Algorithmic Accountability Act Text · Table (p.19)
From algorithmic_accountability_act_text.pdf · page 19
| 0 | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| 7 | |||||
| based on the level of risk; and | |||||
| 8 | |||||
| (C) assessing and documenting the current | |||||
| 9 | |||||
| and potential future or downstream positive and | |||||
| 10 | |||||
| negative impacts of such system or process on | |||||
| 11 | |||||
| the privacy, safety, or security of consumers | |||||
| 12 | |||||
| and their identifying information. | |||||
| 13 | |||||
| (4) Perform ongoing testing and evaluation of | |||||
| 14 | |||||
| the current and historical performance of the auto- | |||||
| 15 | |||||
| mated decision system or augmented critical decision | |||||
| 16 | |||||
| process | using measures | such | as | benchmarking | |
| 17 | |||||
| datasets, representative examples from the covered | |||||
| 18 | |||||
| entity’s historical data, and other standards, includ- | |||||
| 19 | |||||
| ing by documenting— | |||||
| 20 | |||||
| (A) a description of what is deemed suc- | |||||
| 21 | |||||
| cessful performance and the methods and tech- | |||||
| 22 | |||||
| nical and business metrics used by the covered | |||||
| 23 | |||||
| entity to assess performance; | |||||
| 24 | |||||
| (B) a review of the performance of such | |||||
| 25 | |||||
| system or process under test conditions or an |
Source Metadata