Public service will manage risks of using AI while making the most of it

Forum Reply: Public service will manage risks of using AI while making the most of it

We appreciate the op-ed by Ms Ong Li Min and Mr Jason Grant Allen (ChatGPT will be a tricky puzzle piece in public service, July 6). Indeed, they spotted the problems and the potential of using Large Language Models applications (LLMs) like ChatGPT in the public sector.

To harness the benefits, we have started to provide our public officers with LLM-embedded tools, not only to boost productivity but also to learn more about potential applications in their line of work. We have set up rules for cyber and data security, and made it clear that every officer remains responsible for his work. We will also be extra careful with using LLMs for services that interact with the public.

It has only been about seven months since ChatGPT’s launch, and the technology is changing rapidly every week. We are committed to managing the risks while making the most of this innovative tool to help public servants serve Singapore better. For example, this reply was improved with help from Pair, our internal plugin built on GPT-3.5. But we are fully accountable and responsible for it.

Weng Wanyi
Director, National AI Office
Smart Nation and Digital Government Office
Prime Minister’s Office