Coming to terms with generative AI’s early potential for government

Following the recent international AI Summit, many government executives are assessing how to get started exploring the technology’s potential to improve citizen services 

One year after generative AI publicly emerged with the launch of ChatGPT and similar tools, it is clear the rapid evolution from AI to generative AI is opening major new frontiers for improving customer service, streamlining processes, and enhancing efficiency in government services.

The early stages of AI

The evolution of AI during the pandemic, especially to power chatbots serving massive increases in citizen service demands, paved the way for generative AI as a new type of government tool.

From the early large-scale implementations of virtual agents to alleviate demands on overwhelmed call centres, agencies experienced massive gains in their ability to serve citizens in need.

Generative AI marks a further big leap by eliminating the need for extensive training of virtual agents and enabling greater use of data to create personalized content and experiences for constituents.

While generative AI is beginning to open new possibilities in government, its use also has vast limitations. It needs the powers of human, real-world understanding and struggles with specialized, domain-specific topics. However, this should be considered in the transformative promise it holds for public service agencies, with current uses empowering call centres with more information and improved assistance capabilities.

Utalising generative AI

Navigating the complex web of government departments can often be daunting for constituents. Other early generative AI use cases target understanding of user intent and direct stakeholders to appropriate departments. AI-generated recommendations for services and queries regarding assistance programs can enhance government customer experience.

To begin to tap the power of generative AI, public service organizations should adopt a systematic approach. It is an emerging technology area, generally requiring expert-level testing and refinement to unlock its potential. Encompassing the need for intense due diligence and sensitivity around any deployment of generative AI, getting started involves three phases of activity: explore, experiment and execute.

The exploration phase involves defining the vision and identifying the most promising use cases. The experiment phase focuses on developing prototypes and assessing their impact. At the same time, the execute phase involves scaling up selected solutions and may also depend on related strategic priorities such as workforce training and public communications.

Starting small and thinking big is the key to successful government exploration of generative AI. By breaking down service processes into manageable phases, public service agencies can identify relevant use cases and pursue responsible deployment of generative AI solutions. Moving forward with any deployment rests on careful preparation, including organizational leadership’s ultimate responsibility for assessing readiness and measures of adoption and implementation.

Public services using generative AI

Implementing generative AI in public service requires intense focus on responsible AI controls, data readiness, and, likely, workforce upskilling. To help meet such needs, government entities should consider establishing or tapping a generative AI “center of excellence” to help accelerate deployment and minimize risks. Such centres facilitate collaboration with experts and learning from other industries’ experiences, offering insights and guidance throughout the implementation journey.

Generative AI is showing early promise to revolutionize public services through some organizational processes and activities, including better-serving government customer needs. While many concerns, challenges and limitations exist, early use cases are showing generative AI encompasses a new set of tools for government, warranting a systematic approach to assess value and manage experimentation and deployments as forward-looking leaders seek to begin using generative AI to expand and improve public services.

This piece was written and provided by Eyal Darmon and Avik Batra. Eyal Darmon leads Accenture public service customer experience and generative AI work in North America and Avik Batra leads public service customer experience for Accenture Song in North America.