Transforming Public Sector Services using Generative AI

with Robyn Scott

Robyn Scott, CEO and Co-founder of Apolitical, uses examples from around the world to explore the potential for generative AI to transform government operations.

Following the publication of the new report ‘Transforming Public Sector Services using Generative AI: Global Case Studies’, from Apolitical and Microsoft, Robyn Scott, the co-founder and CEO of Apolitical joined Public Sector Future to explore the findings.

Scott explained that Apolitical is a network and learning platform exclusively for the public sector, used by over 200,000 public servants and policymakers in 160 different countries. Their mission is to build 21st-century governments that work for people and the planet, with a focus on challenges such as AI.

Rapid increases in public servants exploring generative AI

Scott shared that the Apolitical team are seeing enormous interest in leveraging AI and regulating it, with public servants being more hopeful and excited about AI than fearful of it. Scott shared that among Apolitical’s civil service members, there is a “balance towards optimism”. When they were polled in 2023, 56% of members were hopeful in regard to its potential.

Scott characterized the adoption of generative AI in governments as being distinguished by both a top-down and a bottom-up dynamic. “In a top-down sense, it is more geared towards how to regulate and control AI. The bottom-up adoption is very much about how AI can be used as a useful technology to handle challenging and sometimes overwhelming jobs better”. She also highlighted the steep adoption curve for generative AI. As early as March 2023, 39% of members polled by Apolitical were already using it in their work, but that had jumped to 54% by July, with 22% of their members using it for research and 19% to help with writing.

Case studies from around the globe

The Report highlights three case studies of civil servants using generative AI tools in their organizations, covering different regions of the world and levels of government. Scott shared a flavor of these case studies:

• Portugal, through its Administrative Modernization Agency, created a virtual assistant to help the public navigate their digital authentication system. The avatar’s answers rose to 90% accuracy in just the first two weeks through iterating and testing. They have plans to develop the avatar to speak between 15 and 20 languages, which will be used to help new citizens who might not speak Portuguese or English fluently.
• The Tokyo Metropolitan Government created its own private version of ChatGPT with the goal of increasing internal productivity and creativity for their team. They started using it to reduce clerical work and help with tasks like idea and text generation. In a recent ideathon, their digital services team identified 200 use cases that the tool could help them with.
• The City of Kelowna in Canada created a generative AI-powered chatbot to accelerate the permitting process for new homes and apartments. Kelowna is one of the fastest-growing cities in Canada and is facing a significant housing shortage. By fielding thousands of permitting questions, the chatbot is tackling backlogs and reducing waiting times, which has many positive implications for the economy and people’s lives.

Success in implementation

Reviewing these examples, Scott identified several common factors in the success of these teams, including:

• “A drive for innovation and curious yet cautious approach to testing the capabilities of generative AI.”
• The importance of collaboration, “they all thought about clear communication with their different stakeholders, involving diverse voices, regular meetings, shared sense of purpose.”
• “Transparency, acknowledging risk, tackling them early on in the projects”, and using this to further build trust.

Scott emphasized the importance of understanding the risks of data input and governance, and the appropriate use of humans in the loop. She also highlighted the need for the right kind of enabling governance, with international cooperation, jurisdictional interoperability, and inclusive governance.

Addressing an asymmetry of understanding

Reflecting on the capability and skilling requirements, Scott shared the need to build understanding of generative AI tools across all levels of organizations: “you’ve got an asymmetry where you’ve got this bottom-up adoption and capability and ease with younger generations. And then you’ve got people at the leadership level who are much less comfortable.”

She expanded “I gave the nearly 60% figure across the wider workforce [for those who] are using it or have used it in some way. Anecdotally, it seems that at the leadership level, senior leadership, it’s more like 20%. And anyone who’s played with generative AI knows that you can’t really understand it in the abstract. And I think if you treat it only as something to be understood in the abstract, it’s much easier for it to become adversarial. Whereas if you’re using it and harnessing it, you understand it, and you also see its benefits.”

Discovering future trends and opportunities

Sharing the optimism of public servants on the potential benefits of generative AI, Scott explained how she expects to see opportunities develop in government operations, policy making, and at the citizen interface level. She believes that generative AI has the potential to move digital services in governments from being efficient and good, to being genuinely smart. 

“I like its application potential to policy making. When you’re making policy, you’ve got to think through so many use cases in a complex system, and brainstorming with a generative AI tool as your partner can be really good at that if you haven’t done it, I would recommend doing it.”

In addition to the direct benefits, Scott shares her perspective that “it’s also really exciting from a second order benefits perspective. Because one of the things I think I will do that’s not talked about that much, is the speed of change that it is necessitating is going to potentially change the way that governments operate if they adopt it.”

“We saw the potential for that with COVID. Whole governments started operating a whole lot faster. They embraced digital a whole lot more. But in many cases, that tide has receded and there’s been a lot of reversion to the mean, and actually, public servants are quite frustrated with that reversion to the mean in many cases.”

“Generative AI is a tide that’s not going to recede. So I think you’re going to get a new precedent set for the speed at which things can be done and improved in government, and that will actually improve government overall, improve its muscle for agility.”

To find out more:

• Read the report

Listen to this episode on any of these podcast platforms:

Spotify

Apple Podcasts

Google Podcasts

Deezer

Libsyn