The statistics come from the latest report from the Lloyd’s Register Foundation World Risk Poll, powered by Gallup, titled A Digital World: Perceptions of risk from AI and misuse of personal data. The data, which features responses from over 125,000 people in 121 countries, has revealed nearly two thirds (65%) of people living in Scandinavian countries believe AI will mostly help people in the future. However, these figures drop when asking those who have experienced discrimination based on their race, skin colour or sex.
In Norway and Denmark, where attitudes towards AI are among the most positive in the world, 26% of respondents who had experienced discrimination said AI would ‘mostly harm’ people. Meanwhile, only 14% of Norwegians and 15% of Danes who had not experienced discrimination felt the same way. This gap is slightly smaller in Sweden (21% versus 11%) – though still among one of the largest in the world.
The report notes efforts governments to develop ‘trustworthy’ AI, such as Norway’s National Strategy for Artificial Intelligence – inviting public debate on the ethical use of AI to address potential bias.
Dr Sarah Cumbers, Former Director of Evidence and Insight at Lloyd’s Register Foundation, said: “AI technologies are rarely deliberately discriminatory. Nonetheless, programming on the basis of data that does not represent the whole population can often be the source of serious unwanted consequences which greatly affect portions of the population, and the World Risk Poll has shown the extent of concern among these communities.
“What may be perceived as a design flaw – such as the sensitivity of facial analysis software being influenced by skin tone – is often the symptom of a larger inclusivity problem and can greatly contribute to negative perceptions of AI. One of the most striking cases of this can be found in Scandinavian countries.
“While the population as a whole feels optimistic about AI – more so than much of the rest of the world – responses from those who have experienced discrimination show a very different perception that policymakers cannot ignore. While we applaud some of the steps countries such as Norway have already taken, there’s still much to do to combat AI bias. This starts with inclusion of those most affected from the earliest possible stages of design, but also with policies and practices that make the benefits of these technologies accessible for all.”
Data snapshots for each of the countries can be downloaded here: Sweden, Norway, Denmark.