Three ways foundation models promise to put generative AI to work for governments

The Singapore Government is actively pursuing the use of generative AI, planning to create 100 generative AI solutions within 100 days and moving public sector chatbots onto large language models (LLMs). Foundation models, like LLMs, hold the potential to offer AI capabilities to governments and enterprises by being trained on large unlabelled datasets and adapted for specific tasks. This article discusses three key ways in which foundation models can utilize AI for the government sector:

  1. Improving Citizen Experiences: By utilizing generative AI and relevant domain knowledge, governments can build applications, such as conversational AI chatbots, to handle citizen inquiries effectively, enabling public sector agencies to focus on more value-added tasks.

  2. Raising Productivity, Transforming Operations: Foundation models can support government organizations in increasing productivity and transforming operations by sifting through and summarizing vast amounts of documents, applications, and files. This includes geospatial foundation models that can analyze satellite data for various purposes.

  3. Modernizing Applications: Generative AI can assist government agencies in modernizing their technology applications, helping upgrade legacy applications to run in a cloud-native way and generating automation code, thereby simplifying the workload of government IT teams.

The article emphasizes starting with identifying initial use cases and pilots to achieve quick gains and then scaling the AI strategy across the organization. IBM, among other tech companies, is actively working on training foundation models and providing workshops and pilots to help government agencies in implementing generative AI effectively.