More US Investment, Talent Needed for AI Field, But Cybersecurity a Concern, Hearing Told
More investment in artificial intelligence and more talented researchers are needed to keep the U.S. globally competitive, but policymakers also should be aware of cybersecurity and regulatory issues that could hamper AI advancement, witnesses told the Senate Commerce Subcommittee on Space, Science and Competitiveness at a Wednesday hearing.
Sign up for a free preview to unlock the rest of this article
Communications Daily is required reading for senior executives at top telecom corporations, law firms, lobbying organizations, associations and government agencies (including the FCC). Join them today!
Eric Horvitz, managing director of the Microsoft Research Lab, said AI can present new kinds of vulnerabilities that need to be analyzed. He said organizations need to think about implementing security in "classical" ways such as authentication models and encryption. But, in response to a question from full Committee Chairman John Thune, R-S.D., Horvitz said AI also can be used to enhance cybersecurity, such as by detecting anomalous patterns with low false positive rates. Responding to a question from Subcommittee Chairman Ted Cruz, R-Texas, about regulations that could hamper AI advancement, Horvitz said more access to public datasets is needed. He said it's very difficult to get medical data shared and the Health Insurance Portability and Accountability Act needs to be thought through more. A system possibly could be developed to maintain privacy while providing access to large amounts of data, he added.
Andrew Moore, dean of Carnegie Mellon University's computer science school, said there's potential ambiguity in responsibility with new AI products. He said a new collision avoidance system might be developed for a self-driving vehicle that greatly reduces accidents, but the "potential danger" in deploying that technology is the impasse among insurers, lawmakers, motorists and carmakers regarding their responsibilities related to the technology. Moore worries about a talent war. He said there aren't many researchers available to work in AI given the industry demand, and even professors are going to industry. He said Americans should start teaching more mathematics to middle school students as a way to get more talent in the U.S. Greg Brockman, chief technology officer of the nonprofit Open AI, said the U.S. has the lead in the AI field now, but not a monopoly. Canada, China, South Korea and other countries are making large investments and getting results.
Brockman told lawmakers U.S. schools need to supply more trained AI researchers because industry has a big need, but a diversity of professionals also is needed within that community. He said experts in agriculture, law and philosophy need to work with AI researchers to provide more perspective. He added the government needs objective measures to gauge AI progress and should sponsor contests to spur innovation.
Experts said AI could save thousands of lives and billions of dollars when applied to many sectors, such as agriculture, education, healthcare, manufacturing and transportation, and provide accessibility for those with special needs. Cruz cited a 2016 Accenture report that predicted AI could double annual economic growth by 2035 and increase labor productivity by up to 40 percent. He noted Forrester predicted AI will get a 300 percent increase in investment in 2017 compared with this year.