Mixture of Experts

In this discussion, AI technical solutions architect Vyoma Gajjar sheds light on the latest advancements from NVIDIA GTC, particularly the groundbreaking Groot N1 humanoid robotics model. Principal research scientist Kaoutar El Maghraoui explores Baidu's new AI reasoning models and their controversial closed-source nature. AI researcher Nathalie Baracaldo addresses the reliability of Chain-of-Thought reasoning in AI, highlighting potential biases. The conversation also touches on Google’s Gemini innovations in AI image generation, exploring competition in the evolving tech landscape.

Join Chris Hay, a distinguished engineer and CTO of Customer Transformation; Kaoutar El Maghraoui, principal research scientist at the AI Hardware Center; and Vyoma Gajjar, AI technical solutions architect, as they delve into the exciting world of AI. They discuss Manus AI's potential to disrupt the tech landscape and the peculiar rise of vibe coding. Also on the agenda are new insights into scaling laws that challenge traditional views and the innovative collaboration behind the upcoming AI phone from Perplexity and Deutsche Telekom.

Blake Johnson, a distinguished engineer and quantum engine lead, shares insights on when quantum computing might make its way into consumer devices. Chris Hay, CTO of Customer Transformation, discusses the Model Context Protocol, which streamlines developer workflows. Volkmar Uhlig, AI infrastructure expert, talks about CoreWeave's shift from crypto to AI cloud services and its competitive edge. They also dive into Sesame AI's innovative voice companion, revealing how these technologies are reshaping user interaction with AI.

In this insightful discussion, Kate Soule, a veteran in AI, and Chris Hay, an experienced AI analyst, dive deep into the unveiling of OpenAI's GPT-4.5. They explore whether pre-training is becoming obsolete, examining the shift toward inference-focused models. Insights on model selection and the balance of cost versus performance are highlighted. Additionally, they tackle the evolving dynamics of AI pricing and the impact of sophisticated tools on user experience. This conversation is a must-listen for anyone interested in the future of AI.

Granite 3.2 is officially here! In episode 44 of Mixture of Experts, host Tim Hwang is joined by Kate Soule, Maya Murad and Kaoutar El Maghraoui to debrief a few big AI announcements. Last week we covered small vision-language models (VLMs), and this week Granite 3.2 dropped with  new VLMs, enhanced reasoning capabilities, and more! Kate takes us under the hood to understand the new features and how they were created. Next, Anthropic dropped a new intelligence model, Claude 3.7 Sonnet, and a new agentic coding tool, Claude Code. Why did Anthropic release these separately? Then, as we cannot have an episode without covering agents, Maya takes us through the new BeeAI agents! Finally, can fine tuning on a malicious task lead to much broader misalignment? Our experts analyze a new paper released on ‘Emergent misalignment.’ All that and more on this week's episode! 00:01 – Intro  00:41 – Claude 3.7 Sonnet 11:58 – BeeAI agents  20:11– Granite 3.2 29:23 – Emergent misalignment The opinions expressed in this podcast are solely those of the participants and do not necessarily reflect the views of IBM or any other organization or entity. 

In this discussion, Kate Soule, an AI and vision language models expert from Granite, Volkmar Uhlig, who leads AI infrastructure at a major firm, and Shobhit Varshney, a senior consultant on AI in the Americas, dive into the burgeoning world of deep research in AI. They explore exciting innovations from OpenAI and Google, the prospects of an OpenAI inference chip, and the rise of smaller vision-language models. The conversation wraps up with an intriguing look at a startup's quest for AI agents, pondering their role in the job market of the future.

Anastasia Stasenko, CEO of pleias and open-source AI advocate, joins AI experts Marina Danilevsky and Chris Hay to dive deep into the Paris AI Summit highlights. They discuss the global advancements in AI safety and a remarkable €109 billion investment to boost European infrastructure. The trio explores the implications of an intriguing new test-time scaling technique for AI models, as well as insights from Sam Altman's latest observations. They also dissect Anthropic's Economic Index, revealing the actual scope of AI adoption across industries.

Joining the conversation are Marina Danilevsky, a senior research scientist focused on AI ethics; Chris Hay, a distinguished engineer and CTO specializing in AI product development; and Nathalie Baracaldo, an expert in AI safety and security. They explore OpenAI’s recent deep research and the o3-mini model, discussing its capabilities and limitations. Insights on the upcoming AI Action Summit highlight its significance in navigating AI governance and ethical implications. The trio also dissects Anthropic’s Constitutional Classifiers and Microsoft’s new unit studying AI's impact on society.

In this engaging discussion, Kate Soule, Director of Technical Product Management at Granite, Chris Hay, Distinguished Engineer and CTO of Customer Transformation, and Aaron Baughman, IBM Fellow and Master Inventor dive into the realities behind DeepSeek R1. They debunk myths surrounding its hype and discuss the true implications of model distillation for AI competition. The trio explores the evolving landscape of open-source AI and how recent advancements can reshape industry strategy, shedding light on efficiency and innovation in model training.

Join experts Abraham Daniels, a senior technical product manager specializing in AI and open-source models, Kaoutar El Maghraoui, a principal research scientist leading AI hardware innovations, and Skyler Speakman, a senior research scientist focusing on AI technology. They unravel the implications of DeepSeek's open-source model launch, Mistral's IPO plans, and the controversial FrontierMath benchmarks. They also discuss IDC's findings on coding assistants, highlighting the shift towards specialized versus generalist tools in the programming landscape.