In this second part of a two-part episode on the dangers of AI and how to deal them in the context of large language models (LLMs), David is joined by our #ArmchairArchitects, Uli and Eric (https://www.youtube.com/channel/UCbQ5FSOvQ311EM3v_e7QwlQ) for a conversation that touches on how LLMs are different from traditional models, the challenge of trusting the model architecture, ethical considerations including bias, privacy, governance, and accountability, plus a look at some practical steps, such as reporting, analytics, and data visualization. Be sure to watch Armchair Architects: The Danger Zone (Part 1) https://aka.ms/azenable/146 before watching this episode of the #AzureEnablementShow. Resources • Microsoft Azure AI Fundamentals: Generative AI https://learn.microsoft.com/training/paths/introduction-generative-ai/ • Responsible and trusted AI https://learn.microsoft.com/azure/cloud-adoption-framework/innovate/best-practices/trusted-ai • Architectural approaches for AI and ML in multitenant solutions https://learn.microsoft.com/azure/architecture/guide/multitenant/approaches/ai-ml • Training: AI engineer https://learn.microsoft.com/credentials/certifications/roles/ai-engineer • Responsible use of AI with Azure AI services https://learn.microsoft.com/azure/ai-services/responsible-use-of-ai-overview • Blog: Considerations for Ethical and Responsible Use of AI in Applications https://techcommunity.microsoft.com/t5/azure-architecture-blog/armchair-architects-considerations-for-ethical-and-responsible/ba-p/4082133 Related Episodes • Armchair Architects: The Danger Zone (Part 1) https://aka.ms/azenable/146 • Watch more episodes in the Armchair Architects Series https://aka.ms/azenable/ArmchairArchitects • Watch more episodes in the Well-Architected Series https://aka.ms/azenable/yt/wa-playlist • Check out the Azure Enablement Show https://aka.ms/AzureEnablementShow Chapters 0:00 Introduction 0:18 Open questions from Part 1 2:15 Practical tips to get started 3:14 LLMs are not traditional models 3:33 Financial services approach 5:06 Commoditization of Generative AI 6:45 Feedback for Prompt tuning 9:10 Content safety 12:10 Teaser for next episodes