When a tool contributes to a decision about someone's benefits, services or family, the person needs to know that, understand in plain terms how it affected the result, and have a real way to challenge it. The Idaho K.W. v. Armstrong litigation is a clear example. After people's disability service budgets were cut by a formula, the state initially declined to disclose how the amounts were calculated, citing the formula as proprietary. The ACLU of Idaho reported that the court found it a due process violation to reduce services based on an undisclosed method and later found the formula produced arbitrary results. The settlement described by Idaho's health department includes a new budget methodology, help to file appeals and the ability to see one's full assessment. For DSD, where budget methods are part of Waiver Reimagine, the lesson is direct: people must be able to understand and contest how a number was produced.
The White House Office of Science and Technology Policy published a Blueprint in the early twenty-twenties describing principles for automated systems, including notice and explanation and access to a person who can consider and remedy problems. The page is now archived and is not current federal policy, but its principles remain a useful checklist. Minnesota law adds concrete duties. Under the Minnesota Government Data Practices Act, when a government entity asks a person for private or confidential data, it must tell them the purpose and intended use, whether they may refuse, the known consequences of supplying or refusing, and who else is authorized to receive it. People have the right to see data about themselves and to contest its accuracy or completeness. If data feeding a tool is wrong, those rights are a path to correcting it, and staff should know how to route such requests to the responsible authority.
An explanation must be understandable to the person, not only technically accurate. It should say which information was used, which factors mattered most, what the person can do if information is wrong or incomplete, and how to appeal. Avoid explanations that restate the output, such as your score was high.
Meaningful staff review means a trained person with time and authority considers the case and can reach a different result. A signature on a pre-filled decision is not review. Track how often staff depart from tool outputs, why, and whether departures differ across groups. Build in easy ways for people to reach a person when a tool does not fit their situation, including people with disabilities who may need an accommodation to use or respond to an automated process.
