I'd suspect that a major difficulty of applying AI to real estate is limited fungibility and liquidity. With a stock or currency, there's usually enough of a l

I'd suspect that a major difficulty of applying AI to real estate is limited fungibility and liquidity. With a stock or currency, there's usually enough of a live market that you can sell quickly at roughly market prices. Much of this is due to fungibility-- one share of INTC is worth the same as any other, so they can trade sight unseen with narrow spreads. With real estate, the AI dashboard can flash and beep and scream "Sell the houses in Roanoke now!" but there's no easy way out. You're either spending weeks or months closing out individual retail sales, and if anything goes wrong you're back to square one while the price is deflating around you, or you try to sell it commercially sight-unseen, for which you're going to take a major price haircut. Might work somewhat better with commercial real-estate which seems a bit more oriented to grading and classification to allow more sight-unseen trading.