Artificial intelligence is likely to be the defining technology of the century, affecting everything from war to jobs to health care. So, understanding what the general public wants from AI is important. A new survey suggests that while there’s no strong consensus on the topic, more Americans are in favor of AI than actively oppose it.
In polling organized by the University of Oxford’s Future of Humanity Institute, forty-one percent of respondents said they somewhat or strongly supported the development of AI, while 22 percent said they somewhat or strongly opposed it. The remaining 28 percent said they had no strong feelings one way or the other. The survey defined AI as “computer systems that perform tasks or make decisions that usually require human intelligence.”
These results can probably be chalked up as a win for the pro-AI party, but not by a lot. Respondents’ answers were also correlated noticeably with demographics. Young, educated, and male individuals were all more likely to be in favor of AI development. For example, 57 percent of college graduates were in favor compared to 29 percent of individuals with high school educations or less. This is a notable split considering that much research suggests AI and robotics will increase social inequality.
Interestingly, one area where there was strong consensus was regulation. 82 percent of respondents somewhat or strongly agreed with the statement “robots and artificial intelligence are technologies that require careful management.” What form this management might take was not specified, though when presented with a number of governance challenges involving AI, respondents ranked data privacy as the most important, closely followed by cyberattacks and autonomous weapons.
When it came to who should be responsible for placing limits on AI development, there was no clear answer. Respondents said they trusted university researchers the most (50 percent reporting “a fair amount of confidence” or “a great deal of confidence”), closely followed by the US military (49 percent). Tech companies fared pretty well, with 44 percent of Americans saying they had a “great deal” or even a “fair amount” of confidence in these organizations, though a notable outlier was Facebook. This was the least trusted firm, with four in 10 respondents saying they had no confidence in the company’s abilities.
Other highlights from the survey include:
- Americans are divided on whether they want to see smarter AI. Most respondents (29 percent) neither supported nor opposed the development of “high-level machine intelligence” with the common sense ability of a human, 31 percent somewhat or strongly supported it, and 27 percent somewhat or strongly opposed it.
- We’re less likely to notice AI in mundane technology. While respondents were quite confident AI is used in robots, virtual assistants, and self-driving cars, they were less likely to recognize that the technology also powers Facebook’s photo-tagging feature, Google’s search engine, and Netflix’s movie recommendations.
- AI is considered much less of a problem than nukes or recessions. When asked to rank the impact and likelihood of 15 “global risks,” the threat from AI was judged to be relatively low. Only the threat from a failure to act on climate change was predicted to have less impact, while nukes, terrorist attacks, infectious diseases, and a global recession were all judged to be a much greater threat and more likely to be a problem in the next 10 years.
The polling was conducted by YouGov on behalf of the Future of Humanity Institute’s Center for the Governance of AI. 2,000 Americans were surveyed, with the individuals selected to reflect the demographics of the US, balanced in terms of age, race, education, sex, income, and political preferences.