Throughout this course, we've discussed current applications of AI in philanthropy: proposal screening, impact measurement, trend analysis. But AI technology is advancing rapidly. In 2026, we're at an inflection point where near-term developments will dramatically reshape the sector. Artificial general intelligence (AGI), multimodal AI systems that process text, images, audio, and video, and autonomous agents that can operate independently present both extraordinary opportunities and significant risks.
This lesson might seem speculative. But as grant professionals committed to thought leadership and long-term strategic planning, we must engage with emerging technologies now—before they reshape our sector. Organizations that wait until next-gen AI is proven will be late. Organizations that engage thoughtfully now can shape how these technologies are developed and deployed.
Next-generation AI technologies will reshape philanthropy within 3-5 years. Grant professionals should engage with these developments now, not wait until change is forced upon them. The question isn't whether to prepare, but how to prepare responsibly.
Current AI systems (including large language models like Claude, GPT, and others) are powerful but limited. They excel at pattern recognition, language understanding, and text generation. But they lack true understanding, cannot reliably reason about novel situations, and require human oversight for consequential decisions. They're tools that augment human capability.
Next-generation systems represent different architectures and capabilities. Artificial General Intelligence (AGI) refers to AI systems with human-level or superhuman intelligence across diverse domains. Rather than being specialized (good at language, or images, or chess), AGI systems would be generalists capable of learning and performing any intellectual task humans can.
Multimodal AI systems integrate multiple data types—text, images, audio, video—into unified representations. Current systems are largely specialized: language models process text; image models process images. Multimodal systems process all simultaneously, understanding not just what words say but how they're inflected in video, what's visible in images, and how audio reinforces meaning.
Autonomous agents are systems that can operate independently, making decisions and taking actions without constant human direction. Current AI systems are passive: they respond to prompts. Autonomous agents actively pursue objectives, decompose complex tasks, make decisions, and execute actions. They might monitor foundations' RFPs, identify opportunities, draft proposals, and submit them—all autonomously.
When will AGI arrive? Expert opinions vary dramatically. Some AI researchers believe AGI is 2-5 years away. Others believe it's 20-50 years away. Still others doubt AGI is possible at all. This uncertainty itself matters: we're operating under genuine uncertainty about transformative technology.
The implications of AGI for philanthropy are profound. If AGI arrives, strategic planning becomes fundamentally different. Who decides what AGI systems work toward? How is power distributed? Can the nonprofit sector influence AGI development? These questions are currently being debated at the highest policy levels. Grant professionals should understand the stakes.
In more moderate scenarios, transformative but sub-AGI systems might be deployed in philanthropy by 2027-2030. Imagine an AI system that could autonomously manage foundation operations: receiving and reviewing proposals, recommending funding decisions, managing grantee relationships, monitoring outcomes, and adapting strategy based on results. This doesn't require AGI—it requires advanced multimodal systems with strong reasoning capabilities.
Such systems would raise profound questions: What is the program officer's role if AI manages most grantmaking functions? How do we ensure equity and accountability? How do we prevent algorithmic approaches from displacing human values? These questions should be debated now, in advance of such systems being deployed.
Multimodal AI represents near-term transformation. Imagine a foundation receiving a nonprofit's grant application: text proposal, video of program in action, audio of beneficiary testimonies, photographs documenting work. Current systems would analyze each separately. Multimodal systems understand them integratively.
The foundation's AI could watch the video while reading the proposal, understanding tone of voice, visual evidence of program quality, authenticity of testimonies. It could cross-check claims in the written proposal against evidence in video and photographs. It could assess the nonprofit leader's credibility by analyzing communication across multiple modalities. This integration of information creates richer understanding than any single modality alone.
For impact evaluation, multimodal AI is transformative. Foundations could submit cameras to grant sites, and AI would analyze video, photographs, and data streams in real-time, assessing program quality, identifying challenges, recommending adaptations. This near-constant monitoring would enable genuinely adaptive grantmaking: when problems are detected, they're addressed immediately rather than discovered at year-end evaluation.
But multimodal AI also introduces new risks. Deepfakes—convincing but fabricated videos—become harder to detect and could mislead foundations about program quality. Nonprofits might face pressure to present polished video evidence rather than authentic reality. The convenience of multimodal assessment might lead foundations to over-rely on AI analysis and under-value the human judgment that comes from site visits and relationships.
Multimodal AI appears objective because it integrates multiple information sources. But it's no more objective than its underlying data and algorithms. Deepfakes, biased video analysis, and manipulation risk are real. As multimodal systems become prevalent, skepticism and human judgment remain essential.
The most controversial next-gen development is autonomous agents. Current AI requires human instruction: "Analyze this proposal." "Draft a recommendation." An autonomous agent would be given an objective—"Identify promising funding opportunities in climate adaptation"—and independently pursue that objective, making decisions along the way without asking permission.
For nonprofits, autonomous agents present opportunity: imagine an AI that continuously monitors foundations seeking climate adaptation funding, identifies opportunities aligned with your organization's work, drafts compelling proposals, and submits them. The AI becomes your development director, working 24/7 to secure funding. This could be transformative for small organizations lacking dedicated development capacity.
For foundations, autonomous agents present distinct scenarios. A positive vision: AI agents autonomously manage routine grantmaking, freeing program officers to focus on strategy, relationship-building, and deep learning with grantees. The foundation's impact deepens even as staff size remains constant. A concerning vision: autonomous agents optimize for measurable metrics (proposal approval rate, funds deployed, short-term outcomes) while losing sight of values-driven judgment that makes philanthropy distinctive.
The key tension with autonomous agents is control and alignment. Can we ensure agents pursue our values? Or will they pursue narrow objectives in problematic ways? The "paperclip maximizer" thought experiment illustrates the risk: an AI tasked with maximizing paperclips might consume all available resources producing paperclips, potentially harming everything else. Similarly, an agent optimizing for "maximize grants funded" might fund low-impact organizations if they're easy to deploy resources toward, rather than high-impact organizations requiring more complex engagement.
Next-gen AI creates genuine opportunities. Efficiency: foundation staff could shrink while funding volume increases, enabling more resources to reach nonprofits. Speed: funding decisions could happen in days or hours rather than months. Equity: algorithmic approaches could reduce individual bias if properly designed. Scale: foundations could serve geographies and organizations they currently can't reach due to capacity constraints. Learning: multimodal monitoring could enable continuous learning about what works, dramatically accelerating organizational development.
Additionally, next-gen AI could democratize funding. Currently, large foundations have sophisticated analytical capabilities while small foundations and grassroots funding mechanisms lack them. Next-gen AI tools, if accessible and affordable, could enable smaller foundations and community-based funders to operate more effectively.
As systems become more autonomous, control becomes more difficult. A program officer can explain why she rejected a proposal. An algorithm can show its reasoning. But an autonomous agent making complex decisions across multiple domains? Its reasoning becomes opaque. How do we ensure it's acting consistent with our values? How do we hold it accountable if it goes wrong?
If AI handles most routine grantmaking, will humans lose capacity to make complex funding decisions? There's a documented phenomenon in aviation: pilots who rely too heavily on autopilot lose their ability to fly manually. Similarly, foundation staff who rely on autonomous agents might lose judgment-making ability. When the system fails, humans can't step in effectively.
Philanthropy is fundamentally values-driven. But values are hard to quantify and algorithm-optimize. Money. There's genuine risk that over-reliance on algorithmic systems causes mission drift away from values and toward whatever is algorithmically optimizable.
If only sophisticated foundations can afford next-gen AI, the gap between well-resourced and under-resourced foundations widens. Foundations with AI gain competitive advantage, draw more top talent, fund more organizations. Smaller foundations fall further behind. This concentrates power in large institutions, potentially undermining sector diversity.
If you're a foundation considering autonomous AI agents, ask yourself: What decisions are genuinely better made algorithmically? Which decisions require human judgment and values-orientation? Keep humans in control of high-stakes decisions. Use AI to augment human capability, not replace it.
Rather than predicting the future, sophisticated organizations engage in scenario planning. What if autonomous agents are widely deployed by 2028? What if multimodal AI transforms evaluation by 2027? What if AGI arrives by 2030? How should we prepare for each?
Scenario 1: Autonomous Agents (2028-2030) By 2028, nonprofit development directors use AI agents to continuously monitor and apply to funding opportunities. Foundation staff members who previously spent 50% of time reviewing proposals now spend 10%, with algorithms handling initial screening. Program officers focus on relationships, strategy, and organizational development. Foundation governance requires explicit policies about agent autonomy, override mechanisms, and accountability. Organizations without access to agent technology fall behind.
Scenario 2: Multimodal AI (2027-2029) By 2027, foundations deploy cameras and sensors at grantee sites, with multimodal AI continuously analyzing program quality and outcomes. Real-time dashboards show program metrics to foundation and nonprofit staff. This enables incredibly rapid learning and adaptation. But it also requires nonprofits to accept constant surveillance and submit to algorithmic assessment of program quality. Power dynamics shift toward foundations.
Scenario 3: AGI (2030+) If AGI arrives, everything changes. We don't know what AGI would mean for philanthropy. Current governance frameworks become obsolete. Entirely new questions about power, values, and human flourishing become central.
Rather than predicting which scenario unfolds, intelligent organizations prepare for multiple possibilities, remain flexible, and maintain the human judgment and values-orientation that make philanthropy distinctive regardless of technology.
The nonprofit sector needs to engage with next-gen AI now. This means: research on AI's impacts on philanthropy, development of governance frameworks before systems are widely deployed, ensuring diverse voices shape AI development (not just technologists, but nonprofit leaders, communities affected by funding), and establishing clear accountability mechanisms.
Grant professionals have important roles. As thought leaders, you can articulate the risks and opportunities. As practitioners, you can ensure that AI systems serve authentic philanthropic values. As advocates, you can demand transparency and accountability from foundations deploying AI. As researchers, you can study what works and what doesn't, building evidence to guide the sector.
New governance approaches are emerging for AI systems. Participatory governance brings affected stakeholders into decisions about system design and deployment. Algorithmic impact assessments evaluate potential harms before systems are deployed. Algorithmic auditing enables ongoing monitoring of real-world performance. Transparency and explainability requirements ensure stakeholders understand how systems work.
These approaches are more demanding than deploying technology quickly and cheaply. But they're essential if the sector wants AI to advance equity rather than entrench existing power imbalances.
The future of philanthropy is not predetermined. The choices we make now about how to develop and deploy next-gen AI will shape what philanthropy becomes. Grant professionals—who understand both the sector and the technology—have crucial roles to play in shaping that future thoughtfully and responsibly.
Ready to master AI in philanthropy? Enroll in the complete CAGP Level 5 course and earn your certification in advanced grant leadership.
Explore Full Course