Americans are turning to AI to survive a brutal housing market—37% would let it buy their next home with ‘minimal human involvement’

The average 30-year fixed mortgage rate rose to 6.95% for the week ended Sept. 17, up 19bps week over week and 69bps from a year earlier, exacerbating affordability pressures amid a limited housing supply. AI adoption is increasing as 72% of LendingTree survey respondents would use it for at least one home-buying or selling task, while 37% would allow an AI to manage a purchase with minimal human involvement. Real-estate platforms are rolling out AI assistants and 92% of agents use or plan to use AI, but concerns over accuracy (63%), legal compliance (49%), and misleading AI-generated listing imagery remain material.

Analysis

Near-7% mortgage rates make affordability tools highly engaging, but engagement alone is not monetization. Z can convert AI-assisted search into higher-intent Premier Agent and mortgage leads, while richer consumer data should improve ad targeting and reduce customer-acquisition costs; the relevant proof point over the next 1-3 quarters is revenue per monthly active user, not assistant adoption. CoStar (CSGP), through Homes.com, is the more direct competitive read-through: AI lowers the cost of matching buyers to inventory, but also risks commoditizing portal search unless proprietary listing, neighborhood, and transaction data remain differentiated.

TREE faces a more ambiguous setup. AI-guided mortgage comparison could increase funnel completion and lead volume, but it also reduces information asymmetry between lenders and borrowers, potentially pressuring lead pricing and lender take-rates; watch lender marketing budgets and TREE's revenue per consumer session. BAC has limited near-term earnings sensitivity because rate lock activity, not consumer research, drives mortgage banking revenue; broader refinancing or purchase-volume recovery still requires a sustained decline in rates rather than better digital tools.

The non-consensus risk is that AI deployment initially raises industry costs and liability rather than removes them. Hallucinated affordability estimates, fair-lending concerns, and misleading AI-generated listing imagery could force human review and increase compliance expense, favoring scaled platforms with legal/data infrastructure over smaller brokerages and proptech vendors. A 6-18 month structural consequence is lower-value agent disintermediation, but high-stakes negotiation, local inventory access, and liability should preserve agents' role while shifting commission economics and lead-generation power toward portals.

Key Decisions for Investors

• Maintain a 3-6 month relative long Z / short CSGP position only if Zillow demonstrates improving revenue per user or Premier Agent conversion; target 10-15% relative upside from faster AI-led lead monetization, with a stop if Z's quarterly traffic growth fails to translate into flat-to-up lead revenue.
• Avoid adding directional TREE exposure on survey-driven AI adoption. Set an alert for quarterly revenue per session, lender count, and variable marketing margin: initiate a tactical short if AI-assisted comparison coincides with declining lead pricing or lender demand, while a sustained improvement in conversion would invalidate the bearish setup.
• Use BAC as a rates expression rather than an AI beneficiary: mortgage-digitization optimism should not materially alter estimates without lower rates and higher originations. Reassess only if the 30-year mortgage rate holds below 6.25% for 6-8 weeks, which would create a more credible volume catalyst for bank mortgage income.
• Monitor Z, CSGP, and residential brokerage peers for regulatory or litigation disclosures around AI-generated valuations, disclosures, and listing imagery. Any mandated human-review requirement or fair-lending enforcement would favor scaled incumbents but could compress near-term AI ROI; reduce portal longs if compliance expense rises faster than advertising yield.