One of the problematic considerations of avm use was the validity of such usage.
When the lenders were in a potential position of disclosing such findings for each and every instance, a new question emerged about the process, and there was quite a bit of news and speculation on it.
What are the confines of the avm, what are the various settings of the software, how does the user influence, validate, or proof check the findings?
It's truly a dummy user program, for all but the controllers of the program. Many utilizers of such avm software are as clueless towards process involved as someone who did not know how to use a computer. It's 'next level' difficulty because avm can involve complex math and complex data base utilization. There are also a lot of somewhat old stories about appraisers being called to task for their use of avm style assimilation or database tools, while at the same time being clueless towards proper usage and high math or calibration involved in the avm database assimilation processes.
On top of that, there were emerging questions about how and if and when and why to share such info with appraisers. Before completion, after completion, to challenge work product, etc.
Lack of transparency with these systems is a consequence of the alternative use argument for valuation products. It's work around mentality at best. If the avms line up, no appraisal necessary. That's built in biased, if I've ever heard it. Now with demins up to 250k or less stellar credit ranges, well that's a scenario where anyone below a certain middle class threshold, does not receive true valuation service from the lender and/or human appraiser. Rather, they get a desktop style product, which one would be hard pressed to reverse engineer the process and prove the avm result is founded or unfounded either way without an actual appraiser services after the fact anyways.
"The settings peramiters? Those are secret bro, we don't share that information!"