AI has a serious image problem

8th August 2026

When even AI bosses are worried about the impact of the technology, is it any wonder the public are scared?. Analysing their major survey of British public opinion, Bobby Duffy and Elena Simperl argue that, while the risks should not be downplayed, if we want to realise the benefits of AI it is important that the public are not already set against it.

Dario Amodei and Sam Altman, the bosses of Anthropic and OpenAI, have recently changed their tunes on what AI will do to jobs. Just a year ago, Mr Amodei predicted that AI will eliminate half of all entry-level white-collar jobs in the next five years. Mr Altman made similar doomy prophecies. Now, they are both more optimistic: Altman, for example, is “delighted” to lean into the alternative vision – that AI will take tasks but enhance roles, freeing up time to service and create further demand.

The challenge is that the public have already heard much more about the scary vision for AI’s impact on their jobs, and that it is started to take a grip. Evidence for this can be seen in a major survey of the British public, workers, young people, university students and employers undertaken by King’s College London.

What our survey showed
We tested opinions about Mr Amodei’s statement about entry level jobs against the World Economic Forum’s more positive prediction that AI will create twice as many jobs as it eliminates by 2030. There is a clear winner: 56 per cent of the people we surveyed think Amodei’s 2025 vision is right, only 25 per cent think the WEF’s is likely.

This deep concern runs throughout: seven in ten of the British public are worried about the economic impacts of AI, six in ten think it will eliminate more jobs than it creates and half think its impact will be worse than a normal recession. One in five even think it will be bad enough to create civil unrest.

And this negative mindset bleeds into general suspicions about AI. Half the public say they would rather completely avoid technologies that are based on AI, and four in ten think it has more disadvantages than advantages. Or they say they are actually just afraid of it. Only a quarter think it will improve their own life or that it is a positive for humanity as a whole.

Worried for themselves, and each other
This negative balance of opinion won’t only be about the jobs impact. There is significant suspicion about what it’s doing to our ability to think for ourselves. As we often find in surveys, the public have a bias towards thinking they themselves will be fine, and it’s others that will be negatively impacted: 60 per cent agree that AI tools are worsening other people’s critical thinking, while only 27 per cent say the same about themselves.

We are quite comfortable thinking we are using AI extremely thoughtfully, while it’s simultaneously making the average person intellectually dependent. But this “I’m better than average” gap is actually not the most striking finding here. It is very unusual for a quarter of the public to admit that they themselves are being dumbed down by such a new technology, and is a clear signal of worry.

The public, workers, students and even employers also have deep equity concerns. Majorities of each group think any benefits from AI will mainly go to wealthy investors and large companies, and six in ten employers – who are often the most cheerleading group – think AI will increase inequality and polarisation in society.

Still time to change their minds
Before it’s even really started then, AI has a serious image problem, partly encouraged by its own leaders. Of course, this is not to downplay the reality of the risks – that it really does pose a threat from incredible disruption as jobs transition, and for equity, as wealth and power are concentrated. But, if we want to realise the benefits of the technology, it is also important that the public are not already set against it.

The good news is that a lot have not made up their minds yet: the public have not been paying nearly as much attention as those working in and around the sector, and around a third of people are still pretty neutral. But the window to influence views will close quickly, and it is much more difficult to switch people to positive engagement once they’ve set against something.

But there is some hope that this is still possible. For example, six in ten university students think educational institutions can prepare them for a job market shaped by AI – but only a third think their own university is currently delivering on that. People don’t think it is hopeless, we’re just not shifting fast enough to support people.

There’s a similar disconnect in communications and practice in workplaces. Eight in ten employers think they’ve seen productivity gains from AI already, and nine in ten think they have a good understanding of how their employees are using the technology. But six in ten workers say they’ve had no clear guidance on using AI, and only a third say it’s made them better at their jobs (11 per cent say it’s made them worse).

How to overcome the fear
We’ll need to do much more on both telling the positive story and providing real practical support which engages people. There is clear majority support for more training and help with transitions, but if these don’t materialise, public opinion will harden: people are already twice as likely to favour regulation and slowing down over innovation and growth, when the impacts may be fewer jobs and other harms.

Getting the balance right is a real and unenviable challenge, given so much is unknown and confident predictions from those closest to the technology keep changing. Some more positive messages, backed by actual action, is not boosterish gloss in this case, but rather a sensible balance to a growing pessimism that could become self-fulfilling. We need to communicate publicly on the best, while preparing, in less headline-grabbing ways, for the worst.

This blog is based on [url=https://www.kcl.ac.uk/policy-institute/assets/ai-and-the-future-of-work-digital.pdf]AI and the Future of Work Attitudes, beliefs and behaviour among the British public, workers, university students and employers[/url]
Pdf 93 Pages

Authors
Bobby Duffy is a co-director of the King’s Institute for Artificial Intelligence at King’s College London.
Elena Simperl is a co-director of the King’s Institute for Artificial Intelligence at King’s College London.

Note
This blog article is from The London School of Economics published on 7th August 2026. To read it with links to more information go HERE