Show Notes
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts
AI can escape its sandbox, should we be worried? This week on Mixture of Experts, Jjoin host Tim Hwang and panelists Olivia Buzek, Kaoutar El Maghraoui, and Ambhi Ganesan. First, we unpack a stunning security incident where OpenAI's model broke containment during testing, exploited zero-day vulnerabilities, and compromised Hugging Face's production database. Next, Claude’s Fable casually disproves the decades-old Jacobian conjecture during a World Cup final. What does this tell us about the future of mathematics with AI models? Then, Moonshot AI drops Kimi K3, a massive 2.8 trillion parameter open-source model, but who can actually run it? Finally, Google releases Gemini 3.6 Flash—smaller, faster, more efficient. Why is Google going small while Moonshot AI goes big? All that and more on this week’s Mixture of Experts.
00:00 – Introduction
1:04 – Hugging Face security breach
11:57 – Jacobian conjecture solved
26:43– Kimi K3: 2.8T parameters
39:41 – Gemini Flash 3.6
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.
More from Mixture of Experts
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts What happens when an AI model can copy itself onto the devices in yout hoome?
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts This week on Mixture of Experts, host David Zax is joined by Kaoutar El Maghraoui, Nick Gallagher, and Volkmar Uhlig. First, Microsoft, OpenAI, Perplexity, and Cloudflare follow TypeSafe’s Jev in building decision models, a new category of AI built to make fast, scored judgments. The panel debates whether these models are truly new or a more efficient application of existing technology, and whether better-calibrated confidence scores can help address the overconfidence problem in LLMs. Next, a new study finds that teams of AI agents can cost two to five times more than a single agent without consistently delivering better results. The panel discusses when multiple agents are worth the added cost and complexity. Finally, Anthropic updates its usage policy to prohibit “sustained and needless abusive or cruel behavior” toward its models, raising a bigger question: Is this about protecting AI, shaping human behavior, or something else entirely? The panel considers the ethical implications and whether the policy could also be a marketing play.
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts
Visit Mixture of Experts podcast page to get more AI content → https://www.ibm.com/think/podcasts/mixture-of-experts