Across the development and social impact sector, organizations have been acquiring AI tools at a remarkable pace. Subscriptions are active. Accounts have been created. AI assisted workflows have begun to appear across communications, data, and program teams. AI capability building, as most organizations are currently approaching it, looks a lot like tool procurement, and that is precisely where the problem begins.
The gap between acquiring AI tools and building genuine AI capability is significant, measurable, and widening. According to the 2025 AI Benchmark Report from TechSoup and Tapp Network, 85 percent of nonprofits are already exploring AI. Only 24 percent have a formal strategy for implementing it.¹ The 2024 Nonprofit Standards Benchmarking Survey found that while 82 percent of nonprofits now use AI in some form, less than ten percent have formal policies governing its use.² These figures describe a sector-wide pattern: rapid adoption without the organizational infrastructure to make that adoption meaningful.
Understanding what AI capability building actually requires beyond the tools, is the question that will determine which organizations in this sector build genuine institutional capacity and which ones accumulate subscriptions they are underusing.
The AI Capability Gap: Why Most Organizations Are Falling Short
There is a useful distinction, rarely made explicitly in conversations about AI in the sector, between AI use and AI capability.
AI use is the act of deploying a tool in a workflow. A team member uses a language model to draft content. A data analyst runs reports through an AI assisted platform. A program officer summarizes field data with an AI powered tool. These activities carry real value, but they are discrete and individual. They do not add up to organizational capability on their own.
AI capability is the organizational condition in which artificial intelligence is integrated into how the institution thinks, decides, operates, and learns. It is not located in any single tool or any individual user. It is distributed across the organization’s culture, its leadership practices, its data infrastructure, and its governance systems. Building it requires a fundamentally different investment than purchasing a subscription.
Three shifts distinguish organizations moving toward genuine AI capability building from those remaining at the level of surface adoption.
The first is from tool to capability; treating AI not as a set of prompts but as a way to redesign how work actually happens.
The second is from experimentation to capacity; investing not just in training but in ownership, time, and leadership alignment.
The third is from task-level use to mission-level application; beginning with the most important problems the organization is trying to solve, rather than the easiest applications of the technology.
Most organizations in the development sector are still on the wrong side of all three.
AI Readiness for Nonprofits: Why It Is an Organizational Learning Problem
The framing that has dominated most conversations about AI in the social sector is that organizations need better access to the right tools. Research tells a more complicated story.
Frameworks for mapping AI capacity across organizations consistently identify five dimensions: culture and leadership, human capital and operations, data architecture, systems and infrastructure, and governance and regulatory compliance. Of these five, only two are primarily technical. The other three are organizational.
AI readiness requires building a technology-literate culture, strengthening leadership ownership of technology strategy, and linking AI efforts directly to mission and organizational ambition. These are prerequisites, not additions to a technology implementation plan.
For nonprofit and development organizations specifically, AI readiness faces an additional layer of complexity. One third of nonprofit managers cite employee resistance and ethical concerns as primary barriers to AI adoption alongside the more commonly discussed obstacles of limited knowledge, infrastructure, and funds.² That figure signals something important: the human and cultural dimensions of AI adoption are generating as much organizational friction as the technical ones, and yet they receive a fraction of the investment and attention.
AI readiness is, at its core, an organizational learning challenge. The technology is advancing faster than most organizations have had the time to adapt. The organizations that close the gap will do so by building the internal conditions that make meaningful adoption possible.
The Three Dimensions That Actually Matter for Development Organizations
For development organizations operating across complex, high stakes environments with accountability to communities, funders, governments, and civil society, AI adoption carries dimensions that the generic technology adoption literature does not fully address.
Data quality and context:
AI systems are only as useful as the data they operate on. Development organizations frequently work with fragmented, incomplete, or context-specific data, field data, beneficiary records, and operational logs that require significant preparation before it can meaningfully inform AI-enabled processes. Organizations that invest in AI tools before addressing their data infrastructure will find that the tools surface the problem rather than solve it.
Ethical and community accountability:
The communities that development organizations serve have legitimate stakes in how their data is used, what kinds of decisions AI systems inform, and whether the values embedded in those systems reflect the values of the communities themselves. Responsible AI adoption in this sector requires an explicit framework for these questions as a foundational design consideration, not an afterthought.
Institutional learning capacity:
The organizations that make the most progress in AI adoption are the ones that build the internal conditions, leadership alignment, psychological safety, and shared learning practices that allow experimentation to translate into lasting organizational knowledge. AI capability is not deposited through training sessions. It grows through the sustained accumulation of organizational practice.
AI Capacity Building as a Leadership Practice
AI capacity building is, at its foundation, a leadership question.
The organizations most effectively advancing AI capability are distinguished by the clarity and seriousness with which their leadership has defined what AI should do for the organization, why it matters at the level of mission, and what the institution is willing to invest in time, attention, and structural change to build genuine capability.
This requires something that most AI adoption conversations underemphasize: an honest assessment of the organization’s current state. What is the actual level of AI literacy across the staff? Where are the data quality gaps that will limit what any AI system can produce? What are the ethical boundaries the organization will not cross, and are they articulated clearly enough to guide decision making in practice? What governance structures exist or need to be built to ensure that AI adoption remains consistent with the organization’s values and its accountability relationships?
These are questions that leadership must ask of itself. Organizations willing to engage them honestly will build a foundation on which genuine artificial intelligence capacity building becomes possible. Organizations that skip this work will acquire a great many subscriptions and very little capability.
The Governance Gap Most Development Organizations Are Not Addressing
The governance dimension of AI capability building is the most consistently underinvested and the most consequential.
A vast majority of nonprofits have no formal AI policy.¹ Among the organizations that do have one, most focus on narrow operational guidelines rather than the broader questions of accountability, data ethics, and institutional values that AI adoption raises. This is a significant gap because their absence reflects a deeper organizational condition: AI is being treated as a technical matter rather than a strategic and ethical one.
For development organizations, the stakes of this gap are particularly high. These are institutions that work with vulnerable communities, sensitive data, and complex accountability relationships. AI systems operating without clear governance frameworks in these environments create risks that extend beyond the reputational. They are operational and ethical, the potential for bias in program targeting, privacy vulnerabilities in data handling, and the erosion of community trust that the organization’s effectiveness depends on.
Governance in AI capability building is not a bureaucratic exercise. It is the organizational expression of the values that should be guiding adoption. An institution that has articulated what AI should and should not be used for, how the data of the people it serves will be protected, and how accountability will be maintained at every level of the organization, has a foundation for genuinely responsible AI adoption. An institution that has not is building on unstable ground, regardless of how sophisticated its tools are.
What This Means for the Development and Social Impact Sector
The AI landscape in this sector is dividing. The organizations pulling ahead are applying AI to core mission challenges, structuring their internal practices around learning and iteration, and treating AI not as a set of products to acquire but as a capability to build over time.
If AI remains a side tool deployed by individuals in isolated workflows, ungoverned by any coherent organizational framework, and disconnected from the institution’s strategic priorities it will produce marginal gains. The organizations that treat AI capability building as a genuine institutional priority, one requiring cultural investment, leadership clarity, governance infrastructure, and sustained learning, are the ones that will operate at a meaningfully different level of effectiveness in the years ahead.
The question for leaders in this sector is whether the adoption happening inside the organization is building genuine capability or simply accumulating activity. The answer will determine whether AI becomes one of the most significant institutional assets they have ever developed, or another set of tools it is paying for and underusing.
At Iwà Consulting, we work with development and social impact organizations navigating the strategic, communication, and governance dimensions of AI capability building. The conversation starts here.
Start a Conversation → iwa-worldwide.com
Sources
¹ TechSoup and Tapp Network. State of AI in Nonprofits: 2025. January 2025. Based on responses from over 1,300 nonprofit professionals. blog.tappnetwork.com
² 2024 Nonprofit Standards Benchmarking Survey. Cited in Whole Whale, Top Nonprofit AI Policies 2025: Analysis and Trends (July 2025) and Bridgespan Group, AI Can’t Be Ignored: Exploring the Opportunities for Nonprofits and the Social Sector (July 2025). wholewhale.com / bridgespan.org
