The U.S. needs to figure out how to grapple with AI while humans are still behind the wheel.
President Obama thinks artificial intelligence could solve many of the world’s biggest problems — like disease, climate change, even economic inequality. To that end, his administration is recommending more investment in the technology across all levels of government, including funding STEM education to have a prepared workforce, advanced research projects, local grants and new federal infrastructure.
The White House released a 48-page report today featuring 24 recommendations for how the government can be involved in an increasingly AI-powered future, as well as ways to regulate the budding technology.
But the report isn’t without warnings.
For one, the White House predicts artificial intelligence and robotics will upend some jobs, noting that low- and medium-skilled workers are most vulnerable to threats of automation. The administration doesn’t offer a solution, but says it’s an issue that deserves careful attention and pledges to investigate appropriate policy responses.
The report further underscores the extremely fine line between necessary regulation and providing the kind of flexibility necessary to innovation.
Drones provide a useful example. Without necessary regulations, drones will potentially crash into airplanes and each other. But without the flexibility for some form of initial testing, the technology will never become safe enough, and America could lose its position as a global leader.
It’s not just about getting the regulations right. It’s also about federal funding and infrastructure. The Obama administration has already promised $3.9 billion for research and deployment efforts to bring self-driving technology safely to American roadways in the 2017 budget.
One theme that emerges throughout the White House’s recommendations is the need for policy making to involve people with technical expertise and deep understanding of how AI systems work, or else policymakers will be making laws about technology they doesn’t understand and can’t enforce.
This issue came to a head last year when Volkswagen was caught circumventing federal emissions standards in its code. With proper technical expertise in regulatory agencies, the government might have been able to catch VW’s cheating software.
Specific recommendations include:
- Government agencies need to invite (and recruit) technical experts into senior-level policy-making discussions that involve AI-enabled products.
- The U.S. government should invest in and develop a new automated air traffic control system that can manage both autonomous drones and manned aircraft.
- The Department of Transportation should develop an evolving framework for the regulation and safe integration of fully autonomous cars for U.S. roads.
- All federal agencies should adopt open data standards to enable the use of AI-powered research by the government, academics and the private sector.
- Federal agencies that use artificial intelligence to help make consequential decisions should be careful to evaluate outcomes to ensure fairness and prevent discrimination.
The White House is hosting a day-long event tomorrow, The Frontiers Conference, in Pittsburgh on the future of U.S. innovation, including topics in robotics, medicine and space exploration. You can catch the livestream here.