AI News Today

Today's episode discusses the launch of Mistral's advanced AI model with multilingual capabilities that challenge established models like GPT-4o. We also delve into Apple's development of two compact AI models, DCLM 1.4B and DCLM 7B, with performance efficiencies surpassing models from major players like Google and Meta. Finally, we'll explore Microsoft's Bing Generative Search, a response to Google's AI-powered search experience. Tune in for these exciting AI developments.

On this episode, we delve into Stability AI's latest generative model, Stable Video 4D, which allows for the generation of videos from 8 different perspectives. We also discuss Reworkd's new focus on web scraping AI technology, introducing a multimodal code generation approach that aids in data extraction from multiple smaller websites. Lastly, we unpack OpenAI's new fine-tuning feature for GPT-4o mini, providing enhanced performance for specific use cases while also offering significant cost savings and improved intelligence over its predecessor.

This episode discusses Meta's launch of Llama 3.1 405B language model with apparent restrictions, raising questions about its open-source status. We also delve into Adobe's latest update, bringing AI features, including a generative shape fill tool to Illustrator and an AI Image Generator to Photoshop, promising a creator-friendly approach. Lastly, we cover Elon Musk's xAI, training AI models in its new supercomputing center in Memphis with plans for Grok 3, anticipated to be the world's most potent AI model.

Today, we discuss Cohere's recent achievement of securing a massive $500 million investment for its privacy-centric enterprise AI technology. We also cover Figma's temporary disabling of its AI feature, First Draft, due to some system issues. Lastly, we'll talk about Perplexity's updates to improve user navigation and overall experience.

This episode delves into Tenstorrent's new Wormhole AI processors and workstation solutions, which offer cost-effective AI computing power. Also discussed is actress Scarlett Johansson turning down OpenAI chatbot voice gig over ethical concerns and Meta's decision to withhold its AI model, Llama, in EU amidst regulatory challenges and privacy concerns.

This episode covers major movements in the AI landscape, beginning with the acquisition of Commure by health tech player, Augmedix. We then shift focus to TSMC, which registers over half of its revenue from high-performance computing, thanks to a burgeoning demand for AI. Lastly, we discuss Google's ongoing negotiations with EssilorLuxottica for a potential partnership, hinting at the integration of Google's AI system in future smart glasses.

This episode covers OpenAI's launch of GPT-4o mini, its most cost-efficient model, offering advantages in language translation, cost-efficiency and security. We also discuss Google's partnership with NBC Universal as the official AI sponsor for Team USA at the 2024 Olympics, featuring Google's technology during live broadcasts. Finally, we delve into the launch of Mistral AI and NVIDIA's advanced language model, the Mistral NeMo 12B, set to revolutionize enterprise applications in terms of accuracy, efficiency and flexibility.

In this episode of AI News Today, we discuss Meta's decision to not offer future multimodal AI models in the EU due to regulatory concerns, though they plan to launch them in the UK. We also talk about Microsoft's release of an AI-powered Designer app for iOS and Android, which aligns with the company's goal of enhancing the creative sector with AI tools. Finally, we clarify rumors around Apple's OpenELM model, confirming that it does not power any of their AI or ML features and was never trained on YouTube subtitles, contrary to prior reports.

This episode looks at French AI startup, Mistral's new large language models - Codestral Mamba, ideal for coding projects and Mathstral 7B, specialized in math reasoning and scientific discovery. We also discuss Anthropic's release of its AI chatbot app, Claude, designed to compete with ChatGPT and accommodate a wider user-base. Lastly, we dive into the controversial topic of AI systems being trained using YouTube content, implicating tech giants like Apple and Nvidia without necessary permissions, thereby raising issues of transparency and potential terms violation.

In this episode, we delve into Meta's upcoming launch of Llama 3 405B, an advanced, open-source language model boasting 405 billion parameters with image and text processing capabilities. We also discuss the Hong Kong government's trial of a ChatGPT-style tool in public services, developed by the Hong Kong University of Science and Technology. Finally, we cover the ongoing investigation by the UK Competition and Markets Authority into Microsoft's $650 million investment in AI startup Inflection AI, amid concerns of potential undue influence within the tech sector by big firms.