It’s been a big year for AI.

Early in January, Sam Altman declared that ‘we know how to build AGI’ and that 2025 would be a transformational year for AI’s capabilities and application. In the UK, the Prime Minister unveiled a major new AI strategy that he claims will help make the country one of the world’s pioneers in the technology. The release of Chinese LLM DeepSeek has provided organisations with a roadmap for building relatively low cost AI platforms. And analysis by CB insights reckons that progress is accelerating on Agentic AI, with 40% of US based developers now actively deploying their agents.

But what do the public think?

Last year, in Trajectory’s subscriber-only AI Citizen report we found that both knowledge of and interest in developments in AI were severely lacking.

In 2024, only 8% of UK adults had heard of LLMs, and only 42% of Generative AI. Attitudes to AI deployment generally were unfussed at best, hostile at worst, with only 22% thinking it would make society fairer and just 33% thinking it would make their own lives easier. Perhaps most significantly, the public still want humans making all the decisions.

We tested eight scenarios where a human or AI could make decisions, from setting the speed limits on smart motorways to managing hospital appointments and reaching a verdict in a criminal case. In each scenario, the public want humans to have responsibility, typically by a huge margin.

In our next Trends Briefing we’ll provide an exclusive update to our AI Citizen research to identify whether or not public attitudes to AI are moving with development or staying suspicious. We’ll also explore what’s driving public attitudes and the wider landscape of AI implementation, including the impact of tariffs, moves towards regulation and compliance frameworks and the likely role of China in any AI future.