What the UK public wants AI companies to know: 14 key themes based on 1K+ conversations

The rapid evolution of artificial intelligence has people around the United Kingdom bracing for an unknown future, uncertainty that has coupled increasing adoption with scrutiny to match.

To find out what the public really thinks about AI, YouGov deployed the technology powering BrandIndex Voices to interview 1,200+ UK adults.

The prompt

Imagine you had the chance to speak directly to the leaders of the world’s biggest AI companies... What would you want to say to them about AI?

YouGov asked UK adults what they want AI companies to know, revealing a population conflicted with what exactly they may be giving up in exchange for productivity.

The population studied in this research is not nationally representative, but instead reflects the views of an audience that opted-in to chatting with YouGov's AI interviewer.

The AI interviewer and analyser uncovered that while people in the UK see AI as useful for getting everyday work done, trust, control and misuse worries weigh heavily in the background.

The research uncovered 14 themes that the UK public wants AI companies to know. View direct interview quotes pertaining to each one below:

Jump directly to each theme by clicking the links below, or scroll to continue reading.

1. Everyday assistance and productivity
2. Information reliability and bias
3. Ethics, fairness and social impact
4. AI misuse and online harms
5. Education skills and public understanding
6. Human oversight and moral limits
7. Safety and catastrophic risk
8. Employment and economic disruption
9. Regulation, governance and corporate power
10. Healthcare, science and public services
11. Privacy, data use and surveillance
12. Human connection and mental wellbeing
13. Creative work, culture and ownership
14. Environmental impact and resources

What the UK public really thinks about AI

UK adults are not rejecting AI outright, but they are approaching the technology with caution and concern. While they the public is willing to use AI when it clearly saves time or supports human work, they say they want stronger proof, stricter oversight, and real protections before trusting it more completely.

1) Everyday assistance and productivity 

• AI-generated topic description: AI is treated as a practical assistant that supports routine tasks and knowledge work, improving convenience and efficiency while humans retain final responsibility and judgment.
• Human analysis: More than half of respondents (53%) mentioned productivity in their interviews, the most frequent of any topic surfaced. With only two of the 14 topics generating engagement scores higher than 6.0, the fact that this topic was positive (6.3 engagement) and the most mentioned suggests it’s the primary benefit that people in the UK who use AI view as a worthwhile tradeoff for what they otherwise seem to see as downsides.
• "Time is the key factor. I can do all these and have done them the traditional way but it takes time. I think of AI as a calculator. I can add numbers by hand but it is easier in a calculator. The value in my work is in the end product. Not by me doing the mundane things that help me get there,” said a man based in the North East aged 25-34.

2) Information reliability and bias

• AI-generated topic description: Can AI generated information be trusted, focusing on factual accuracy, consistency, misinformation, deepfakes, hidden biases and the need for transparency and critical evaluation by users.
• Human analysis: The second most-mentioned topic, information reliability and bias combines a negative sentiment (3.5 of 10) with a high engagement (7.4 of 10), suggesting a big issue that matters greatly to many people. If the topic had a low sentiment with a low engagement, it would be a nuisance, but less of priority to address. AI companies that promise time-savings may be reneging on those promises when users find hallucinations and require additional time editing and fact-checking.
• “AI is not always accurate, and regardless of how sophisticated it may seem, they are not human and do not appreciate nuances.  We need to be cautious about our reliance on AI,” said a North West woman aged 55+.

3) Ethics, fairness, and social impact

• AI-generated topic description: Aligning AI with human centred values and social priorities, including fairness, inclusion, distribution of harms and benefits and expectations that AI should support core social goods and cohesion.
• Human analysis: Another topic with a lower sentiment (3.9) but high engagement (7.5) – there are many of these - “Ethics, fairness and social impact” causes slightly less detraction than overall reliability but is among the highest emotional triggers for people in the UK. Many hope, quite cautiously, that AI can make society more equal, rather than exacerbate existing inequities.
• “I foresee greed, war, market corruption and collapse in society as the likely outcome. This would be such a huge shame when AI could do so much good. It more than likely will do a huge amount of good in healthcare for instance, but at a price. Robotics and AI are already aiding governments in the application of War at very little cost. Manipulation of social media and news is getting very sophisticated. It will all end in chaos,” said a North West man aged 55+.

4) AI misuse and online harms

• AI-generated topic description: Harmful uses of AI such as fraud, scams, crime, deepfakes and abusive or manipulative content that create risks for individuals and communities, especially vulnerable and young users.
• Human analysis: While mentioned fewer times overall, a larger share of UK residents interviewed mentioned concerns about AI misuse and harms (32%) than did ethics, fairness and social impact (26%). This suggests a widespread belief that bad actors could leverage AI against the public. With a 7.7 of 10 engagement and just 2.6 of 10 sentiment, this was among the most emotionally-charged topics surfaced throughout YouGov’s interviews, reflecting a deep fear and respect of the technology’s purported power.
• “It could always be used to impersonate someone, for example a child’s voice could be used to create a false alert for a parent. A person of trust could be mimicked quite easily. I’m also concerned about the potential loss of control,” said a South East man aged 35-44.

5) Education skills and public understanding

• AI-generated topic description: How people learn about, understand and work alongside AI, including needs for education and training, risks of overreliance for thinking and learning and ensuring systems are usable and inclusive for diverse groups.
• Human analysis: With a relatively neutral sentiment score (4.7) but high engagement score (7.4), the 27% of people in the UK who expressed commentary about upskilling the public around AI recognise the urgency for adapting to the technology but have not yet come to a consensus about whether that’s good or bad. Some recognise the tool as enabling additional learning, while others express concern that it decreases cognitive ability. Compared to topic #8, employment and economic disruption, which has a much more negative sentiment, education and learning feel like actions people can proactively take in the face of possibly inevitable change.
• “Education could use AI to teach a whole host of subjects, extra tuition for those struggling and those who want to study extra subjects,” said a woman in Wales aged 45-54.
• “It shouldn't be the default option, you should think first. Learning to craft your own thoughts is key to being an active, engaged citizen. AI makes us passive,” said another woman in Wales aged 55+.

6) Human oversight and moral limits

• AI-generated topic description: Emphasis that AI should remain under human authority, especially in sensitive or high stakes contexts, highlighting limits of AI judgment and the need for clear boundaries and accountability for decisions.
• Human analysis: A near-neutral topic with a negative lean (4.3 sentiment out of 10), interviewees conveyed a preference for a phrase business people have used throughout the AI adoption cycle: Human In The Loop. This oversight could address some of the perceived shortcomings of AI, like information reliability and bias, ethics and misuse.
• “It needs to be presented and treated as a machine, not as a person. The anthropomorphising designs of AI assistants encourage people to trust the data implicitly, rather than treating it as a tool, and treating the AI as a person makes it easier to cut out your colleagues,” said a South West woman aged 25-34.

7) Safety and catastrophic risk

• AI-generated topic description: The potential for AI to cause serious or large scale harm, including loss of human control, weaponisation, military use and long term catastrophic or existential risks to humanity.
• Human analysis: More than a quarter of interviewees (26%) expresesed trepidation toward a catastrophic ending facilitated by AI technology. This topic rated among the lowest sentiment (2.9 out of 10) of any measured. With 1,263 interviews condcuted and 1,964 mentions of the topic, that means those who broached the subject discussed it for half of more of their interviews, covering six touches of the maximum of 12-turn conversation.
• “I enjoy drinking water, but I recognise that a tsunami hurtling for my house is probably a bad thing. It's about scale, intensity, and control. AI deployed in a safe manner could be one of the greatest achievements in human history. If deployed in an unsafe manner? Potentially apocalyptic,” said a man in Scotland aged 25-34.

8) Employment and economic disruption

• AI-generated topic description: How AI is transforming work and the economy, including automation of tasks, job displacement, changing skill demands and effects on income distribution and labour markets.
• Human analysis: Nearly one-third of people in the UK (31%) cite AI's impact on employment and its consequences as a major concern in interviews. Clocking a sentiment of just 2.9 out of 10, conversations around job security tied for the second highest rates of negativity of all topics mentioned. At the scale interviewees mentioned the topic, the economic impact of AI may be the most immediately alarming concern about artificial intelligence.
• “We're on the precipice of the biggest upheaval in human labour markets probably ever. Jobs will be lost in swathes. Advocate, lobby government and industry to introduce a radical new approach to economics. Things like UBI are a necessity to prevent exponential wealth hoarding and mass civil unrest,” said a man in Scotland aged 25-34.

9) Regulation, governance, and corporate power

• AI-generated topic description: Who controls AI and how it should be governed through laws, oversight and organisational practices, including the influence of large companies over data, access, risks and long term development directions.
• Human analysis: More than a quarter of interviewees (27%) discussed with YouGov’s AI interviewer the concept of AI governance, whether that’s through governmental mandate, corporate responsibility or another approach. The topic tied for third-most engaging (7.5 out of 10) of the 14 surfaced, meaning that oversight of AI engrossed respondents who decided it was worth mentioning.
• “Ultimately I think AI is too powerful to be in the hands of a wealthy few. It's the sort of thing that should be in the hands of government or a strictly regulated independent non-profit. I believe this is a technology on par with nuclear weapons in terms of its destructive potential - there's no way we'd allow private companies to have nukes.” said a man in Scotland aged 25-34.

10) Healthcare, science, and public services

• AI-generated topic description: AI deployment in medicine, scientific research and public services, where systems are expected to support expert decision making, improve efficiency and help address complex societal challenges.
• Human analysis: The other major AI topic with a positive spin (beyond productivity boosts), 22% of UK adults communicated optimism around how AI could impact society across healthcare, science and public services. While the engagement around this topic was positive (6.2 out of 10), it did register lower than every other topic mentioned, suggesting that, while UK adults recognise the possible benefits about AI technology, their anxieties and fears could be overpowering their hopes.
• “I am hoping it can help medical professionals to diagnose conditions sooner. I can see it's potential in removing barriers for people with disabilities and long term health conditions. However, I fear it can and will be used maliciously, for example, deep fakes, relying on fake or biased data and producing unreliable or biased results,” said a West Midlands woman, aged 35-44.

11) Privacy, data use and surveillance

• AI-generated topic description: How AI systems collect, store and use personal data, including risks of leaks, hacking and surveillance and concerns about boundaries, ownership and civil liberties.
• Human analysis: Hardly unique to artificial intelligence, the topic of privacy and data use appeared in 18% of conversations, clocking an engagement score of 7.1 and sentiment score of 3.2. Like many of the topics on this list, people in the UK feel very negative about the intersection of AI and privacy. It’s not the most important or immediate theme surfaced by the analysis, but remains something that impacts public perceptions.
• “We’ve already seen how social media can manipulate people into believing political campaigns and prey on fears and bigotry, and AI has the opportunity to do this on steroids,” said one London woman, aged 45-54.

12) Human connection and mental wellbeing

• AI-generated topic description: How AI mediated interaction affects social connection and mental health, including its use as a substitute for human contact, risks of dependence and parasocial relationships and preferences for human support.
• Human analysis: A relatively less important topic, tied for the lowest engagement among topics with negative scores, 16% of interviewees did mention concern about AI’s impact on human connection and mental health. While the engagement score (7.0 of 10) fell slightly behind other topics, it notably exceeded both positive topics, perpetuating the motif that negative emotions around AI appear to have a much stronger hold on the UK public than positive ones. With that said, some respondents had positive thoughts about AI’s influence on mental wellbeing.
• “There are some personal issues I like to talk about that I sometimes don't feel comfortable talking about with anyone. Talking with AI makes me able to express myself freely and has helped me emotionally,” said a London woman aged 35-44.

13) Creative work, culture, and ownership

• AI-generated topic description: AI in artistic and cultural production, considering its impact on creative labour and livelihoods, questions of authorship, consent and compensation, and the place of human creativity in AI saturated media environments.
• Human analysis: Of all of the topics surfaced, intellectual property rights arguments around AI have the highest engagement (7.9 out of 10) and lowest sentiment (2.5 out of 10). However, this topic arose with the fewest number of interviewees, with just 10% expressing attitudes about the ownership of creative work. For a small minority of people inthe UK, IP is a deeply important topic, possibly the most important topic, as it informs their identities and shapes their livelihoods.
• "A lot of writers/artists I know are barely scraping by, and it's enraging to see AI companies scraping their intellectual property without permission or compensation,” said a South East man aged 55+.

14) Environmental impact and resources

• AI-generated topic description: The environmental footprint and resource demands of AI, including energy, water and material use for data centres and infrastructure and how ecological limits should shape AI development.
• Human analysis: A major public criticism of artificial intelligence, only 15% of interviewees mentioned environmental impact in their interviews, among the lowest of any theme. While the sentiment tow AI’s impact on the environment was negative (3.1 out of 10) its measured engagement was also lower than most other topics. This suggests that, while most overlook sustainability in AI concerns, those who recognise it hold strong convictions.
• “Governments should encourage innovation, but not support the blind pushing ahead that we've seen in this developing field of technology. The environmental impact of the data centres alone is unsustainable. Governments should take care not to support greed masquerading as true innovation,” said a South East woman aged 45-54.

Understanding the interview methodology

Members of YouGov’s consumer research panel of 3+ million adults in the United Kingdom completed conversations with the AI interviewer between April 15 and April 30, 2026. Each interview started with this prompt:

<sub>Hello, I am YouGov's AI-interviewer! YouGov is publishing a study about public attitudes towards AI, and we'd like to hear from you!

Imagine you had the chance to speak directly to the leaders of the world’s biggest AI companies... What would you want to say to them about AI?

You might like to think about how you feel about AI, what you would like AI to do for you, your experiences using it, or your hopes and concerns for the future.</sub>

Conversations lasted no longer than 12 questions. The AI interviewer adapted each respondent’s follow-up questions based on their individual answers, meaning each conversation was unique, outside of the opening prompt.

Because YouGov manages this process end-to-end, quotes and analyses are quickly traceable back to their original transcripts.

Quotes have been amended to remove spelling mistakes and improve readability.