20 Minute AI Appraisals Are Coming

[ucbruin]

REGULATORY BODIES' & THEIR STINKING REGULATIONS....
Are only for libtards....

 

[J Grant]

Poor choice of words on my part. The AI agent will grid X number of comps based on your parameters from the prior reports and guidelines it was trained with.

SPARK and similar software companies already do that.

AI is over hyped for most applications, yet people line up like lemmings for it -

 

[alebrewer]

If you think what you do doesn't have any value over automated processes then yes you should be worried.

As I see it, AI (or whatever nomenclature you want to use) currently struggles with external, physical, and functional issues. Give an AVM a home in a homogeneous market - it probably outperforms an appraiser, as appraisers bring bias into the model - whether they want to or not. Its identifying proximity to a dump site that it struggles with. Or the bathroom being located off a bedroom (which - if we're honest - appraisers struggle with also). The physical piece is quicky being addressed with photo recognition software. And you can bet the external piece isn't too far behind - teaching LLM's to recognize golf courses, or busy arterials, or commercial, or whatever the factor is. Then you're only left with the functional piece - which is a very small % of the overall appraisable population...

 

[shrubberyvaluation]

As I see it, AI (or whatever nomenclature you want to use) currently struggles with external, physical, and functional issues. Give an AVM a home in a homogeneous market - it probably outperforms an appraiser, as appraisers bring bias into the model - whether they want to or not. Its identifying proximity to a dump site that it struggles with. Or the bathroom being located off a bedroom (which - if we're honest - appraisers struggle with also). The physical piece is quicky being addressed with photo recognition software. And you can bet the external piece isn't too far behind - teaching LLM's to recognize golf courses, or busy arterials, or commercial, or whatever the factor is. Then you're only left with the functional piece - which is a very small % of the overall appraisable population...

Some external factors can be a very major part %.

 

[Joe Flacco]

As I see it, AI (or whatever nomenclature you want to use) currently struggles with external, physical, and functional issues. Give an AVM a home in a homogeneous market - it probably outperforms an appraiser, as appraisers bring bias into the model - whether they want to or not. Its identifying proximity to a dump site that it struggles with. Or the bathroom being located off a bedroom (which - if we're honest - appraisers struggle with also). The physical piece is quicky being addressed with photo recognition software. And you can bet the external piece isn't too far behind - teaching LLM's to recognize golf courses, or busy arterials, or commercial, or whatever the factor is. Then you're only left with the functional piece - which is a very small % of the overall appraisable population...

Yeah, but how will it determine the impact of those differences?

The reaction to those differences are very micro market specific.

 

[alebrewer]

Some external factors can be a very major part %.

that is correct.

 

[alebrewer]

Yeah, but how will it determine the impact of those differences?

The reaction to those differences are very micro market specific.

Theoretically the same methodologies that appraisers use... I know for a fact that AI uses regression better than most appraisers. Teaching it to perform paired sales would be much easier than regression I'd think.

 

[Joe Flacco]

And sometimes being on a busy street doesn't even matter. Sometimes a two lane big street is a huge deal.

 

[alebrewer]

And sometimes being on a busy street doesn't even matter. Sometimes a two lane big street is a huge deal.

yep.

 

[Joe Flacco]

Theoretically the same methodologies that appraisers use... I know for a fact that AI uses regression better than most appraisers. Teaching it to perform paired sales would be much easier than regression I'd think.

I totally disagree with you. I will believe it when I see it.