There is widespread agreement among ethicists and tech advocates that responsible AI principles require fairness, transparency, privacy, human safety, and explanability. But it is not always clear how to operationalize these broad principles or how to handle situations when conflicts arise between them. Moving from the abstract to the concrete when developing algorithms often presents challenges as a focus on one goal can come at the detriment of alternative objectives. A new Brookings report offers six steps to responsible AI in the federal government.
On April 5, the Center for Technology Innovation at Brookings hosted an expert panel that will cover ways to operationalize responsible AI and move toward more concrete standards. Panelists also discussed how to design appropriate algorithms and build technical capacity in the workforce.