Protecting the Public from Algorithmic Discrimination

A recent White House Executive Order (on “Further Advancing Racial Equity and Support for Underserved Communities Through the Federal Government”) alerts Americans to an emerging threat most probably hadn’t noticed: algorithmic discrimination. Seriously. Always on the lookout for new forms of discrimination, the Biden administration’s February 16 executive order directs federal agencies to undertake efforts “protecting the public from algorithmic discrimination.”

Correctly suspecting that will leave most Americans scratching their heads, the administration includes this definition:    

The term “algorithmic discrimination” refers to instances when automated systems contribute to unjustified different treatment or impacts disfavoring people based on their actual or perceived race, color, ethnicity, sex (including based on pregnancy, childbirth, and related conditions; gender identity; intersex status; and sexual orientation), religion, age, national origin, limited English proficiency, disability, veteran status, genetic information, or any other classification protected by law.

Critics are already slamming this effort as the administration’s attempt to “reprogram artificial intelligence to ‘advance equity’.” But the executive order is about much more than that. As National Review’s Rich Lowry puts it, it is basically “a Five-Year Plan for the implementation of diversity, equity, and inclusion across a huge swath of government agencies.”

Implementing that plan will naturally take a lot of bureaucrats—starting at the top. Thus the executive order creates a new “White House Steering Committee on Equity,” which is expected to “ensure that agencies undertake ambitious and measurable steps to deliver equitable outcomes for the American people.” Federal agencies will each have their own new “equity team,” which President Biden wants you to know will be “led by a designated senior official (senior designee) charged with implementing my Administration’s equity initiatives, and shall include senior officials from the office of the agency head and the agency’s program, policy, civil rights, regulatory, science, technology, service delivery, financial assistance and grants, data, budget, procurement, public engagement, legal, and evaluation offices, as well as the agency’s Chief Diversity Officer, to the extent applicable.” All that ensuring, delivering, leading, and charging will be in the service of each agency’s crafting an annual “equity action plan.”

As Lowry suggests, “The new order, like so much of the Left’s projects in 21st-century America, is equally sinister and parodic.” But the new order leaves out a fundamental question. When it comes to welfare programs, what exactly does equity mean?

For example, what will the Department of Agriculture’s new equity team make of the fact that non-Hispanic whites constituted 38.4 percent of all food stamp recipients in mid-2020, below that group’s share of the US population in poverty (42.8 percent) then? Or that 2.4 percent of food stamps recipients were Asian Americans, similarly below that group’s share of the US population in poverty (4.4 percent) that year? Would a more equitable social safety net provide more benefits to those groups—despite their having the lowest overall poverty rates of any racial or ethnic categories?

Similarly, what will the Social Security Administration’s equity team make of the fact that two thirds of children receiving Supplemental Security Income benefits are boys? Would equity require equal numbers of girls to qualify?

Once equity teams sort out who is being “disfavored,” they will have to determine what their vision of equity in benefit programs actually looks like. Would it require adding new recipients or removing current ones until their desired balance of “actual or perceived race, color, ethnicity, sex” and so on is achieved? The Executive Order doesn’t say, but unsurprisingly suggests that more benefit provision is the goal: agency equity action plans “shall include” a review of “potential barriers that underserved communities may face in accessing and benefitting from the agency’s policies, programs, and activities” along with “strategies, including new or revised policies and programs, to address the barriers.”

Predictably, no mention is given to how much all this will cost taxpayers of all backgrounds. But the short answer is a lot, especially if it includes reviving the expanded Child Tax Credit temporarily paid in 2021, which the executive order touts for having delivered “more equitable outcomes” and the president called for in his state of the union address.  That one policy would cost well over $1 trillion in the first decade alone, without considering any of the countless other potential expansions from this latest executive order.

Perhaps the administration’s next executive order will tell Americans how much all these “new or revised policies and programs” will cost and just who will pay for them. Because one thing is sure: algorithms won’t be picking up the tab.

The post Protecting the Public from Algorithmic Discrimination appeared first on American Enterprise Institute – AEI.