The Reform thinktank has set out eight ideas – including a new pilot evaluation process and funding for scaling up projects – for using AI to increase productivity in government.
It has published a report, AI and the productivity revolution in government, which says deployment of the technology has been patchy and too reliant on individual public servants being willing to champion its use.
This is not due to funding as large language models (LLMs) are cheap to use off-the-shelf, there are plenty of areas in which LLMs or machine learning software could easily provide benefits, and plenty of guidance is available on the use of AI.
Instead, the report says: “The problem is mainly cultural. Government departments and public bodies are not set up to maximise the benefits of innovations like AI. There are few incentives to shift towards AI enabled delivery models, an aversion towards critical evaluation, limited access to specialist skills, and bureaucratic processes fail to quickly translate ideas into new products and tools.
“If AI pilots succeed they are not consistently scaled up; if AI pilots fail they are not adequately learned from.”
Attitude shift
It adds: “Across the public sector attitudes towards AI must shift. Leaders need to take charge of this shift, ensuring that innovations are widely adopted, and look afresh at what the functions of the state are and how AI can best enable them.
“Cross-functional teams collaborating across policy, operations and technology will be essential in embedding this shift and getting real world value out of AI.”
In order to achieve the cultural change it proposes eight ideas, beginning with a requirement on all government departments to report on plans to adopt AI as part of their outcome delivery plans.
Any pilot projects would be subject to a standardised evaluation process when they are complete to determine the lessons learned and whether they should be rolled out more widely, and there should be funding for the scaling up of successful pilots. This would be backed up by streamlining the procurement process to make it easier for government bodies to buy off-the-shelf products and for SMEs to provide their services to the sector.
These would be accompanied by a cross-government competition for identifying the best new ideas for using AI in Whitehall.
Pay and training
In response to the shortage of AI skills, Reform proposes an increase in pay allowances for relevant positions, and a standardised AI Essentials training course, covering the basic skills and ethical considerations for using AI, for Grade 6 and 7 civil servants.
In addition, legacy system updates and improvements in data quality and access should be made through iterative, agile processes alongside expanding the use of AI in frontline services.
