People have nuanced views about AI that can change in relation to perceived risks, benefits, harms, contexts and uses, according to a new review from the Ada Lovelace Institute.
It has brought together research from academia, public institutions and the private sector into What do the public think about AI?, which is aimed at supporting policy makers in the context of this week’s AI Safety Summit.
Its main findings include that studies generally convey the nuanced views among the public on AI.
These include positive attitudes towards some uses, such as in aspects of healthcare and advancing science. But they are countered by concerns around the technology being used to replace human judgement in life affecting decisions on issues such as financial support and job recruitment.
This comes with worries about privacy and the normalisation of surveillance, fuelling a belief that regulation is needed.
Benefits and risks
Overall, the public can simultaneously perceive the benefits as well as risks in most applications of AI, and is generally neutral on whether it will have a positive or negative impact on society, the review says.
It recommends that policy makers should use diverse methods to engage with different groups from the public to understand attitudes, and use the evidence to strengthen regulation and oversight. Applications that could affect civil rights require particular attention.
Octavia Reeve, associate director at the Ada Lovelace Institute, said: “Understanding public attitudes towards AI, and how to involve people in AI decision making, is becoming ever more urgent in the UK and internationally.
“Governing AI requires the meaningful involvement of people and communities, particularly those most affected by technologies.”
The organisation’s public participation and research lead, Anna Colom, added: “Decisions about AI cannot be made legitimately without the views and experiences of those most impacted. However, public voices are still frequently overlooked or absent in decision making.
“Our rapid review provides a timely synthesis of evidence, however there is still a need for more extensive and deeper research and public participation addressing the many uses and impacts of AI across different publics, societies and jurisdictions.”
