7. Conclusion and recommendations
As seen throughout this report, governments in the LAC region have made tremendous progress in strategizing for and experimenting with AI in the public sector. In order to ensure that these efforts are informed, trustworthy and enhance public value, the right factors and capacities must be in place to ensure a strong foundation on which to build out AI efforts and ambitions. In particular, governments need to:
Develop a responsible, trustworthy and human-centric approach, including through data ethics, ensuring fairness and mitigating bias, providing for transparency and the explainability of algorithms, promoting safety and security, putting in place accountability mechanisms, and ensuring an inclusive and user-centred approach.
Build key governance capacities, including through leading, co-ordinating and building support for AI; data leadership and strategy; creating space for experimentation; understanding public sector problems and the potential for AI solutions, and ensuring future preparedness through anticipatory governance.
Put in place key enablers, including data, funding, internal and external expertise, and digital infrastructure.
The volume of considerations that public officials must take into account may seem overwhelming. However, governments around the world and in the LAC region have devised approaches to addressing them in their own context. While countries in the region vary significantly in their current capacities for AI in the public sector, and in their digital maturity more generally, the potential for adoption of AI for public sector innovation and transformation remains significant across the board. As with other regions around the world, opportunities also exist to create a united regional approach to AI, with each country making contributions based on their own comparative strengths.
To help governments in the region seize this potential, this report makes 13 key recommendations.
Recommendations for adopting a strategic approach to AI in the public sector
In order to maximise the positive potential impact of the use of AI in the public sector, and to minimise negative or otherwise unintended consequences, national governments in the LAC region should:
1. Explore the collaborative development and execution of a LAC regional strategy and roadmap for AI in the public sector.
a. Work with other national governments in the region to identify a collective forum for strategy development, such as the Electronic Government Network of Latin America and the Caribbean (GEALC Network).
b. Explore leveraging third-party support for co-ordinating and facilitating strategy and roadmap development, such as by the OECD, CAF or the IDB.
c. Include a commitment and deadline for the national government of each regional adherent to develop their own national strategy and roadmap for AI in the public sector.
d. Ensure all participating countries have a voice in the design and implementation of the LAC regional AI strategy and roadmap.
e. Include collective commitments, objectives and goals in a manner that is high-level and open to be adapted by each country relative to their own unique context and AI aspirations as part of their national strategy for AI in the public sector.
f. Continue co-operation among LAC countries after the regional strategy is issued in order to help ensure implementation, monitor progress and promote regional collaboration.
g. Create a mechanism for the understanding and documentation of AI use cases in the public sector in the region, and the continuous sharing of good practices and lessons. Consider the OECD.AI Policy Observatory and GlobalPolicy.AI as outlets for sharing and co-operation.
2. Develop and adopt a national strategy and roadmap for AI in the public sector, for countries that have not done so already.
a. Move forward with developing a national strategy for AI in the public sector, even if a LAC regional strategy is not yet in place.
b. Strive for alignment between the national strategy for AI in the public sector and the LAC regional public sector AI strategy.
c. Ensure the public sector AI strategy is aligned with and contributes to the economic and societal goals and objectives of existing and future national AI strategies.
d. Consider in the public sector AI strategy’s development and implementation the need to reassess existing legal and regulatory frameworks to address the social, ethical and legal challenges related to the strategic and responsible use of AI in the public sector.
e. Adopt a collaborative and inclusive approach, both inside government and with the broader digital ecosystem and the public, in the development of the public sector AI strategy and the related and resulting policies and initiatives.
f. Include in the strategy or roadmap clear objectives and specific actions, measurable goals, responsible actors, time frames, monitoring instruments, and funding mechanisms, as appropriate.
3. For countries that have not done so already, develop a national public sector data strategy covering the different aspects of data, to serve as a foundation for progress of AI use.
a. Ensure the strategy is clear, aligned with the OECD’s framework for a data-driven public sector (OECD, 2019[1]), and includes all relevant aspects (e.g. data governance, public sector data assets and data sharing, data security and privacy, data infrastructure, data skills, fostering demand for data-driven decision-making, prioritisation of data investments and making public sector data open by default).
b. Strive for alignment with the national strategy AI in the public sector, the broader national AI strategy and the regional AI strategy for AI in the public sector.
c. Consider the need to reassess existing legal and regulatory frameworks to address the opportunities and challenges associated with leveraging data for AI in the public sector and secure alignment with relevant data protection laws.
d. Develop the strategy and all related and resulting policies and initiatives in an open and inclusive manner, both inside government and with the broader digital ecosystem, including the public.
e. Include clear objectives and specific actions, measurable goals, responsible actors, time frames, monitoring instruments, and funding mechanisms, as appropriate.
4. Explore regional co-operation and collaboration for public sector AI projects and initiatives.
a. Pinpoint specific public sector problems that could benefit from cross-border collaboration using AI in the public sector, and establish methods and processes for regional collaboration to address them.
b. Regional leaders (as identified in this report) should identify ways to assist and share lessons and insights in areas important for exploring and adopting AI in the public sector with countries that have less developed capacities in these areas.
c. Countries that have been identified in this report as having limited capacities in certain areas should take actions to enhance their capacities through increased attention and resources (e.g. training of civil servants, staffing, funding).
d. Consider leveraging external expertise, such as through procurement or partnering with inter-governmental organisations, especially for areas in which no countries have been identified as regional leaders in this report.
e. Seek out replication of proven models and ideas from others, so long as they are openly and appropriately adapted to each country’s own context and values.
5. Support AI efforts within the public sector at sub-national level and account for them in broader AI policies and initiatives.
a. Promote experimentation and adoption of AI at sub-national (e.g. local) levels, where governments are closer to citizens and their needs.
b. Consider developing AI hubs in cities across the region to focus on AI adoption in the public sector at the local level.
c. Explore how existing or new AI principles, guidelines and other tools can be used to ensure AI in the public sector is pursued in an informed and trustworthy manner at the sub-national level.
d. Empower sub-national and local leaders to have a voice in areas related to national and regional considerations for AI in the public sector (e.g. standards-setting groups, networks, strategy design and implementation working groups, etc.).
e. Facilitate dialogue and the sharing of good practices at the sub-national level.
6. Strengthen the overall focus on implementation to ensure pledges, commitments, and strategic objectives are realised.
a. Ensure the establishment of adequate processes and mechanisms to convert high-level objectives and commitments into real, implementable initiatives through a sustained focus on each item and accountability measures that ensure progress is made.
b. Look into the development of benchmark mechanisms to monitor public sector AI implementation in public institutions and lay the foundations for an impact evaluation mechanism.
7. Take steps to ensure the long-term sustainability of public sector AI strategies and initiatives.
Recommendations for promoting a responsible, trustworthy and human-centric approach to AI
8. Implement the OECD AI Principles and develop a detailed and actionable national ethical framework for trustworthy AI, for countries that have not done so already.
a. Implement the OECD AI Principles, which explicitly invites current non-adherents, including OECD non-member countries, to take note of the principles and adhere to them.1
b. Develop a national-level framework that is in alignment with the OECD AI Principles as well as the country’s context and norms in order to enable the development and implementation of trustworthy AI systems by public sector organisations. As seen in this report, such a framework can be embedded within a national AI strategy or developed as an independent document. Explore leveraging ongoing OECD work on the classification of AI systems, risk impact assessments and tools for trustworthy AI.
c. Explore the potential for developing practical instruments to help guide the implementation of the framework in the public sector, including through AI impact assessments, and establish an approach to AI implementation that takes into account various trade-offs and alternatives to using AI in the public sector.
d. Ensure the development of the framework and any associated policies and instruments is done in an open and inclusive manner, both inside government and with the broader digital ecosystem, including the public.
9. Ensure a focus on considerations for the use of trustworthy AI in the public sector as identified in this report, taking into account countries’ relative strengths and gaps in different areas.
a. Put in place mechanisms and capacities to support:
i. The inclusion of perspectives that are multi-disciplinary (different educational backgrounds, professional experiences and levels, skillsets, etc.) and diverse (different genders, races, ages, socioeconomic backgrounds, etc.) in a setting where their opinions are valued in the design and implementation of public sector AI strategies and initiatives (including AI-enabled projects).
ii. The practical implementation of ethical frameworks that safeguard against bias and unfairness, foster transparent and explainable AI systems, ensure robust, safe and secure processes, and establish clear accountability structures and clear roles and responsibilities for humans when it comes to AI use and AI-enabled decision-making in the public sector.
Recommendations for building key governance capacities to support AI in the public sector
10. Provide for sustained leadership capacity at the central and institutional levels to guide the development and ongoing implementation and oversight of public sector AI and data strategies and related initiatives.
a. Ensure senior political and government career leadership is actively involved in and supports the development and implementation of the national public sector AI strategy.
b. Put in place a Government Chief Data Officer (GCDO) or equivalent position responsible for developing and iterating upon a government data strategy and building the public sector ability to extract value from its data (including open government data, advanced analytics, algorithms and artificial intelligence).
c. Install institutional Chief Data Officers (iCDOs) or data stewards at each major public sector organisation to connect the strategic vision of the central government with data management practices at the institutional level and promote inter-institutional data co-ordination.
d. Ensure the GCDO and iCDOs have knowledge, skills and abilities relevant to AI in the public sector (e.g. data science, machine learning, trustworthy AI, etc.) and/or put in place positions with such skills to work in close co-ordination with the GCDO and iCDOs.
11. Leverage anticipatory innovation governance techniques in the public sector to prepare for the future.
a. Ensure strategies, roadmaps, and implementations are flexible and leave options open for the future.
b. Take into account the needs of sub-national government and local communities and strive for alignment between national frameworks for AI in the public sector and things that have a local impact.
c. Explore the potential for applying OECD Anticipatory Innovation Governance frameworks and approaches to public sector AI efforts (Tõnurist and Hanson, 2020[2]).
d. Equip public servants and citizens with the tools and capabilities to adapt to the changes that AI, including in the public sector, may bring in the future (including through the promotion of digital literacy and enhanced preparedness for the future of work).2
12. Ensure a solid focus on governance considerations identified in this report, taking into account countries’ relative strengths and gaps in different areas.
a. Put in place mechanisms and capacities to support:
i. Cross-government co-ordination for promoting strategic alignment synergies across public sector organisations that can support a purpose-oriented, problem-driven and trustworthy adoption of AI in the public sector, including through formal mechanisms (e.g. committees, councils, ethics boards) as well as less formal ones (e.g. communities of interest, networks).
ii. Internal and external communications to share the uses and benefits of AI in the public sector to both build trust among citizens as well as secure buy-in among public servants.
iii. Exploration and experimentation with different AI methods and approaches and data in an environment where public servants can take controlled risks (e.g. sandboxes and labs relevant to AI in the public sector).
iv. The systemic identification and understanding of public sector problems and the evaluation of multiple technology options to determine the needs and how AI can assist.
Recommendations for putting in place key AI enablers
13. Ensure a solid focus on the critical enablers for AI in the public sector as identified in this report, taking into account countries’ relative strengths and gaps in different areas.
a. Put in place mechanisms and capacities to support:
i. Access to accurate, reliable and appropriate data, and the provision of government data to fuel AI in all sectors.
ii. The provision of funding for public sector AI exploration and implementation.
iii. Bringing about the right expertise in government through upskilling and recruitment.
iv. Clearing a path for accessing external expertise and services through procurement and partnerships.
v. Access to the digital infrastructure necessary for AI in the public sector, such as hybrid cloud, computing power and interoperability services.
References
[1] OECD (2019), The Path to Becoming a Data-Driven Public Sector, OECD Digital Government Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/059814a7-en.
[2] Tõnurist, P. and A. Hanson (2020), “Anticipatory innovation governance: Shaping the future through proactive policy making”, OECD Working Papers on Public Governance, No. 44, OECD Publishing, Paris, https://dx.doi.org/10.1787/cce14d80-en.
Notes
← 1. For more information, visit https://OECD.AI/ai-principles or contact [email protected].
← 2. See www.oecd.org/future-of-work for the OECD’s efforts related to the future of work.