The Ada Lovelace Institute has recently reported on foundation models in the public sector and offers an in-depth look at the diverse formal and informal applications of AI tools like ChatGPT.

It has a broad set of recommendations that are helpful for public sector leaders. The section on “proposed use cases” is particularly worth reviewing.

Overall the report explores AI tools potential for enhancing public services in various areas, including document analysis, policy draft review, public enquiry management, and knowledge consolidation.

Foundation models are a form of artificial intelligence (AI) designed to produce a wide variety of outputs that are being adopted across various sectors. There is some optimism in policy, public sector and industry settings about the potential for these models to enhance public services by:

• automating review of complex contracts and case files (document analysis)
• catching errors and biases in policy drafts (decision support)
• powering real-time chatbots for public enquiries (improvements in management of public enquiries)
• consolidating knowledge spread across databases into memos (knowledge management).

However, the report also addresses the inherent risks associated with these technologies, such as biases, privacy breaches, misinformation, and security threats. Aimed at policymakers and public-sector leaders, the report provides guidelines for the responsible and beneficial use of foundation models in government settings.

More information on this can be found in the full report.