ICTworks

Artificial intelligence (AI) technologies are increasingly prevalent in our lives—from digital voice assistants to personalized advertising and entertainment or advanced medical diagnostics. Access to the data and computing power that fuel AI is growing year-over-year, both in developed and developing economies.

At the same time, new techniques are being developed that enable the use of AI in domains where data is scarce, greatly widening the applicability and accessibility of AI in the near future. AI applications’ projected impact on the global economy by 2030 is equivalent to an increase in global GDP by 16 percent, and it is predicted to play a role in addressing each of the 17 UN Sustainable Development Goals.

However, we have also begun to see the ways in which the use of AI can inflict harm. AI tools can be used to infringe on basic human rights by enabling surveillance, censorship, or more effective targeting of individuals for malign or exploitative purposes.

The USAID Artificial Intelligence Action Plan outlines how we can engage responsibly with AI in our programming and in the countries where we work.

Responsible AI

AI applications lies at the center of emerging national security concerns: malicious cyber activity, the creation and spread of disinformation, microelectronics supply chain vulnerabilities, or the energy demands and climate impacts of cutting-edge AI systems. In addition, well-intentioned AI can nonetheless exacerbate inequities and amplify bias.

For example, AI technologies may systematically give lower credit scores to women compared to men, in part because there is less available financial data for women, thereby amplifying existing disparities in financial access. Because non-dominant or minority languages are often not well represented in digital form, language-based AI tools may systematically fail for, or misrepresent, speakers of these languages.

As the use of AI technologies proliferates across a broad range of markets, sectors, and country contexts, so too will the benefits, but also misuse and AI-related harms. Development actors have a responsibility to ensure that appropriate measures are taken to promote AI’s positive potential and to protect those most at risk of being negatively impacted by its use.

This includes advocating for the use of AI in USAID programming when its benefit is clear—but importantly, also advocating against its use when the risk of harm is too great.

Even beyond our programming, as AI technologies are built and deployed globally, development actors have an opportunity to influence important questions around AI’s evolution—how it’s constructed, by whom, and to what end. Today’s global AI ecosystem is increasingly centralized, with investment, talent, data assets, and decision-making power concentrated in a handful of major North American, European, and Asian firms.

AI Ecosystems

Development actors, including USAID, can encourage inclusive, equitable AI development and application, striving to serve those who stand to benefit the most, rather than solely those who have the strongest market presence or largest digital footprint.

As AI technologies are embedded and intertwined in digital ecosystems, a responsible approach to AI should include strengthening key aspects of the enabling ecosystem. This includes data systems, connectivity, and local workforce capacity. In addition, there must be a focus on strengthening the civil society structures holding AI systems and actors accountable, and shaping policy environments that in turn encourage open, inclusive, and secure digital ecosystems.

Together, these investments will support governments, businesses, and individuals to sustainably and equitably benefit from the use of AI technologies.

As the leading international development donor, USAID must continue to prioritize the rights-driven, responsible use of digital technologies and data in our work. When we turn to AI, we must commit to do so with full appreciation of the potential for benefit and harm, as outlined in Executive Order 13960, Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government. For AI, this includes constructing appropriate safeguards, investing in relevant talent, and understanding how AI technologies are connected with the broader digital ecosystem and the different stakeholders therein.

AI Partnerships

It also includes working closely with digital rights partners and experts to help determine when AI technologies should not be used in our work—and when the rights-infringing use of AI by others may merit USAID lines of effort to address potential harms.

Our approach is, and must continue to be, aligned with the two mutually reinforcing objectives of the USAID Digital Strategy: promoting the responsible use of digital technology for development, and working toward more inclusive, open, and secure digital ecosystems. As a technology that can identify, interpret, and act on patterns in data, AI is a natural extension of the international development community’s push to become more empirical and data-driven, and to drive innovation to where it is needed most.

AI’s ability to derive insights from disparate data sources offers potential for greater impact even across contexts where data scarcity or system fragmentation is a challenge. From AI-enabled health chatbots to satellite imagery analysis for resilience and disaster response, AI holds great promise for increasing efficiency, boosting productivity, expanding the reach of needed services, and empowering countries and communities.

USAID AI Action Plan

The USAID Artificial Intelligence Action Plan outlines how we can engage responsibly with AI in our programming and in the countries where we work with concrete steps to:

1. Commit to Responsible AI in USAID programming

A formal commitment to responsible use, more systematic exploration of the risks and rewards of using AI in development programming, and investments in staff capacity will help USAID to responsibly harness AI technologies for improved development and humanitarian assistance.

2. Strengthen digital ecosystems to support responsible use of AI

USAID’s efforts to strengthen digital ecosystems should include focused attention on key AI components: investment in digital data infrastructure and data governance, strengthening in-country systems of AI accountability and oversight, and promotion of a responsible, capable global AI workforce. These efforts should be grounded in a holistic understanding of the broader digital ecosystem; Digital Ecosystem Country Assessments can be a valuable precursor to AI-focused work.

3. Partner to shape a global Responsible AI agenda

Policy setting, research partnerships, and multi-stakeholder engagement play critical roles in promoting accountability and rights protection in the AI ecosystem, and USAID should continue to elevate the importance of including developing country actors in key global AI decision-making or policy-setting fora.

A lightly edited synopsis of the USAID AI Action Plan executive summary.