Is OpenAI going to enter the social media game? In episode 52 of Mixture of Experts host, Tim Hwang is joined by Gabe Goodhart, Kate Soule and Marina Danilevsky. First, Sam Altman is rumored to be testing an internal prototype social network; why is this a potential next move for the AI giant? Next, for our paper of the week, we analyze Anthropic’s study on chain-of-thought reasoning, “Reasoning Models Don’t Always Say What They Think.” Then, AI scraping puts a strain on Wikimedia; what’s the impact of this? Finally, China held a humanoid robot half-marathon, where humans raced alongside robot competitors. Who wins this AI race? All that and more on today’s Mixture of Experts. 

00:41 -- OpenAI social network 

10:02 -- Anthropic’s reasoning study 

20:56 -- AI bots strain Wikimedia 

31:33 -- Humanoid half-marathon 

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. 

OpenAI just dropped o3 and o4-mini! In episode 51 of Mixture of Experts host, Tim Hwang is joined by Chris Hay, Vyoma Gajjar and special guest John Willis, Owner of Botchagalupe Technologies. Today, we analyze Sam Altman’s new AI models, o3 and o4-mini. Next, Google announced that by Q3 you can run Gemini on-prem; what does this mean for enterprise AI adoption? Then, John is on the show today to take us through AI evaluation tools and why we need them. Finally, NVIDIA is planning to move AI chip manufacturing to the U.S. Can they pull this off? All that and more on today’s Mixture of Experts. 

00:01 – Intro 

00:56 – OpenAI o3 and o4 mini 

14:57 – Google Gemini on-prem 

23:43 – AI evaluation tools 

34:59 – NVIDIA's U.S. chip manufacturing 

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. 

IBM z17 is here! In episode 50 of Mixture of Experts, host Tim Hwang is joined by Kate Soule, Shobhit Varshney and Hillery Hunter to debrief the launch of a new mainframe with robust AI infrastructure. Next, Meta dropped Llama 4 over the weekend;, how's it going? Then, Shobhit is recording live from Google Cloud Next in Las Vegas, along with Gemini 2.5 Pro. What are some of the most exciting announcements? Finally, the Pew Research Center shows perception of AI, how does this impact the industry? All that and more on today’s 50th Mixture of Experts. 

00:01 -- Intro 

00:55 -- IBM z17 

11:42 -- Llama 4 

25:02 -- Google Cloud Next 2025 

34:29 -- Pew's research on perception of AI 

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. 

Explore the new features of IBM z17: https://www.ibm.com/products/z17

Read the Pew Research: https://www.pewresearch.org/internet/2025/04/03/how-the-us-public-and-ai-experts-view-artificial-intelligence/  

Subscribe for AI updates: https://ibm.biz/Think_newsletter

Visit Mixture of Experts podcast page to learn more AI content: https://www.ibm.com/think/podcasts/mixture-of-experts 

Will OpenAI be fully open source by 2027? In episode 49 of Mixture of Experts, host Tim Hwang is joined by Aaron Baughman, Ash Minhas and Chris Hay to analyze Sam Altman’s latest move towards open source. Next, we explore Anthropic's mechanistic interpretability results and the progress the AI research community is making. Then, can Apple catch up? We analyze the latest critiques on Apple Intelligence. Finally, Amazon enters the chat with AI agents. How does this elevate the competition? All that and more on today’s Mixture of Experts.

00:01 -- Introduction

00:48 -- OpenAI goes open  

11:36 -- Anthropic interpretability results 

24:55 -- Daring Fireball on Apple Intelligence 

34:22 -- Amazon’s AI agents

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.

Subscribe for AI updates: https://www.ibm.com/account/reg/us-en/signup?formid=news-urx-52120

Learn more about artificial intelligence → https://www.ibm.com/think/artificial-intelligence

Visit Mixture of Experts podcast page to learn more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts

What’s the best open-source model? In episode 48 of Mixture of Experts, host Tim Hwang is joined by Kate Soule, Kush Varshney and Skyler Speakman to explore the future of open-source AI models. First, we chat about the release of DeepSeek-V3-0324. Then, more announcements coming out of Google including Gemini Canvas and Gemini 2.5. Next, Extropic has entered the chat with a thermodynamic chip. Finally, AI image generation is on the rise as OpenAI released GPT-4o image generation. All that, and more on today’s Mixture of Experts.

00:01 – Intro

00:42– DeepSeek-V3-0324

09:48 – Gemini 2.5 and Canvas

21:27– Extropic’s thermodynamic chip

30:20 – OpenAI image generation

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.

What’s the most exciting announcement coming out of NVIDIA GTC? In episode 47 of Mixture of Experts, host Tim Hwang is joined by Nathalie Baracaldo, Kaoutar El Maghraoui and Vyoma Gajjar. First, we dive into the latest announcements from NVIDIA GTC, including the Groot N1 model for humanoid robotics. Next, Baidu released some new AI reasoning models, and they’re not open source? Then, for our paper of the week we discuss the flaws of Chain-of-Thought reasoning. Finally, Gemini Flash 2.0 has released image generation models for developer experimentation., Iis Google catching up on the AI game? Tune -in to today’s Mixture of Experts to find out! 

00:01 – Intro  

01:27– NVIDIA GTC 

14:18– New Baidu AI models 

21:19– Chain-of-Thought reasoning 

32:18 – Gemini image generation 

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. 

Is Manus a second DeepSeek moment? In episode 46 of Mixture of Experts, host Tim Hwang is joined by Chris Hay, Kaoutar El Maghraoui and Vyoma Gajjar to talk Manus! Next, the rise of vibe coding—what started as a joke has now become a thing? Then, we dive deep into the future of scaling laws. Finally, Perplexity is teaming up with Deutsche Telekom to release an AI phone—what’s the motivation here? Tune-in to today’s Mixture of Experts to find out more! 

00:01 – Intro 

00:37 -- Manus 

14:09 – Vibe coding 

30:13 – Scaling laws 

39:07 – Perplexity's AI phone  

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. 

When can we expect quantum to reach consumer devices? In episode 45 of Mixture of Experts, host Tim Hwang is joined by special guest, Blake Johnson, to debrief the quantum noise in the news. Blake helps us understand the intersection between quantum and AI and how far we are from this technology. Then, veteran experts Chris Hay and Volkmar Uhlig hash out some other news in AI this week. We cover Anthropic’s Model Context Protocol, CoreWeave filing for an IPO and Sesame AI’s new voice companion. All that and more on today’s Mixture of Experts! 

00:01 – Intro  

01:06 – Quantum leap 

20:08 -- Model Context Protocol 

28:24 -- CoreWeave IPO 

40:12 -- Sesame AI voice companion 

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. 

Is pre-training dead? In this bonus episode of Mixture of Experts, guest host Bryan Casey is joined by Kate Soule and Chris Hay. On Thursday, Sam Altman dropped GPT-4.5 just after we wrapped our weekly recording. We got a few of our veteran experts on the podcast to analyze OpenAI’s largest and “best” chat model yet. What’s the hype? Tune-in to this bonus episode to find out! 

00:01 – Intro  

00:25 – GPT-4.5 

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. 

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.