How Politics Shapes UK Attitudes to AI
The artificial intelligence chapter of British Social Attitudes 42, published by NatCen and written by researchers at the Alan Turing Institute, the Ada Lovelace Institute and the Oxford Internet Institute. It uses a survey run on the NatCen Opinion Panel to test how political orientation shapes what people think about eight specific uses of AI. The headline finding is that support for regulation is high on every side of the political map, even where opinions about the technology itself diverge sharply. Free PDF.
What the report is
This is a chapter from _British Social Attitudes: The 42nd Report_, published by the National Centre for Social Research in July 2025. The authors are Tvesha Sippy and Aditi Dutta (Alan Turing Institute), Helen Margetts (Oxford Internet Institute and Alan Turing Institute), and Roshni Modhvadia, Oliver Bruff and Jorge Perez (Ada Lovelace Institute).
It draws on the second wave of the _How do people feel about AI?_ survey, run through the NatCen Opinion Panel with an unweighted base of roughly 2,500 respondents. Rather than asking about "AI" as a blob, it asks about eight named uses: cancer detection from scans, facial recognition in policing, large language models, loan repayment risk, robotic care assistants, welfare eligibility, mental health chatbots and driverless cars. The 2025 wave deliberately oversampled low-income, digitally excluded and minoritised ethnic groups.
What it finds
- Four uses carry a positive net benefit score: cancer detection, facial recognition in policing, large language models and loan risk. Four come out negative: robotic care assistants, welfare eligibility, mental health chatbots and driverless cars.
- 57%* of people from a Black minority ethnic group are concerned about facial recognition in policing, against *39% of the public overall.
- People with right-wing views record higher net benefit scores than those with left-wing views for nearly every application, driven less by seeing more benefit than by reporting less concern. On mental health chatbots, 36%* on the left and **35%** on the right see a benefit, but **68%** on the left are concerned against *59% on the right.
- 63%* of those with left-wing views worry facial recognition could produce false accusations, against **45%** on the right. Split the other way, it is **61%** of libertarians and *47% of authoritarians.
- Job loss worries split too: 62%* on the left fear it from robotic care assistants, against *44% on the right.
Why it matters
The regulation numbers are the story. 72%* say laws and regulations would make them more comfortable with AI, up from **62%** in 2023, at **76%** on the left and *70% on the right, with over half on both sides also wanting sign-off from a government regulator.
So the "public is split on AI" framing that gets recycled in policy coverage is mostly wrong. People differ on which uses they will tolerate. They broadly agree the thing needs rules. That is a useful stick to have to hand when an AI rollout is sold as inevitable.
Key takeaways
- Ask about specific uses, not "AI". The disagreement lives in the applications, especially policing and welfare.
- Demand for regulation crosses the political spectrum and has grown by ten points since 2023.
- Groups most likely to be on the sharp end of a system are the most worried about it, which is a data point, not a mood.
Photo by Yoksel 🌿 Zok on Unsplash
Download
Download the file· End of dispatch ·
About the correspondent
Dead Good Club
The House Curator brings together stories worth knowing about. Our editor selects and rounds up useful resources for the library, while summaries (usually AI-generated) offer a quick overview and helpful context. Every post links to the original article or video, so you can explore the full story for yourself. If you'd like to write for The Dead Good Club and see your work featured here, get in touch by email.
