An artificial intelligence tool named "Consult" has been trialed by the Scottish government to analyze public feedback on the regulation of non-surgical cosmetic treatments like botox and fillers, marking the first use of such technology for a government consultation in the UK. The tool processed around 2,000 responses, identifying common themes and categorizing feedback, with human experts verifying its accuracy at each stage. This approach aimed to streamline the labor-intensive task of sifting through large volumes of public opinion, which often results in tens of thousands of submissions during consultations.

Consult is part of a broader suite of government AI tools known as "Humphrey," designed to enhance the efficiency of civil servants and reduce administrative costs. The AI’s performance was closely matched to that of human analysts, with negligible differences in how themes were ranked, and officials noted that the tool provided a useful starting point for deeper policy analysis. The government estimates that widespread adoption of such AI technology could save up to £20 million annually by reducing the need for manual review and outsourcing to contractors.

One of the key advantages highlighted by officials is the AI’s ability to minimize individual analyst bias, as it systematically processes responses and reduces opportunities for personal preconceptions to influence results. Additionally, Consult was tested for its capacity to handle responses with spelling errors and other language issues, though it currently operates only in English and requires translation for other UK languages. Human oversight remains integral, as experts review the AI’s thematic sorting and ensure the quality of analysis through an interactive dashboard.

Despite promising results, Consult is still in its trial phase, with further evaluation planned before broader deployment. Officials have addressed concerns about AI "hallucinations"-the generation of inaccurate information-by limiting the tool’s scope to a narrowly defined task, thereby reducing risks associated with erroneous outputs. The government sees this as a significant step toward a more agile and responsive public sector, capable of efficiently gathering and analyzing public sentiment on important policy issues.