Unconvinced: how the world feels about AI

What did the World Risk Poll find out about people’s attitudes towards AI?

Author: Dr Jan Przydatek, Director of Technologies, Lloyd’s Register Foundation

_____________________________________________________________

Globally, almost three in 10 people (28%) say AI will mostly harm’ people in their country in the next 20 years.

That’s according to the Lloyd’s Register Foundation World Risk Poll – the first global study of people’s perceptions and experiences of risks to their safety.

Conversely, 39% – less than two in five – say it will mostly help’. The remaining majority is unsure, showing a substantial degree of global public scepticism about the benefits of AI technologies.

According to the poll, people in the UK are only slightly more optimistic than the global average, with 44% saying AI will mostly help’ and 28% that it will mostly harm’. While the poll only asks about AI in a general sense, this finding provides broader context for the findings of the Ada Lovelace Institute and The Alan Turing Institute’s recent joint report, which investigated the attitudes of people in the UK to specific AI use cases.

The study showed mixed feelings among British citizens, with cautious optimism: a majority of respondents saw 10 of the 17 use cases asked about as more beneficial than concerning, two as similarly beneficial and concerning, and expressed more concern than perceived benefit about the remaining five.

Undoubtedly, AI technologies have huge potential to realise a multitude of economic and societal benefits, including within Lloyd’s Register Foundation’s own mission space to engineer a safer world, as demonstrated by its ongoing investment in the Data-Centric Engineering programme at The Alan Turing Institute.

But to safely realise the full benefits of these technologies for the whole of society, we must bring the whole of society along on that journey, and to do that we must understand the contextual factors that lead some to be more concerned than others. The World Risk Poll provides insights into some of these.

At a global level, the poll reveals that experience of discrimination (for instance on the basis of ethnicity, sex, religion or disability) galvanises opinion on AI, making people slightly more likely to say AI will either help or harm people in their country (rather than that they don’t know or have no opinion). Moreover, there are also a number of countries where experience of discrimination significantly increases worry about harm, including 16 countries where the increase is more than 10 percentage points.

Meanwhile, experience of discrimination also makes people across the world much more likely to say they are very worried’ about governments misusing their data, from a third (34%) of those who have not experienced it to almost half (48%) of those who have experienced two or more forms of discrimination. This relationship is less pronounced in the UK but still present, with concern rising from 14% to 22%.

Discrimination is not the only factor at play. The World Risk Poll also reveals how income plays a role in attitudes to the use of personal data, the collection of which is indispensable to programme AI technologies. More than half (56%) of internet users worldwide who state they are finding it very difficult’ to get by on their current income say they are very worried’ about their personal information being stolen online, compared with about a third (35%) of those who state they are living comfortably’.

These findings suggest that in many countries around the world, the use of both personal data and AI technologies may be seen by vulnerable groups as means of reinforcing and amplifying existing biases and discriminatory power structures, causing them more harm than good. As we have seen in other areas of the World Risk Poll, such a perception of risk can also cause harm in itself, by excluding already marginalised groups from the social and economic opportunities that digital technologies offer.

It is therefore critical that policymakers, businesses and digital innovators construct and regulate AI systems in ways that counter biases and that they engage with vulnerable communities to give them the confidence to use data programmes and technologies that can benefit them.

To speak to the Lloyd’s Register Foundation team about its World Risk Poll, please get in touch via the Poll website.