In 2018, several major controversies involving AI served as an alarm bell for technologists, policy makers and the public. Technology may have made significant progress in many areas, but it can also fail catastrophically when constructed sloppily or improperly.
It is not surprising then that Americans have mixed support for the continued development of artificial intelligence and that most agree that it should be regulated, according to a new study by the Center for the Governance of AI and the Future of Humanity Institute at Oxford University.
For Allan Dafoe, the center's director and co-author of the report, these lessons are important for policy makers and technologists to consider in the debate on how to best advance and regulate AI. "There is currently no consensus for the development of advanced AI, nor is it good for humanity," he says. "This kind of perception could make the development of AI seem illegitimate or provoke adverse political reactions to its development."
Sign up for the L & # 39; algorithm
Artificial Intelligence demystified
By registering, you agree to receive newsletters and emails.
notifications from MIT Technology Review. You can change your preferences at any time. See our
It is clear that decision makers in the United States and around the world need to better understand public concerns and how to respond to them. Here are some of the key findings of the report.
Americans are not sure that AI is a good thing
Although more Americans argue that they do not oppose the development of AI, there is no strong consensus in either case.
A higher percentage of respondents also think that high-level artificial intelligence would do more harm than good to humanity.
When asked to rank their specific concerns, they indicated that the weakening of data privacy and the increased sophistication of cyberattacks were at the forefront – issues of great importance and issues likely to arise. 39 affect many Americans in the next 10 years. Self-contained weapons closely followed, but were ranked with a lower probability of large-scale impact.
Americans want better governance of IA
More than 8 out of 10 Americans believe that AI and robotics need to be managed carefully.
It's easier said than done because they do not trust a particular entity to assume that role either. Of the various options presented by federal and international agencies, corporations, non-profit organizations and universities, none received more than 50% of respondents' confidence to responsibly develop and manage AI . However, US military and academic researchers have had the greatest confidence in the development of technology, while high-tech companies and non-profit organizations have more confidence than government actors to regulate it.
"I believe that artificial intelligence could be a huge benefit," says Dafoe. But the report shows a major hurdle in achieving this: "You need to make sure that you have a broad, legitimate consensus on what the company will undertake."
Correction: An earlier version of this story had a typo in the first chart, showing that 31% of Americans "somewhat oppose" the development of AI. It has been corrected to 13%.