My opinion: we still need humans for an appraisal. These automated valuations don't take into account the condition of the property, which is a huge factor. The comp selection process is more than matching sf within a given radius of the subject property. Yes, the algorithms use more variables, but they can't assess the intangibles that a skilled appraiser takes into account for valuations. The idea that these AVM's are going to save money for the consumer is disingenuous. The fees will correspondingly increase on the banks side...just calling it something else instead of Appraisal Fee.

I see this as another money grab by lenders. Similar to the introduction of AMC's which were originally supposed to be independent 3rd parties between the lender and the appraiser. So the banks formed their own AMC's so they could keep half of the appraisers fees. This step the lenders are proposing would effectively let the banks keep the other half of the fee charged to the consumer. Also, this idea that AI "appraisals" will mitigate risk to the lender is completely bogus. A lower LTV mitigates risk.

Denita is correct. AMC's were created to gather data on every residential structure every built. Think Big Brother. Once they capture that data, theoretically they can create a desk appraisal. Presto, they can fund a loan.

There are large problems with this idea. Structures, neighborhoods and market demand change. Their data can and will be faulty. It requires eyes on the subject and comparables. The farther you get from a city center or suburban development, the sketchier the data.

While AMC's have taken the lender out of the assignment, they haven't resulted in better appraisal results. Data is mother's milk in an appraisal. With bad data, you get bad appraisals. It's pretty simple, garbage in, garbage out. AMC's produce bad product because at some point their data becomes aged and obsolete.