Brought to you by Fannie & Freddie

Appraising Our Future

I finished watching "Sherlock" with Benedict Cumberbatch a few weeks ago and started perusing through Hulu, where I came across "Terminator: The Sarah Connor Chronicles." Personally, I think Lena Headey looks better as a brunette, although her portrayal of Cersei, the cruel mother of Joffrey, in "Game of Thrones," really showcases her talent as an actress. In the second season of "The Sarah Connor Chronicles," Skynet is in its infancy and still learning, thanks to Shirley Manson, who I think may even be the same Shirley Manson from the '90s grunge band Garbage. Anyway, in the show, Skynet is being fed images and often finds itself perplexed by what it sees, much like how I envision the newly announced AI world into appraisals will play out. Fannie Mae and other authorities plan to introduce AI into their appraisal processes, where it will analyze images from appraisers' reports, potentially questioning the accuracy of their assessments.

?

In an era dominated by technological advancements, the integration of artificial intelligence (AI) into various industries has become inevitable. However, as we witness the emergence of AI-driven appraisal systems, it's crucial to tread cautiously and recognize the potential pitfalls that accompany this transition. The allure of AI lies in its promise of efficiency, accuracy, and objectivity. However, the system is only as good as its parameters will allow. While our reality may not involve sentient killer robots, there are parallels. AI, in its infancy, grapples with understanding nuanced visual data, often leading to erroneous conclusions.

?

Enter the world of real estate appraisals, where the stakes are high, and accuracy is paramount. Fannie Mae's proposal to incorporate AI into the appraisal process raises legitimate concerns. Can an algorithm truly capture the intricacies of a property's value, discerning between normal wear-and-tear and structural deficiencies? More importantly, can it replicate the human intuition and expertise honed through years of experience?

?

Furthermore, the introduction of AI poses a fundamental question: who holds accountability when errors occur? While proponents argue that AI reduces human bias, the reality is more nuanced. Bias can seep into AI systems through flawed training data or algorithmic design, perpetuating existing inequalities and systemic injustices.

?

In 2018, the American Civil Liberties Union (ACLU) conducted a test using Amazon's Recognition software. The software falsely matched 28 members of Congress with individuals who had been arrested. Notably, the false matches disproportionately targeted people of color, raising concerns about racial bias in the technology.

?

In 2020, the Detroit Police Department wrongfully arrested and detained a man based on a false facial recognition match. The man was held for 30 hours before being released, highlighting the potential consequences of relying on flawed facial recognition technology for law enforcement purposes.

?

Social media platforms like Facebook have faced criticism for their photo tagging features, which use facial recognition algorithms to automatically identify and tag individuals in photos. Critics argue that these algorithms may exhibit racial and gender bias, leading to incorrect or offensive tags.

?

Photo recognition software may misclassify objects in images, leading to errors in image understanding and analysis. For example, a photo recognition system might incorrectly label a dog as a cat or misidentify a person's activity. And photos taken in low light conditions or with obscured objects may be harder to analyze accurately.

?

With the rise of sophisticated image editing tools, detecting manipulated or edited images has become increasingly challenging. Photo recognition software may struggle to identify images that have been digitally altered. This could become a real problem if homeowners, agents, and anyone with a vested interest in the reported value are allowed to provide photographs of the property.

?

The UAD (Uniform Appraisal Dataset) is a standardized format for appraisal reporting used in the United States. Within the UAD, C3 and C4 ratings refer to specific condition ratings assigned to various components of a property.

B4-1.3-06, Property Condition and Quality of Construction of the Improvements (03/01/2023)

?

1. C3 Rating:

?? -The improvements are well-maintained and feature limited physical depreciation due to normal wear and tear. Some components, but not every major building component, may be updated or recently rehabilitated. The structure has been well-maintained.

Note: The improvement is in its first-cycle of replacing short-lived building components (appliances, floor coverings, HVAC, etc.) and is being well– maintained. Its estimated effective age is less than its actual age. It also may reflect a property in which the majority of short-lived building components have been replaced but not to the level of a complete renovation.

?

2. C4 Rating:

?? - The improvements feature some minor deferred maintenance and physical deterioration due to normal wear and tear. The dwelling has been adequately maintained and requires only minimal repairs to building components/mechanical systems and cosmetic repairs. All major building components have been adequately maintained and are functionally adequate.

Note: The estimated effective age may be close to or equal to its actual age. It reflects a property in which some of the short-lived building components have been replaced, and some short-lived building components are at or near the end of their physical life expectancy; however, they still function adequately. Most minor repairs have been addressed on an ongoing basis resulting in an adequately maintained property.

?

The key difference between C3 and C4 ratings within the UAD framework is the level of condition: C3 indicates an average condition with minor wear and tear, while C4 indicates a property with noticeable defects or deficiencies. It would be nice if there was a C3.5 option, but no such rating exists at this time. And as any of us in this industry knows a good portion of the time the physical observation of the property does not necessarily equate to the photographs that were uploaded to MLS. A good staging company, a mop and bucket, and other tweaks here and there can make a C4 property appear closer to a C3 in photos. And good luck trying to argue condition with the AI as a real estate agent on behalf of your client.

?

Not every component in a home will have an equivalent condition. What if a home has a remodeled kitchen but a crumbling foundation? If the appraiser reports the condition of the property as C4, does he/she risk Fannie/Freddie sending a letter to the State? What if the kitchen is updated with original painted cabinets and new shaker doors? The cabinets are from the middle of the last century, but the door faces are new. What about all of the various materials? We have porcelain that can look like hardwood and laminate that can be made to look like granite. Will the AI be able to tell whether or not the paint is fresh or if the colors are acceptable in the marketplace? Will it detect smoke in the curtains? How about carpeting that was installed a year ago only to have the owner to then adopt an untrained puppy; how will the AI rate the condition?

I realize that we in the appraisal world are the ugly little stepchild of the whole Lending process. And like the undesirables that we are, we seem to be getting the brunt end of the stick more and more. Accusations of bias, often unfounded and unjust, cast a long shadow over our integrity. Meanwhile, the rise of Appraisal Management Companies (AMCs) brings a new wave of challenges, as they impose higher fees and bureaucratic hurdles. It appears as though the powers that be are attempting to all but phase out the human aspect when it comes to evaluations in the industry. And this at a time when the Appraisal Institute is enacting PAREA and those same government institutions are attempting to expand diversity in the industry.

?

The integration of AI into appraisals heralds a new chapter in the real estate industry's evolution. I, for one, embrace this technology as it saves time with charts, graphs, analysis, etc. Of course, I have had to correct the machine on several occasions following errors in the output. As we take another step toward Fannie & Freddie's version of Skynet, let us not forget the invaluable role of human judgment and expertise.

?

If you are an attorney, agent, loan officer, or homeowner with questions on the appraisal process, please feel free to contact the offices of Asap Appraisals, LLC at 262-442-2323, email us at [email protected], or visit our website at wiasapappraisals.com.

?