The U.S. must lead in artificial intelligence
- By David Hoffman
- Mar 06, 2019
The United States must get serious about a comprehensive strategy on artificial intelligence or risk losing leadership in a technology that is different from any other that's been developed so far. Nations that lead in AI stand to gain tremendous advantages across industry, government and society at large. More than 20 countries and governing bodies have already released strategies to promote the development of AI in an effort to reap its benefits and, in some cases, take the lead in AI. But despite a promising recent executive order, the U.S., notably, has not created such a strategy.
A transformative technology, AI's potential benefits include improving healthcare, streamlining national security and cyber defense, enabling more precise data analytics, and making agriculture more sustainable. A 2017 report by PWC estimated that AI technologies could increase global GDP by $15.7 trillion by 2030, representing a potential 20 percent increase, with $3.7 trillion of that increase occurring in North America, for a 14.5 percent increase in GDP. However, if AI is not designed, deployed and integrated responsibly, it also brings risks, including workforce displacement, data hoarding, threats to privacy and potentially unethical uses of AI technology.
The federal government is in the best position to influence the development, funding and adoption of AI and has often supported innovation by the technology sector from afar. The technology industry traditionally has appreciated this "hands-off" approach, given the danger of prescriptive regulation stifling innovation. The nature of this relationship, in which the government has let the private sector lead, has resulted in the U.S. leading in technologies that drive the internet, mobility and more. With AI, however, there is too much at stake. Government and industry must join forces or risk falling behind other nations that already have launched their national strategies.
The Administration made a commendable start to this effort with the recent release of its Executive Order on Maintaining American Leadership in Artificial Intelligence. This initiative sets out a good blueprint that now needs to be filled in – and quickly. A comprehensive U.S. national strategy on AI should, at a minimum, focus on four key areas.
Foster AI innovation
The executive order calls on federal agencies to prioritize funding for AI research in their 2020 budget requests. There should also be a source of dedicated federal funding for R&D on AI. Additionally, while some agencies, like the Department of Defense, already have outlined their own AI strategies, the federal government should mandate this as a widespread practice, helping other agencies do the same. Agencies with AI strategies should also develop ethical guidelines and policies for AI's use in government. These could serve as guidelines and guardrails around AI's use in academia, civil society and the public and private sectors.
The U.S. also needs either to lead or ensure an international coordinated effort to facilitate cross-border data sharing, intellectual property protection and the use of international interoperability standards. Each of these examples would help the U.S. stay on track with AI innovation.
Invest in education and training and protect people's welfare
While AI could replace some jobs, developing AI for future decades will also require an expert workforce. The recently established National Security Commission on Artificial Intelligence, a public-private body, should be funded. The Commission should be tasked with addressing education, workforce and job impact and ethics training and skills retraining, among other critical issues.
Furthermore, the federal and local governments should explore ways to mitigate potential negative impacts of AI-caused employment displacement. Governments might, for example, expand unemployment insurance and benefits, or explore creating a public-private partnership network of National Service opportunities to alleviate job loss.
Sensibly liberate data
The more data is available to an AI system, the more intelligent it can become. Nations that promote open data sources and data sharing are the ones likely to see the most rapid advancements in AI capability. The U.S. government must continue to lead by example and sensibly increase access to government sources of data. Governments need to incentivize sharing and can do so through tax or policy incentives. In addition, the U.S. government should seek to prohibit the imposition of tariffs on data flows around the world.
Create a supportive legal and policy environment
While the federal government should use caution before adopting new laws, regulations, taxes or controls that may inadvertently impede the responsible development of AI, it is past time for the U.S. to adopt a comprehensive federal privacy law. As data is so key for the development and adoption of AI, governance around robust privacy protections is essential.
A federal privacy law would allow organizations to develop business practices around one clear set of directives and principles rather than having to comply with various state laws. It should also improve the Federal Trade Commission's ability to mitigate individual and societal harm that may occur as a result of AI implementation.
With a national AI strategy, the U.S. has a great opportunity to increase its industrial competitive advantage, improve the quality of life for people and maintain its technological leadership. It's not enough to rely on the private sector to drive AI forward when other nations are jumping in with significant financial and policy commitments. The U.S. needs to put more focus, investment and accountability behind AI, and it needs to do it soon.
David Hoffman is associate general counsel and global privacy officer, Intel Inc.