AI Tool 'Consult' Accelerates UK Government Feedback Analysis Amid Calls for Caution

AI Tool Speeds Up Government Feedback Analysis, Experts Urge Caution

An AI assistant named Consult has recently been trialed by the Scottish Government to handle public feedback on regulating non-surgical cosmetic procedures. It processed over 2,000 responses, identifying main themes and concerns using generative AI. The results closely matched human analysis, allowing officials to focus more on interpreting the findings rather than sorting data.

UK Technology Secretary Peter Kyle highlighted the efficiency gains: “No one should be wasting time on something AI can do quicker and better, let alone wasting millions of taxpayer pounds on outsourcing such work to contractors.” The aim is to extend Consult’s use across government departments to support the government’s ‘Plan for Change’ while cutting costs.

The current system relies heavily on expensive external contractors for feedback analysis. Consult, part of a broader AI toolkit called Humphrey, is intended to free civil servants from routine tasks and reduce contractor expenses. Officials double-checked and refined the AI-generated themes to ensure accuracy. This collaboration between AI and humans preserved quality while saving time.

Trial Results and Benefits

• Consult’s analysis aligned closely with human reviewers, with minor differences that did not affect overall conclusions.
• The AI sorted responses into categories, giving officials more time to focus on policy implications.
• Civil servants found Consult’s work a useful starting point and noted it helped reduce bias by providing consistent analysis.
• The Scottish Government manually reviewed all responses during the trial phase to confirm reliability.

Public Health Minister Jenni Minto praised the tool for speeding up understanding of public views and focusing on policy questions more quickly. If fully adopted, Consult could save the government approximately 75,000 days of analysis annually, given that the UK conducts around 500 consultations each year.

Maintaining Human Oversight

Consult is designed to assist, not replace, human experts. Officials remain involved in reviewing AI-suggested themes and sorting of responses through an interactive dashboard. This ensures transparency and allows for adjustments based on expert judgment.

Handling tens or hundreds of thousands of responses will soon become more manageable with Consult, but the human role in interpretation and decision-making stays central.

Expert Warnings on AI Use in Government

While the benefits are clear, experts emphasize caution. Stuart Harvey, CEO of Datactics, points out that AI effectiveness depends heavily on the quality of underlying data. Poor data management risks biased or unreliable outcomes. Transparency, strong governance, and ethical standards are essential for building public trust.

Professor Michael Rovatsos from the University of Edinburgh agrees, noting risks of AI bias and the potential for manipulation. He calls for strict safeguards and ongoing investment to maintain reliability and fairness in government AI tools.

Stuart Munton, Chief for Group Operations at AND Digital, stresses the importance of keeping people at the core of AI adoption. Investing in training and empowering diverse public sector teams will improve outcomes and ensure AI tools meet real-world needs.

The Road Ahead

The UK government plans to expand Consult’s use across departments before the end of 2025. This move aligns with efforts to create a leaner, more responsive public service that balances cost savings with quality engagement.

For government professionals interested in how AI tools like Consult can support public engagement and policy development, exploring training on AI applications in public service could be valuable. Resources such as Complete AI Training’s courses for government roles offer practical knowledge on integrating AI responsibly.