At a proptech conference last month, a vendor demo showed an AI tool underwriting a multi-family acquisition in under three minutes. The crowd applauded. What no one asked was whether the person using that tool could tell if the output was wrong.
That question now has a data point. MRI Software's 2026 Commercial Real Estate Pulse Check surveyed nearly 600 CRE professionals across leadership, property management, accounting, and asset management roles. The headline finding: 62% of organizations say they are preparing for AI. The catch: 54% offer no AI training at all.
The two numbers together describe an industry that has convinced itself it is ready for a technology it is not teaching its people to use. The gap between adoption and competence is not a future risk. It is a present liability.
MRI's survey breaks down what little training does exist. The most common forms are usage guidelines and responsible use policies. Both are necessary. Neither teaches an employee to get reliable results from an AI tool. Source validation, the ability to critically evaluate AI outputs, is offered by only 16% of organizations. Prompting exercises, which teach employees how to construct queries that produce repeatable results, are offered by just 14%.
Carla Hinson, MRI's VP of Solution and Innovation, put it directly: "We are finally starting to understand the importance of asking AI the right questions in the right way." Prompting is not instinctive. It is a learned skill with a direct bearing on whether outputs are useful, accurate, and consistent enough to act on. When organizations skip that instruction, they leave the quality of AI outputs to chance.
The survey also reveals where AI benefit concentrates. Over half of respondents plan to use AI primarily for improving personal performance over the next 12 months. Automating administrative tasks comes in second at 45%. The pattern holds across roles: executives, property managers, and property accountants all cite personal productivity as their primary AI use case at rates between 53% and 56%.
When AI benefit stays at the individual level, it becomes harder to capture the full return on investment of training and tools. A property manager who uses AI to draft lease summaries faster is a marginal gain. A team that uses AI to flag covenant breaches across a portfolio is a structural advantage. The survey suggests most organizations are settling for the former.
The consumer-facing AI tools that most CRE professionals encounter first are designed to feel intuitive. They respond to plain language, produce polished outputs quickly, and create the impression that effective use requires no expertise. That perceived accessibility has led a significant number of organizations to treat AI deployment as a plug-and-play decision rather than a capability-building exercise.
This is not a technology problem. The tools work. It is a management problem. Organizations are spending on AI licenses and integration while underinvesting in the human layer that determines whether those tools produce insight or noise. In a capital-intensive industry where underwriting errors and lease interpretation mistakes carry real dollar consequences, that asymmetry matters.
The broader implication for capital markets is indirect but real. As institutional investors push for operational efficiency and data-driven decision-making, they are beginning to ask about AI readiness in due diligence. A firm that cannot demonstrate that its people know how to validate AI outputs is a firm with an unmeasured operational risk. That risk will eventually show up in underperformance relative to peers who invested in both the tool and the training.
The conference applause was real. The training gap is real. The question for every CRE organization is whether they are building the capability to make AI reliable or just the appearance of it.