The Digital Transformation Authority (DTA) is looking to dip its toes into the world of Generative AI, issuing a Request for Information (RFI) “to gain an understanding of the market.” The DTA is seeking to uncover Gen AI solutions that could “serve for individual Government organisations or the Australian Government as a whole.”

“Information is being sought to help understand the range of products and services that exist in relation to this rapidly evolving technology and how they may offer value for government,” the RFI states.

The DTA outlines three areas where it expects Government agencies could employ Generative AI solutions:

Customer-facing functions – e.g., using generative AI enabled chatbot or concierge services that help users navigate complex information and services in a dependable and consistent fashion.

Analytical or business intelligence functions – e.g., using generative AI to assist with the processing of large numbers of submissions or consultation responses, assisting in distilling complex information from a range of sources.

Core organisational and corporate functions – e.g., using generative AI to develop tailored learning and development experiences, streamlining complex or burdensome approval processes while maintaining and ensuring human responsibilities and accountabilities.

The RFI emphasises that the government wants ‘safe and responsible’ AI solutions, and underlines this with a series of questions that must be answered by those responding to the RFI:

• Can you describe or comment on how your AI solution ensures traceability and transparency of its generative AI outputs, such as how it can address the issue of incorrect information/'hallucinations' in its responses?
• Do you provide any assurances regarding intellectual property, such as for the source data or outputs of the generative AI solution?
• Does the generative AI solution comply with the Australian Privacy Act? (Do you see there being any particular risks that need to be managed / extra careful of?)
• Can you demonstrate how the foundation model(s)/LLM(s) used by your solution has been trained, the data that was used in training it, and the ethical sourcing of said data?

The DTA will not be creating a shortlist from those vendors who respond to the RFI, and admits it is also approaching other suppliers of GenAI technology to obtain information the same or similar to that requested by this RFI.