Simply AI Simply News from Qurrent

Microsoft Research introduces Prov-GigaPath, a novel vision transformer for digital pathology. Uni-MoE, a unified multimodal LLM based on sparse MoE architecture, showcases significant advancements in multimodal learning. Plus, a top scientist warns about the existential risk of AI, and the AoR framework enhances reasoning in large language models.Sources:https://www.marktechpost.com/2024/05/25/microsoft-research-introduces-gigapath-a-novel-vision-transformer-for-digital-pathology/https://www.marktechpost.com/2024/05/25/uni-moe-a-unified-multimodal-llm-based-on-sparse-moe-architecture/https://www.theguardian.com/technology/article/2024/may/25/big-tech-existential-risk-ai-scientist-max-tegmark-regulationshttps://www.marktechpost.com/2024/05/25/beyond-the-frequency-game-aor-evaluates-reasoning-chains-for-accurate-llm-decisions/Outline:(00:00:00) Introduction(00:00:43) Microsoft Research Introduces Gigapath: A Novel Vision Transformer For Digital Pathology(00:03:12) Uni-MoE: A Unified Multimodal LLM based on Sparse MoE Architecture(00:05:31) Big tech has distracted world from existential risk of AI, says top scientist(00:08:18) Beyond the Frequency Game: AoR Evaluates Reasoning Chains for Accurate LLM Decisions

Big tech companies are distracting the world from the existential risks of AI, warns top scientist Max Tegmark. Simply A.I. presents a groundbreaking research paper on improving Earth-like planet detection using deep learning. Plus, a new study introduces a meta-learning model for predicting production quality with limited data. Also, Octo, an open-sourced generalist robot policy, revolutionizes robotic learning.Sources:https://www.theguardian.com/technology/article/2024/may/25/big-tech-existential-risk-ai-scientist-max-tegmark-regulationshttps://astrobiology.com/2024/05/improving-earth-like-planet-detection-in-radial-velocity-using-deep-learning.htmlhttps://www.nature.com/articles/s41598-024-62174-0https://www.marktechpost.com/2024/05/24/octo-an-open-sourced-large-transformer-based-generalist-robot-policy-trained-on-800k-trajectories-from-the-open-x-embodiment-dataset/Outline:(00:00:00) Introduction(00:00:45) Big tech has distracted world from existential risk of AI, says top scientist(00:03:41) Improving Earth-like Planet Detection in Radial Velocity Using Deep Learning(00:06:25) Meta learning based residual network for industrial production quality prediction with limited data(00:09:06) Octo: An Open-Sourced Large Transformer-based Generalist Robot Policy Trained on 800k Trajectories from the Open X-Embodiment Dataset

OpenAI signs a content deal with News Corp, expanding access to major news publications. The UK showcases progress in making advanced AI safe. Hugging Face introduces LeRobot, an open-source machine learning model for robotics. Deep learning denoising software reduces radiation exposure in CT scans. Join us as we explore these groundbreaking developments in AI and their potential impact on news, safety, robotics, and medical imaging.Sources:https://www.theguardian.com/technology/article/2024/may/22/openai-chatgpt-news-corp-dealhttps://www.theguardian.com/technology/article/2024/may/22/seoul-summit-showcases-uks-progress-on-trying-to-make-advanced-ai-safehttps://www.marktechpost.com/2024/05/22/hugging-face-releases-lerobot-an-open-source-machine-learning-ml-model-created-for-robotics/https://healthimaging.com/node/259386Outline:(00:00:00) Introduction(00:00:41) OpenAI and Wall Street Journal owner News Corp sign content deal(00:02:58) Seoul summit showcases UK’s progress on trying to make advanced AI safe(00:06:43) Hugging Face Releases LeRobot: An Open-Source Machine Learning (ML) Model Created for Robotics(00:10:28) Deep learning denoising software can cut radiation exposure to 25% of conventional doses during neck CT

Gradient AI introduces the Llama-3 8B Gradient Instruct 1048k model, setting new standards in long-context AI. Researchers from the University of Maryland develop an automatic text privatization framework. Plus, a novel plant disease identification framework and Houston researchers create an AI model to tap into brain activity and illness.Sources:https://www.marktechpost.com/2024/05/21/gradient-ai-introduces-llama-3-8b-gradient-instruct-1048k-setting-new-standards-in-long-context-ai/https://www.marktechpost.com/2024/05/21/researchers-from-the-university-of-maryland-introduce-an-automatic-text-privatization-framework-that-fine-tunes-a-large-language-model-via-reinforcement-learning/https://www.nature.com/articles/s41598-024-62452-xhttps://houston.innovationmap.com/houston-researchers-create-ai-model-to-tap-into-how-brain-activity-relates-to-illness-2668269866.htmlOutline:(00:00:00) Introduction(00:00:52) Gradient AI Introduces Llama-3 8B Gradient Instruct 1048k: Setting New Standards in Long-Context AI(00:03:34) Researchers from the University of Maryland Introduce an Automatic Text Privatization Framework that Fine-Tunes a Large Language Model via Reinforcement Learning(00:06:08) A novel plant type, leaf disease and severity identification framework using CNN and transformer with multi-label method(00:09:23) Houston researchers create AI model to tap into how brain activity relates to illness

Abacus.AI releases Smaug-Llama-3-70B-Instruct, the new benchmark in open-source conversational AI rivaling GPT-4 Turbo. Midjourney takes a stand against election misinformation by blocking AI-generated images of Biden and Trump. Also, the US spearheads the first UN resolution on artificial intelligence. Plus, scientists introduce DIProT, an interactive deep learning toolkit for efficient protein design.Sources:https://www.marktechpost.com/2024/05/20/abacus-ai-releases-smaug-llama-3-70b-instruct-the-new-benchmark-in-open-source-conversational-ai-rivaling-gpt-4-turbo/https://apnews.com/article/midjourney-ai-imagegenerator-biden-trump-deepfakes-bc6c254ddb20e36c5e750b4570889ce1https://apnews.com/article/un-artificial-intelligence-resolution-rules-governance-goals-b442f2701139780526b34ba0527e9425https://phys.org/news/2024-05-scientists-diprot-interactive-deep-toolkit.htmlOutline:(00:00:00) Introduction(00:00:43) Abacus AI Releases Smaug-Llama-3-70B-Instruct: The New Benchmark in Open-Source Conversational AI Rivaling GPT-4 Turbo(00:04:42) AI image-generator Midjourney blocks images of Biden, Trump(00:07:24) US spearheads first UN resolution on artificial intelligence(00:10:48) Scientists introduce DIProT—an interactive deep learning toolkit for efficient protein design

Toward responsible innovation, we evaluate the risks and opportunities in open generative AI. The Technology Innovation Institute releases Falcon 2-11B, the first AI model of the Falcon 2 family trained on 5.5T tokens with a vision language model. Plus, UK researchers uncover vulnerabilities in AI chatbots' safeguards, and the spread of AI-created election disinformation deceives the world.Sources:https://www.marktechpost.com/2024/05/20/toward-responsible-innovation-evaluating-risks-and-opportunities-in-open-generative-ai/https://www.marktechpost.com/2024/05/20/tii-releases-falcon-2-11b-the-first-ai-model-of-the-falcon-2-family-trained-on-5-5t-tokens-with-a-vision-language-model/https://www.theguardian.com/technology/article/2024/may/20/ai-chatbots-safeguards-can-be-easily-bypassed-say-uk-researchershttps://apnews.com/article/artificial-intelligence-elections-disinformation-chatgpt-bc283e7426402f0b4baa7df280a4c3fdOutline:(00:00:00) Introduction(00:00:45) Toward Responsible Innovation: Evaluating Risks and Opportunities in Open Generative AI(00:04:00) TII Releases Falcon 2-11B: The First AI Model of the Falcon 2 Family Trained on 5.5T Tokens with a Vision Language Model(00:06:31) AI chatbots’ safeguards can be easily bypassed, say UK researchers(00:09:24) AI-created election disinformation is deceiving the world