TL;DR

The Ada Lovelace Institute’s year-end analysis reveals a growing gap between AI progress narratives and public sentiment. With 84% of UK citizens believing government favours tech companies over public interest, and the promised AI Bill now stalled, questions mount about whether current approaches serve society.

The Trust Problem

Research shows 97% of the UK public have heard about AI, but familiarity hasn’t bred comfort. A striking 89% believe safety should take priority over speed, even if this means slowing development. Meanwhile, 74% believe the ability to ban AI on ethical grounds matters more than competitive advantages.

Yet despite these clear public preferences, the government has signalled no intention to proceed with the AI Bill promised in Labour’s manifesto. This follows a global trend of AI deregulation that puts people at risk of harm from systems increasingly embedded in daily life.

Real-World Deployment Concerns

Emerging research on AI transcription tools in social work reveals concerning patterns. Local authorities are adopting a “try now, test later” approach, with significant variation in oversight and risk management. What one worker considers appropriate human review might take five minutes; another’s might take an hour.

The findings highlight how hallucinations—fabricated references and assertions inherent to generative AI—could affect official records and statutory judgements. An LSE study found one widely used model downplays women’s physical and mental health issues compared to men’s, potentially affecting care provision.

Looking Forward

The Institute calls for AI to be held to the same standards as consequential sectors like food, medicine, and aviation. This means independent evaluation and audit, robust regulation, and inclusive democratic engagement in shaping technology’s role in society. Without these safeguards, the gap between AI promises and public benefit may continue to widen.