This audit measures whether AI tools are replacing your economic reasoning or supporting it. Your answers show whether you retain genuine control over modelling choices, forecasting confidence levels, and policy conclusions.
Before you ask an AI tool any question about your data or model, write down what you expect the answer to be. If the AI answer differs, that is your signal to investigate, not your signal to trust the AI.
Never use a confidence interval generated by an AI forecast tool without reconstructing the logic yourself. Spurious precision in policy forecasts causes real economic harm.
When an AI tool suggests a variable, specification, or transformation, ask yourself: 'Could I defend this choice to an economist who thinks I am data fitting rather than doing science?' If you cannot answer yes, do not use it.
Keep a separate document of assumptions you are aware of but have not tested. Review it monthly. This prevents buried assumptions from shaping policy recommendations.
If your AI-assisted model contradicts economic theory you trust, that is not a reason to abandon theory. It is a reason to stop using that model until you understand why they diverge.