AI Audit in Property Valuation: A Call for Transparency

Before buying and selling real estate, all parties involved usually conduct property valuations. Banks also need to perform valuations when approving mortgage loans to determine the maximum loan amount. In the past, institutions had to hire professional valuers (appraisers) to assess real estate, which was not only costly but also time-consuming.

Therefore, the development of Automated Valuation Models (AVM) in recent years — valuation systems powered by artificial intelligence (AI) — has gained popularity. These systems process vast datasets to generate instant valuations, significantly improving efficiency in related industries.

However, these AI valuation systems lack high transparency and can be considered “black box” operations, meaning even AI programmers cannot explain how certain valuations are derived, with explanations limited to the factors considered.

The opacity of AI valuations has real-world implications. New Zealand’s courts have recently mandated that an appropriate person must explain the reliability of AI-generated materials in court proceedings. This new requirement underscores the urgent need for a robust AI governance system. It is essential to build an independent professional body, akin to an auditor in businesses, to audit the biases of AI estimates.

In our ongoing research, we are developing a framework that evaluates these automated valuations. The rapid integration of AI into property valuation in New Zealand is no longer just about innovation and efficiency. It is a pressing issue of trust, transparency, and accountability. Our goal is to establish a system that ensures transparency and accountability in property AI valuations.

1. Consultation with Experts:

We began with consultations among leading experts in property valuation to identify key issues and gaps in the current AI valuation systems.

2. White Paper on AI Governance:

This consultation led to the development of a White Paper outlining the necessity of AI governance frameworks to ensure transparency and accountability in property valuations.

Yiu, Edward Chung Yim and Cheung, William, How to Bring AI-Enabled Valuations to Meet Professional Duty of Care Standards? (February 8, 2024). Available at SSRN: http://dx.doi.org/10.2139/ssrn.4720254

3. Roundtable Forum:

We held a roundtable forum that brought together experts from property valuation, computer science, and commercial law to discuss the implications of AI in our sector and the importance of establishing a robust auditing framework.

D. Franklin et al. (2024, September 3), AI Governance and Explainability in Property Markets — a Roundtable Forum. https://cassyni.com/events/S2x5yMu9DdXSUp5HxjnDVo

4. Research Translation: The Conversation Article:

Building on these insights, we authored an article on The Conversation that highlights our findings and the pressing need for independent audits of AI-generated valuations.

Cheung, K.S. & Yiu, C.U. (2024) Use of AI in Property Valuation is on the Rise — but we need more Transparency and Trust, Conversation, Oct 11. https://theconversation.com/use-of-ai-in-property-valuation-is-on-the-rise-but-we-need-greater-transparency-and-trust-240880

5. Journal Article:

Our research continues with a journal article detailing an AI audit framework and presenting initial results from the New Zealand housing market, focusing on bias correction and enhancing the reliability of AI valuations.

to be updated …