Artificial intelligence can process data.
It can structure large data sets, recognise patterns and make anomalies visible.
It does not replace the quality of the data basis.
Bad data remain bad data.
Incomplete data remain incomplete.
Non-comparable data remain non-comparable.
For real estate valuation, the data basis is decisive. Relevant information includes resilient data on purchase prices, rents, vacancy, operating costs, property condition, use, location, risk and transaction activity.
AI can evaluate such data more quickly.
It cannot automatically make them correct.
An incorrect rental assumption does not become resilient because it has been processed digitally.
An unsuitable comparable transaction does not become comparable because a model has included it.
An incomplete data basis does not become complete because it has been evaluated automatically.
Data quality is particularly important for investment properties.
Small deviations can have significant effects on value, risk and transaction capability.
Valuation therefore requires more than data processing.
Valuation requires data scrutiny.
Data must be complete, plausible, comparable and value-relevant. They must fit the property, the valuation date and the market environment.
Artificial intelligence can support the process.
It can prepare scrutiny.
It can provide indications.
Responsibility for the data basis remains.
Without resilient data, there is no resilient valuation.
There is only faster processing of uncertain fundamentals.
Bad data remain bad data.
AI does not change that.
That is precisely the limit of automated valuation.
