The Centre for Data Ethics and Innovation took a global approach to gauge public opinion on AI safety by commissioning Deltapoll for a comprehensive survey, published on October 27 and targeting nine diverse nations in anticipation of the UK’s AI Safety Summit 2023. Participants from Canada, France, Germany, Italy, Japan, Singapore, South Korea, the UK, and the US provided their perspectives, with a clear trend of endorsing rigorous AI safety testing.
Findings indicated robust support for the involvement of independent experts in evaluating powerful AI, with approval ratings spanning from 59% in Japan to a striking 76% in both the UK and Singapore. Additionally, a government-backed AI safety institute received backing across the board, with 51% in Japan up to 65% in Italy endorsing the idea, and 62% support noted in the UK.
The survey highlighted surveillance for crime detection and medical diagnosis as AI’s most applauded applications. However, it also uncovered substantial concerns regarding AI’s darker potential, such as cyberattacks, biological weapon design, and the overarching fear of humans losing control over AI.
A noteworthy consensus emerged around the urgency to mitigate existential risks posed by AI, drawing parallels with global challenges like pandemics and nuclear warfare. Even though the level of agreement varied — from 56% in South Korea to 40% in Germany — opposition remained minimal across all nations surveyed.
Opinions were more polarized when weighing AI’s risks against its benefits. The majority in countries like Canada, France, Germany, Japan, the UK, and the US perceived AI as more risky, whereas in Italy, Singapore, and South Korea, respondents were more optimistic about its benefits.
Deltapoll conducted the survey online between October 9th and 13th, 2023, ensuring representation from each country’s adult population through weighted data. Participant numbers ranged from 1,090 in the UK to 1,164 in Germany, totalling over 10,000 responses globally.
