Join host Will Jarvis for a solo presentation on the rapidly evolving world of large language models and what they mean for property tax assessors. Will covers the connection between AI and property tax through Sam Altman's Moore's Law for Everything thesis, explains how transformer-based models work and why prompting skill matters, walks through practical use cases like drafting correspondence, summarizing appeals, and enriching parcel data, and demos two free ValueBase tools—ValPal and Ratio Pal—that assessors can start using immediately.

• 00:00 - Introduction and background on ValueBase, Sam Altman's investment, and the macro connection between AI and property tax
• 03:00 - The astonishing pace of AI improvement: frontier math breakthroughs and what rapid model advancement means for society
• 06:30 - Why humans still matter: agency, real-world interaction, and the enduring role of knowledge workers alongside AI
• 08:00 - The wolves-to-dogs analogy and the debate over pausing AI development
• 10:00 - Economic context: slowing productivity growth, rising costs of essential goods, and why we need more technological innovation
• 14:30 - How large language models work: the transformer architecture, Attention Is All You Need, and text prediction fundamentals
• 16:30 - The context problem: why LLMs are like the smartest intern who knows nothing about your office
• 18:30 - Practical use cases for assessors: drafting correspondence, summarization, organizing tasks, and data enrichment
• 25:00 - The critical importance of verifying model outputs and avoiding "AI slop grenades"
• 29:00 - Why good grammar, encouragement, and clear prompting lead to better model performance
• 33:30 - Anatomy of a useful prompt: task, context, output format, checks, and asking for clarification
• 37:00 - Demo of Ratio Pal: uploading sales data, running ratio study statistics, and visualizing vertical equity
• 41:00 - Demo of ValPal: assessor-specific LLM with state property tax codes, IAAO standards, and code execution
• 44:00 - Q&A on learning to use AI tools, recommended resources like Andrej Karpathy's guides, and closing remarks