The blog of the Urban Institute
February 13, 2019

Five recommendations for agencies to best harness Trump’s executive order on AI

February 13, 2019

On Monday, President Trump signed an executive order to help ensure America maintains its leadership in artificial intelligence (AI).  

While the order contains 10 sections, one of the most significant elements, section 4, requires all heads of agencies funding research and development (R&D) to consider AI as an agency priority for R&D and plan accordingly when developing budget priorities for fiscal year 2020 and future years. The order does not contain additional funding to accomplish these goals.

Here at Urban, our office of technology and data science has been leveraging AI to inform decisionmakers and policy research for the past few years. Recently, we’ve published papers using machine learning to create new zoning data, using Twitter data to measure the connection between digital and physical segregation, leveraged cloud computing to produce simulation results at scale, and built a risk assessment clearinghouse to ensure effective use of AI in criminal justice.

Based on our experience with artificial intelligence, here are five priority actions we would recommend agencies prioritize in their AI planning and budgeting processes:

1. Leverage AI to fill knowledge gaps

...and obtain timely information on pressing policy issues

Combining existing gold-standard data with new, real-time data sources—including administrative records—using machine learning can help fill knowledge gaps. Using existing data as the “ground truth,” analysts can apply AI and machine learning methods to more real-time data to provide more timely information on pressing policy issues.

2. Fund AI-powered what-if tools

...to enable policymakers to make better data-driven decisions

When policies are proposed, places like Urban evaluate, or “score,” the outcome. But what if we could score thousands of potential policies at once and let policymakers choose the ones with the best outcomes?

By leveraging the latest cloud computing technology and predictive analytics, policymakers could be empowered with better information on the most effective policies in real time.

3. Invest in research

...to better understand how AI is changing our society and its implications for inclusion, equity, and social justice

Artificial intelligence is changing the way we work, the way we deliver health care, how we buy homes, and how our criminal justice system operates. As AI systems move deeper into the social world, they create opportunities and challenges, and we need to better understand how these technologies are changing society so we can leverage them to improve opportunity for all.

4. Push for greater transparency

...along with accountability and appropriate assessment and oversight of AI systems

Automating processes, data collection, and decisions could bake bias into the system, but AI algorithms could also make biased decisions more transparent. But transparency alone is not enough, and according to the most recent report on the state of AI, and despite the encouraging developments around ethics and transparency in AI, accountability continues to be the most pressing concern that requires the government’s attention.

According to the AI Now 2018 report, “Governments need to regulate AI by expanding the powers of sector-specific agencies to oversee, audit, and monitor these technologies by domain.”   

5. Expand the AI debate

...beyond technology experts and engage a broader coalition of researchers, lawyers, civil society, and advocates to inform the AI debate

As AI has become more widely adopted, advocates, journalists, and technologists have raised concerns about the limits of technological solutions to problems of fairness, bias, and discrimination and the need for more translators who can develop tools and act as guides so that everyone can weigh in on the future of AI.

We need to help advocates, policymakers, journalists, and the general public better understand the trade-offs and limits in existing AI systems and algorithms so we can have an informed public discussion of which trade-offs we want to make for the public good and ensure adequate protection for civil rights.  

AI for all

AI could define our society’s future, and these recommendations are just a few among many ideas to strengthen American competitiveness in this space.

Ensuring that the US continues to lead means that we must take advantage of the incredible advances AI has to offer, while ensuring that technology provides those benefits equitably for all Americans.

Illustration by FrankRamspott via GettyImages.

SHARE THIS PAGE

As an organization, the Urban Institute does not take positions on issues. Experts are independent and empowered to share their evidence-based views and recommendations shaped by research.