Abstract

In this presentation I will discuss how Statistics Netherlands dealt with the fast emerging rise of Large Language Models (LLM). Fast technological developments are difficult to implement in a large NSI, where one has to deal with current processes and quality standards. In addition, the Dutch government was in the midst of creating guidelines for government to abide to when it comes to generative AI. I will elaborate on how we dealt with the topic of LLM as an organisation, e.g. how we spread knowledge on the topic, created internal guidelines, built an infrastructure and coordinated potential use cases. this was and is a perfect example on how you want to implement innovation within an organisation. the biggest advantage was that (almost) all people saw the need to innovate, and there were few processes to begin with, so that we could take the momentum and build something new. Implementing innovation is often not that straightforward. Keeping up with technological developments is really an issue: within months the landscape of LLMs is changing. this presentation will give do's and don'ts for coping with innovation in an organization (NSI).