Generative AI and public sector impact
Generative AI is currently the most disruptive technology, with government leaders recognizing its transformative potential. However, scaling AI beyond small proof of concepts (PoC) remains challenging for government institutions. Unlike private industry, government institutions need a distinct approach to scaling AI applications due to different incentives and risks. Some government leaders are starting to make progress with employee-driven initiatives.
An employee-driven strategy requires each employee to have the necessary AI fluency, with training levels varying by job, level, and role. Additionally, AI investments should balance implementation costs against public benefits. This highlights two challenges for government leaders: the lack of technical expertise can obscure the true costs of AI, and leaders need methods to measure the challenging-to-quantify mission outcomes of AI investments.
EY study, conducted with Oxford Economics, surveyed nearly 500 senior government executives and held in-depth discussions with 46 government leaders and AI experts across 14 countries (Australia, Canada, Denmark, Estonia, France, Germany, Netherlands, Italy, Japan, Singapore, South Korea, United Arab Emirates, United Kingdom, Unites States of America). The survey reveals that despite widespread recognition of AI’s potential in government, although 64% believe AI can save costs and 63% think it can enhance services, its implementation in government is still limited. Only 26% have deployed AI in parts or fully across their organization, and just 12% have implemented generative AI (GenAI) solutions.
In Latin America, Chile’s National Center for Artificial Intelligence (CENIA) and the Economic Commission for Latin America and the Caribbean (ECLAC) presented the results of the 2nd edition of the Latin American Artificial Intelligence Index (ILIA 2024). The study evaluated the AI readiness of 19 countries in the region. Chile ranked first with 73.07 points, followed by Brazil (69.30) and Uruguay (64.98). Other countries, such as Argentina, Colombia, and Mexico, were categorized as “adopters.” Despite doubling the AI talent concentration in the workforce over the past eight years, no country in the region has reached the levels of the Global North. Chile, Brazil, and Uruguay are leading in AI implementation and are focusing on expanding these technologies across various sectors, mainly private sector.
In Latin America, Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Uruguay have national AI strategies. These strategies emphasize key themes such as ethics, AI governance, adoption, cross-sector collaboration, procurement, skills, capacities, data, and technical infrastructure. While these represent significant progress, they vary in terms of action items and enablers. Most strategies have specified actions and objectives, but many lack time frames, funding mechanisms, and monitoring instruments.
Even developed economies face challenges in implementing AI for public services. As shown on EY research, despite recognizing its transformative potential, they encounter barriers such as privacy and security concerns (62%), lack of strategic alignment (51%), weak business cases (41%), inadequate infrastructure (45%), and ethical considerations (42%). Addressing these interconnected challenges across strategic, technical, and organizational dimensions is essential for successful AI implementation.
A recent Deloitte survey revealed that only one-sixth of government leaders believe their organization has high or very high generative AI expertise, compared to 32% to 56% in the private sector. This expertise gap complicates technical decisions around model selection, cost, and data management. While private sector interest in generative AI is often driven by executives, government workers are more eager to adopt it, though leaders remain cautious about risks. Despite some bottom-up interest, only 1% of government leaders reported that over 60% of their workers have access to generative AI, a significantly lower figure than in the private sector.
AI has the potential to significantly impact public good. However, for government institutions to make a meaningful impact, they need to scale AI adoption. This creates a paradox: widespread AI use is necessary for impact, but it also introduces risks that need to be managed.
Unlike private enterprises where leadership drives AI scaling, government employees may uncover transformative AI use cases. However, they need access to appropriate tools and training for their roles. Access to generative AI remains limited, as shown on Deloitte’s survey, with only 1% of government respondents reporting that 60% of their workforce has access to these tools. This is slowly changing as more governments provide wider access to generative AI tools.
Understanding how generative AI can create value for a government institution is crucial. To do it, they need reskilled people, to help identify where to look for results and which metrics to use for measurement. The pathway to value determines the necessary metrics for success and the levels of governance required. When AI automates single tasks, it creates value by increasing efficiency, measured by time and money saved. When AI is integrated into larger workflows, it improves overall performance, measured by mission outcomes such as reducing crime, enhancing benefits delivery, or increasing community longevity.
Navigating the rapidly changing AI landscape can be challenging for governments. Partnering with technical experts is crucial for laying the technical foundations and avoiding vendor lock-in. This collaboration benefits both parties: technical partners help government leaders gain AI fluency, while the government shapes the tech industry’s approach to solving problems, creating demand for solutions like sovereign AI to address security and privacy concerns.
Overall, we could summarize the key pathways to government institutions successfully implement generative AI to support public services:
· Data and AI literacy — they need to reskill their people to understand where generative ai could really bring value to government.
· Technology and Process review — They need to enjoy the partnership with technical experts to stablish the better technical architecture to prevent risks, be scalable and reduce costs.
Bibliography:
The transformative power of Data and AI — https://www.ey.com/en_nl/insights/government-public-sector/how-data-analytics-and-ai-in-government-can-drive-greater-public-value
The big risk of doing nothing — UK Government — https://www.gov.uk/government/publications/ai-in-schools-and-further-education-findings-from-early-adopters/the-biggest-risk-is-doing-nothing-insights-from-early-adopters-of-artificial-intelligence-in-schools-and-further-education-colleges
Delivering on the promise of AI in government — https://www.deloitte.com/us/en/insights/industry/government-public-sector-services/government-trends/2025/scaling-ai-in-government.html
Latin American Artificial Intelligence Index (ILIA) Reconfirms Chile, Brazil and Uruguay as Leaders in the Region — https://www.cepal.org/en/pressreleases/latin-american-artificial-intelligence-index-ilia-reconfirms-chile-brazil-and-uruguay
