The Impact of Embedded Intelligent Agents on Tax-Reporting Decisions
The use of tax preparation software to meet federal tax-reporting requirements has dramatically increased in the last decade. The general assumption is that such software improves the accuracy of taxpayers' returns, in part because embedded intelligent agents identify potential form errors, provide interpretations of tax laws, and highlight potential IRS audit flags. However, it is possible that these intelligent agents may have other, unintended effects as well. In particular, it is likely that the audit warnings embedded in many of these products may cause many taxpayers to take more conservative positions in their tax returns. Taxpayers most likely to be affected in this way are those who are relatively less knowledgeable about tax laws and reporting requirements. This study presents the results of a computerized tax experiment that are consistent with the above expectations. For novice taxpayers, the audit flags embedded in the software led to conservative adjustments that are rather extreme, resulting in significantly higher reported taxable incomes relative to their counterparts who did not have access to the embedded audit flags. Knowledgeable taxpayers, in contrast, maintained essentially the same level of taxable income and corresponding tax liability despite software warnings of potential audit.