Will Jarvis - ValPal at IAAO

The assessment profession stands at a crossroads. While we debate ratio studies and struggle with storage rental rates, Silicon Valley is pouring $100 billion into AI tools that could fundamentally reshape how we work. The question isn't whether to adopt these technologies, it's whether we'll lead the charge or get left behind.

The Smart Intern You Never Have to Train

Think of large language models (LLMs) like ChatGPT as the world's smartest intern who never needs coffee breaks. They've read every property tax code, memorized every USPAP standard, and can recall North Carolina General Statute 105-286 faster than you can open your filing cabinet. But here's the catch: just like that eager intern, they need clear direction and careful oversight.

Will Jarvis, CEO of ValueBase and former assessment researcher, puts it perfectly: these tools are "very sensitive to how you talk to them." Feed them vague instructions, get vague results. Give them context and examples? Now you're cooking with gas.

The beauty lies in their versatility. Need to know Oklahoma's storage rental rates? The AI scours the internet and delivers a report in 30 seconds, complete with visualizations and source citations. Taxpayer shows up with a suspect appraisal? Upload it and watch the AI flag USPAP violations faster than you can say "insufficient adjustment support."

The Hallucination Problem: Trust but Verify

Here's where seasoned assessors earn their keep. LLMs occasionally "hallucinate", they generate plausible-sounding information that's completely fabricated. Remember that attorney who cited non-existent case law? That's your cautionary tale.

Smart implementation means building guardrails. When ValueBase developed ValPal for assessment offices, they bound it to actual property tax codes and assessment manuals. Ask about North Carolina's reappraisal cycle, and it pulls directly from state statutes, not from its imagination. It's the difference between a wild guess and a researched answer.

"You always need to verify and validate whatever the LLM is giving you," Jarvis emphasizes. Think quality control, not blind faith.

Where Deterministic Tools Still Rule

Not everything needs an AI solution. For ratio studies, you want cold, hard math, not creative interpretation. That's why tools like ValPal incorporate deterministic programs for specific tasks. Upload your sales and assessed values, and get IAAO-compliant statistics every time. No variability, no surprises.

This hybrid approach, AI for complex reasoning and research, traditional algorithms for calculations, represents the sweet spot for assessment offices. It's not about replacing your CAMA system; it's about making it sing.

The Local Knowledge Advantage

Here's what should help you sleep at night: AI will never replace the assessor who knows that the Smith property floods every spring, or that the industrial district's soil contamination affects values. These models train on internet data, not on your decades of local expertise.

"You all have context within the office on the particular problems that you have to work with," Jarvis notes. That context, your understanding of local markets, neighborhood dynamics, and property quirks, remains irreplaceable.

What changes is the grunt work. Instead of spending hours researching storage rates or analyzing taxpayer appraisals, you'll focus on interpretation, judgment, and strategy. The AI handles the research; you handle the decisions.

Implementation Without Intimidation

Start small. Pick one workflow that eats time, maybe it's finding comparable sales or checking exemption eligibility. Test an AI tool on that single task. Measure the time saved and accuracy achieved. Build confidence before expanding.

Most importantly, involve your team early. The offices succeeding with AI aren't the ones with the biggest budgets, they're the ones where leadership demystifies the technology and shows staff how it makes their jobs easier, not obsolete.

Key Takeaways

AI is already here. While you're reading this, venture capitalists are pouring billions into tools that could transform assessment work. The question is whether you'll help shape that transformation or react to it.

Hallucinations require human oversight. LLMs are powerful but imperfect. Trust them like you'd trust a brilliant but green employee, with verification and guidance.

Hybrid solutions work best. Use AI for research and complex reasoning, deterministic tools for calculations and standardized processes. Know which tool fits which job.

Your expertise remains essential. Local knowledge, professional judgment, and human interpretation aren't just valuable, they're irreplaceable. AI amplifies your capabilities; it doesn't replace them.

The assessment profession has weathered technological shifts before. We adapted to CAMA systems, embraced digital mapping, and learned statistical modeling. This is simply the next evolution. The offices that thrive will be those that see AI not as a threat, but as the most powerful tool to hit our profession in decades.

The revolution is here. Time to lead it.