Center for the Governance of AI, housed within the University of Oxford, released a report yesterday, titled “Artificial Intelligence: American Attitudes and Trends” based on the findings from a nationally representative survey that they conducted using the survey firm YouGov. The report talks about the results from the survey that provides an insight into the American public’s attitudes and opinion toward AI as well as AI governance. Let’s have a look at some of the major highlights from the report.
Majority of the Americans support than oppose AI development
The report states that Americans express mixed support towards AI development, with the majority of them supporting the development than opposing it. The survey results showed that a substantial 41% of the American respondents strongly support the development of AI and a smaller minority (22%) strongly opposes its development. About 28% of the respondents expressed a neutral attitude towards AI development and 10% stated that they do not know.
Support for AI development varies between gender, race, experience, and education
The report states that support of the American public towards AI development is greater among wealthy, educated, male, or those with experience in technology. The Center for the governance of AI performed a multiple linear regression to predict the support of the American public.
As per the survey results, a majority of the respondents belonging to the following four subgroups expressed support towards AI development:
- The ones with four-year college degrees (57%)
- The ones with an annual household income above $100,000 (59%)
- Those who have graduated with a computer science or engineering degree (56%)
- Those with experience in computer science or programming (58%).
On the other hand, women (35%), those with a high school degree or less (29%), and the ones with an annual household income below $30,000 (33%) showed less enthusiasm towards AI development.
A large majority of Americans want more careful management of AI and robots
The report states that a majority of the American public (more than eight in 10) want AI and robot to be carefully managed, while only 6% disagrees. Center for the governance of AI replicated a question from the 2017 Special Eurobarometer, to compare Americans’ attitudes with those of EU residents. They found out that the 82% of those in the U.S want more careful management of robots and AI, which is not quite far from the EU average, where 88% of the public supports the same notion.
Similarly, 6% of Americans don’t support the notion which is quite close to the EU average where 7% disagreed. The report states that a large percentage of respondents in the survey selected the “don’t know” option.
Americans consider many AI governance challenges to be important
The report states that a majority of the Americans consider AI governance challenges such as prioritizing data privacy and preventing AI-enhanced digital manipulation, etc, of high importance. Respondents of the survey were asked to randomly consider five AI governance challenges out of the given 13.
As per the survey results, the AI governance challenges that Americans think most impactful and important for tech companies to tackle include data privacy and AI-enhanced cyber attacks, and surveillance.
On the other hand, the challenges that are considered on average 7% less likely to be impactful by the Americans include autonomous vehicles, value alignment, bias in using AI for hiring, the U.S.-China arms race, disease diagnosis, and technological unemployment. At last, challenges that are perceived as even more less likely to be impactful includes criminal justice bias and critical AI systems failures.
Americans see the potential for U.S. China cooperation on certain AI governance challenges
As a part of the survey, all the American respondents were assigned three out of five AI governance challenges, on which they see US-China co-operation. The five challenges included:
- AI cyber attacks against governments, individuals, and organizations.
- AI-assisted surveillance violates privacy and civil liberties.
- AI systems that are safe, trustworthy, and aligned with human values.
- Banning lethal autonomous weapons.
- Guarantee of a good standard of living for people who are at risk of losing their jobs to automation.
The survey results showed that China cooperation on value alignment is perceived to be the most likely (48% mean likelihood) and cooperation to prevent AI-assisted surveillance being the least likely (40% mean likelihood).
“In the future, we plan to survey Chinese respondents to understand how they view U.S.-China cooperation on AI and what governance issues they think the two countries could collaborate on”, states the report.
Americans don’t think labor market disruptions will increase with time
As a part of the survey, respondents were asked to select one out of four conditions based on the likelihood of AI and automation creating more jobs than eliminating over the future time frames of 10 years, 20 years, and 50 years.
The survey results showed that on average, American public disagrees with the statement “automation and AI will create more jobs than they will eliminate” more than they agree with it. Also, about a quarter of respondents gave “don’t know” responses. However, respondents’ agreement with the statement increased slightly with the future time frame.
Americans trust the U.S. military, universities, tech firms, and non-gov organizations the most to build AI
The report states that Americans put more trust in tech companies and non-governmental organizations than in governments for the development and use of AI. As a part of the survey, respondents were randomly assigned five actors out of 15 that are not well-known to the public such as NATO, CERN, and OpenAI. Respondents were also asked how much confidence they have in each of these actors to build AI and were again randomly assigned five out of 15 actors.
As per the results of the survey, Americans consider university researchers and the U.S. military as the most trusted groups to develop AI. Half of the Americans responded with a “great deal” or “fair amount” of confidence for this group. Americans expressed slightly less confidence when it came to tech companies, non-profit organizations, and American intelligence organizations. In general, the American public finds more confidence in non-governmental organizations as opposed to the governmental ones.
41% of the American population expressed a “great deal” or even a “fair amount” of confidence in “tech companies,” as compared to the 26% who feel that way about the U.S. federal government. American Public has more trust in intergovernmental research organizations CERN), the Partnership on AI, and non-governmental scientific organizations (e.g., AAAI). Moreover, about one in five respondents in a survey selected a “don’t know” response.
The surveys were conducted between June 6 and 14, 2018, where a total of 2,000 American adults (18+) completed the survey. The analysis of the survey was pre-registered on the Open Science Framework. “Supported by a grant from the Ethics and Governance of AI Fund, we intend to conduct more extensive and intensive surveys in the coming years, including of residents in Europe, China, and other countries”, states the report.