Proposal Description

The paper will describe in detail UNCTAD's motivation, technical approach, and implementation hurdles in introducing its natural language interface to its Official Statistics database.

It will begin by contextualizing NSOs' motivation for and pressure to develop natural language tools, touching on such themes as changes in user expectations in the era of generative AI, as well as competition from common AI tools such as ChatGPT in representing the gateway to organizations' data.

It will then outline the philosophy and technical approach to the system's design, discussing decisions around scope and core functionality as well as technical and cost elements.

Finally, it will discuss experiences and learnings from actually implementing the system and rolling it out to users, as well as the way forward and future improvements.