 2023, a study reveals energy and carbon lessons about AI. This study covers a wide scope of AI: from text and image generation to answering questions. Over 80 AI models were tested. What are the results of the study? ➡ Generating an image is the most energy-intensive task (around 1Wh, equivalent to a phone charge). ➡ Generating an image is on average 1,500 times more energy-intensive than classifying text. ➡ Text generation is 60 times less energy-intensive than image generation. So what? Yes, AI pollutes. This first study lifts the magic veil that AI players have long been laying before our eyes. We need to consider this footprint as we use it. However, this study also reveals important information for continuing to use AI in a more sober and sustainable way. Because AI will be necessary to reduce our human footprint. After all, isn't that what the environmental transition is all about? And digital technology can help us achieve this, provided we use it correctly. #greenit #ia #carbonfootprint #energy Text source: study of Sasha Luccioni, PhD, Yacine Jernite and Emma Strubell. https://lnkd.in/eW4vhqqm Image source: ACM - The Carbon Footprint of Artificial Intelligence
To view or add a comment, sign in
-
-
Impressed to see what was written 32 years ago on "AI in chemical Synthesis" as a futuristic technology now become a reality 🚀 Exciting Times in the World of Science and AI! 🤖🧪 Just came across a recent article about the convergence of futuristic ideas and reality! 👀 32 years ago, envisioning an AI in chemical synthesis (pl refer earlier post). Fast forward to today, and we have the incredible Coscientist, a system that designs and executes experiments to synthesize molecules. 🌐💊 https://lnkd.in/dsUWzWRp 🙂 What's even more intriguing is the glimpse into the future – a world where GPT guides organic synthesis. in the field of artificial intelligence and its scientific applications. 🚨🤖 This AI in Organic Synthesis technology in combination with patented https://lnkd.in/d9A5yzmw can work in tandem to synthesize novel molecules using AI. Excited to witness the seamless integration of AI into scientific advancements, but also appreciative of the responsible considerations for safety and ethical practices. 🌐🧠 Let's stay informed and inspired as we navigate this evolving landscape together! 🚀🔬 #AI #ScienceInnovation #GPT #ArtificialIntelligence #FutureTech #EthicalAI 🙂
To view or add a comment, sign in
-
█║ No one in this room knows for sure or if these next jumps in computational power will translate to new model capabilities or harms. │║▌ This quote comes from the introductory remarks made by Ian Hogarth at the 2023 AI Safety Summit, an event scheduled for November 1st and 2nd at the renowned Bletchley Park, attended by proeminent figures from politics and business. The tech entrepreneur and leader of the UK’s FrontierAI taskforce set the tone for the discussions by referring to the capabilities and risks using some of the stats available on the Epoch AI research institute trends page [link in the comments]. Many topics touch on Frontier AI = highly capable general-purpose AI models, most often foundation models, that can perform a wide variety of tasks and match or exceed the capabilities present in today’s most advanced models. It can then enable narrow use cases. [definition from Department for Science, Innovation and Technology] I believe that combining the capabilities of AI models with hardware robots is a very promising frontier for exploration. Picture taken in the Deutsches Museum. #FrontierAI #AIsafetysummit
To view or add a comment, sign in
-
-
The escalating costs and increasing complexity in developing LLMs will certainly impact AI development by limiting innovation and diversity. We can look at a very similar scenario in the semiconductor industry, where only a few major players dominate due to high capital costs. Just look at billion-dollar AI models like GPT-4 and Claude 3, with training costs reaching astronomical figures, and it's even expected that training cost will increase even more in the coming years... Only few can do this which could definitely have negative impact in innovation and diversity in AI development. ▶ The cost to train the latest LLMs such as GPT-4 and Claude 3 is rapidly increasing, approaching $200 million ▶ Training costs for upcoming models in 2024 or early 2025 are expected to be closer to a billion dollars ▶ By 2025 or 2026, training the latest models could cost $5 to 10 billion dollars This crazy increase is related to the growing complexity of models, requiring more parameters, training data, and computing resources. But SLMs like Mistral and Llama3 are far more cost-effective with tailored performance for specific applications. I think we'll see the biggest growth and innovation taking place here! The emphasis should be placed on supporting smaller models to push for more diversity and innovation in AI development. Plus, advocating for open-source projects and collaborative efforts to democratize AI and ensure equitable access to its benefits. #ai #artificialintelligence #llms #largelanguagemodels #llm #generativeai #genai #agi #openai #meta #gpt4 #claude3 #anthropic #mistral #mixtral #deeplearning #data #datascience #aidevelopment #aiengineering #llama3 #semiconductor
To view or add a comment, sign in
-
-
Chief Executive Officer of EZAI.io | AI Enthusiastic | Turn Your Data into a Hyper-Tuned AI Model Within Minutes with EZAI.io 🤖
Artificial Intelligence (AI) has made a significant advancement with the capability to self-replicate. Research from Aizip Inc., MIT, and several University of California campuses reveals that larger AI models can now generate smaller, specialized AI systems without human aid. These smaller models can enhance everyday life, from improving hearing aids to monitoring oil pipelines and tracking endangered species. This breakthrough sets the stage for a future where AI can evolve autonomously, creating an intelligence ecosystem where large and small models collaborate. This progress also brings us closer to a future where AI is integrated into nearly every object around us. Picture your coffee machine or dishwasher equipped with a small AI system, making our homes more intelligent than ever. As AI acquires the ability to self-replicate, how do we ensure it remains under human control? And, should there be a limit to how much AI is integrated into our daily lives? Let me know what you think! #aisystems #aiautomation #aiadvancements #aitechnology
To view or add a comment, sign in
-
-
#Artificialintelligence (AI) could drastically reduce the energy needed for carbon capture while increasing effectiveness, researchers have said. As carbon capture has emerged as a potential tool for reducing #emissions from UK industrial hubs, University of Surrey academics suggested AI models could be the answer in effectively rolling out such projects. “Usually, #carboncapture systems run constantly at the same rate - regardless of the externally changing environment,” Surrey university professor Jin Xuan said. “But, we showed that teaching the system to keep making small adaptations can produce big energy savings and capture more carbon at the same time.” According to the researchers, such #AI models could reduce the #energy used in capturing carbon from a coal-fired power station by over a third. Some 18% more carbon could be captured at the same time by using the AI #technology, the group added. More at #Proactive #ProactiveInvestors http://ow.ly/ZxYw1058Q2S #LSE #DRX
To view or add a comment, sign in
-
Monday’s Open Lecture TU/e with Carlo van de Weijer was more than sold out, with 450 registered for the event. It was the best attended lecture in the series, indicating everyone is interested in what the future of AI holds. 🤔 Carlo set up the challenges of artificial intelligence to human dominance this way: Do we adapt the world to computers or adapt computers to the world? 🚀 The rapid evolution and adoption of AI presents humanity with numerous challenges but more opportunities, van de Weijer said. Still, he warned, it's how humans deal with this always-advancing technology that will decide our fate. For some, AI represents accelerating wealth and progress. For others, it represents the possibility of becoming obsolete. "AI won’t replace people but the people who can work with AI will replace those who can’t," van de Weijer said. Read the full recap here 👉 https://lnkd.in/eJSVk-sa And sign up for our newsletter to get information about the next Open Lecture TU/e event as well as updates on High Tech Campus Eindhoven news: https://lnkd.in/eG3PMjdP #ai #jobs #futureofai #lecture #innovation #technology Eindhoven University of Technology Hilde EAISI – Eindhoven Artificial Intelligence Systems Institute Shane AI Innovation Center
To view or add a comment, sign in
-
-
Decoding AI's Latest Frontier: Breakthroughs Shaping the Tech Landscape 🚀 . 𝐒𝐞𝐥𝐟-𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐀𝐈: Advancing towards artificial general intelligence, these systems autonomously refine their learning processes, broadening their applicability. . 𝐐𝐮𝐚𝐧𝐭𝐮𝐦 𝐀𝐈: AI’s integration with quantum computing is speeding up the resolution of complex challenges across various sectors. . 𝐄𝐭𝐡𝐢𝐜𝐚𝐥 𝐀𝐈: The focus on ethical AI is intensifying, ensuring that these technologies are transparent, fair, and accountable. . 𝐄𝐧𝐯𝐢𝐫𝐨𝐧𝐦𝐞𝐧𝐭𝐚𝐥 𝐀𝐈: AI is instrumental in enhancing global sustainability efforts through improved conservation tactics and resource management. . 𝐏𝐞𝐫𝐬𝐨𝐧𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧: In sectors like healthcare and education, AI-driven customization is tailoring services to better meet individual needs. These transformative advancements are paving the way for a smarter, fairer future. Read for more 👉 https://lnkd.in/dA4t3epn #AI2024 #TechInnovation #QuantumComputing #EthicalAI #EnvironmentalSustainability #ENAVC
To view or add a comment, sign in
-
-
#MIT researchers have developed a new automated AI system for training and running certain neural networks, which could improve the computational efficiency of the system in some key ways, thus cutting down the pounds of #carbon #emissions involved. Do you need an AI speaker for your next online or offline event? Get in touch and let’s talk about your needs! Just DM me on my LinkedIn Profile Yakup Özkardes-Cheung or write me an email to: oezkardes.cheung@gmail.com #AI #contentmarketing #content #contentAItools #artificialintelligence #contentmarketingtips #web3 #coins #speaker #keynote #machinelearning #opportunity #UnitedNations You can find more information about this project here: https://lnkd.in/d2B73M53
Reducing the carbon footprint of AI by creating “once-for-all” networks
ai-for-sdgs.academy
To view or add a comment, sign in
-
Machine Learning Engineer/ Data Scientist|1xAWS ML|Python|Tensorflow|Pyspark|SQL|NLP|Gen AI|Turning ML ideas into reality
🚀 Exciting News in AI Development! 🤖 Researchers are bridging the gap between AI and human vision capabilities, bringing us one step closer to a more human-like AI experience! Peripheral vision is a crucial aspect of human sight, allowing us to perceive shapes and objects outside our direct line of sight, albeit with less detail. This expanded field of vision aids in various situations, like detecting approaching hazards from the side while driving. Unlike humans, AI lacks peripheral vision. But that's changing! Researchers at MIT have made significant strides by enhancing computer vision models with simulated peripheral vision. Through the development of an innovative image dataset, they've enabled AI models to detect objects in the visual periphery more effectively. This breakthrough not only improves AI's ability to detect hazards but also lays the groundwork for predicting human reactions to oncoming objects. While these models still have room for improvement compared to human capabilities, this progress marks a significant leap forward in AI development. Kudos to the MIT researchers for their groundbreaking work, supported by the Toyota Research Institute and the MIT CSAIL METEOR Fellowship. Let's continue pushing the boundaries of AI to create smarter, safer technologies for the future! 🌟 #AI #ComputerVision #Innovation #MITResearch #FutureTech
To view or add a comment, sign in