The Big Picture: Why Property Taxes Matter for Economic Growth and How AI Is Changing the Office

About this Episode

This episode captures a conference presentation by Will Jarvis, founder and CEO of ValueBase, delivered to a room of assessors and appraisal professionals in Charlottesville, Virginia. Rather than a narrow product demo, Will uses the session to build a sweeping argument: property taxes are the most important lever for American economic growth, and AI is the tool that will keep the profession viable enough to protect that lever. It's a bold thesis — one that connects Henry George, Sam Altman, total factor productivity, and the day-to-day reality of running an assessor's office into a single narrative thread.

What makes this talk worth unpacking is that Will isn't just cheerleading for AI adoption. He's making an economic case for why the work assessors do matters at a civilizational level — and then being refreshingly candid about both the power and the limits of the technology he's selling.

Property Taxes Are an Economic Engine, Not Just a Revenue Source

Will opens with a macro-level argument that most assessors rarely hear articulated this clearly. Since the 1970s, total factor productivity growth in the United States has flattened. The pie stopped getting bigger as fast, and political fights over how to slice it got uglier.

He walks through the interventions we've tried — corporate tax cuts in the '80s, globalization in the '90s, fracking in the 2010s — and lands on tax policy as the next frontier. Specifically, he cites research from the University of Virginia suggesting that shifting half of the national tax burden from income onto property taxes could double the growth rate and halve inequality.

The logic is Georgist at its core. Land doesn't depreciate. As societies get wealthier, more wealth gets absorbed into real estate values. Taxing that value more heavily encourages development, discourages speculation, and funds public goods without suppressing economic activity the way income or sales taxes do. Will points out that America already has the highest rate of property taxation in the OECD — and argues that's a feature, not a bug.

Tax Revolts Are the Real Threat

If property taxes are so important, then tax revolts are an existential risk. Will frames the recent pushes to eliminate or freeze property taxes — DeSantis in Florida, the reappraisal ban in North Carolina — as symptoms of a system under stress. Vertical inequity, capacity shortages, and manual data problems erode public trust. Legislators who don't understand the technical realities respond with blunt instruments.

Will's position is clear: assessors need to get in front of policymakers and make the case. He recounts convincing an Ohio gubernatorial candidate over dinner on Nantucket that eliminating property taxes would be a mistake. Some legislators are unreasonable, he concedes, but many simply don't know. The complexity of the system works against it in the public arena. This is where AI enters the picture — not as a silver bullet, but as a way to help offices maintain quality and fairness with fewer resources, reducing the conditions that trigger revolts in the first place.

AI Is Powerful, but It's Mimicry — Not Reasoning

Will's explanation of how large language models work is one of the more honest accounts you'll hear from someone whose company depends on them. The core insight behind the transformer architecture is statistical: all human language follows patterns, and if you train on enough text, you can predict what comes next with startling accuracy.

But Will is careful to draw the line. These models mimic intelligence. They don't reason. They perform poorly on out-of-sample problems — situations they haven't encountered in training data. They hallucinate. They struggle with math unless forced to use external tools. Their knowledge is frozen at training time. And if you use free versions, your data may be used for further training.

This candor matters because the assessment profession demands defensible, explainable results. Will acknowledges that even machine learning valuation models — gradient boosted trees, GRNNs — can overfit and sales-chase just like a poorly built regression. A COD below five should raise eyebrows, not applause.

The Real Skill Is Management, Not Prompting

Perhaps the most practical insight from the talk is Will's reframing of what AI fluency actually looks like. It's not about clever prompting. It's about management. These models need context about your jurisdiction, your office, your local market quirks. They need to be corrected, validated, and supervised — just like a new hire who's book-smart but has never worked in your county.

Will notes that his own workload has doubled since adopting AI tools, even as productivity has quadrupled. The bottleneck isn't compute or model intelligence. It's finding people who understand problems well enough to describe them to a model and then validate the outputs. That's the skill assessors should be developing: not data entry, but strategic oversight.

Key Takeaway

Property taxes aren't just a local government funding mechanism — they're a growth policy with national implications, and the profession that administers them is under real pressure. AI can help offices maintain quality and equity with constrained resources, but only if assessors approach these tools as managers, not as magic. The hardest and most valuable work ahead isn't running models — it's knowing your jurisdiction well enough to tell the model what matters and catching it when it gets things wrong.