Updated about 14 hours ago on . Most recent reply
AI estimating for construction is becoming more common.
On standardized projects in controlled conditions it can be a useful starting point.
On a new build with site specific conditions it can get you into serious trouble if you treat the output as a finished product rather than a first draft that requires human verification.
I have seen AI generated construction estimates that missed ground stabilization requirements. Along with roughly ten other site conditions that an experienced estimator catches because they have physically stood on enough job sites to recognize what certain soil composition means before the engineer confirms it. The AI produced a number. The number was wrong. Not because the technology failed but because the person deploying it confused efficiency with expertise.
That distinction matters in construction more than almost any other field because the consequences of a wrong number are not theoretical. They show up as change orders after the contract is signed when walking away is no longer realistic.
The broader pattern is worth understanding before you deploy any AI tool in your investment process.
AI deployed to make good work faster is a competitive advantage. AI deployed to replace the judgment that made the work good is a liability that does not announce itself until the project is already committed.
McDonald's runs AI order takers profitably because the transaction is standardized and the human judgment involved was minimal. You cannot apply that logic to a construction estimate on a site with non standard soil conditions. Or to a nurse assessing a patient. Or to a due diligence process on a commercial acquisition where the things that matter most are the ones that don't show up in the data.
Use the tools. Verify the output with someone who has stood on enough job sites to know what the data is missing.
What are you currently using AI for in your investment process and what are you still doing manually?