"Is AI coming for my job?"
I hear this question constantly from people in social impact—social impact professionals in Nairobi, M&E officers in DC, program managers in São Paulo, policy analysts in Brussels, innovative finance experts in London and more.
After spending the past two years watching this unfold across regions and sectors (and making plenty of wrong predictions along the way), here's what I've learned: the answer isn't yes or no. It's more complicated, and honestly more interesting.
The grant writers I thought would struggle? Many are thriving by becoming strategic advisors who use AI as a drafting partner. The data roles I assumed were safe? Some are being hollowed out faster than expected. The lesson: predicting which specific jobs "survive" is mostly a fool's errand.
What I can say with more confidence is this: we're not facing an apocalypse immediately (it may be coming). We're facing a massive reshuffling—and how you position yourself within that reshuffling matters enormously.
The most credible global forecasts don't point to mass extinction. The World Economic Forum's Future of Jobs Report 2025 projects about 92 million jobs displaced by 2030—but 170 million created. That's a net gain of 78 million positions globally.
These aren't just numbers on a slide. They represent real people navigating career pivots, some exciting and some terrifying. They represent entire sectors being rewritten. And critically for the impact field, they represent an opportunity to shape how this transition unfolds—rather than simply being shaped by it.
For social impact professionals, the translation is this: most roles won't disappear, but the tasks inside them will be fundamentally rewritten. The organizations that adapt fastest—learning to blend human judgment with AI capabilities—will have an outsized advantage in attracting talent, delivering on missions, and stretching limited resources further. The ones that don't? They'll likely struggle to compete for both funding and people.
But here's my concern: Based on the research I'm doing and conversations I'm having across the sector, I'm not as optimistic as the WEF headline numbers might suggest. Yes, AI will create new jobs—including roles we can't even name yet, which is fascinating to consider. But a critical question is: What percentage of those jobs will be stable, full-time positions with benefits?
What I'm seeing in practice is the gigification of the economy accelerating: more contract work and fewer permanent positions; more project-based engagements and less job security; more "flexibility" that often means unpredictable income; fewer employer-provided health insurance, retirement benefits, and paid leave; and more workers classified as independent contractors rather than employees.
This pattern isn't new—but AI is likely to accelerate it. When organizations can use AI to handle routine tasks, they often need humans for shorter bursts of high-judgment work rather than ongoing full-time roles. The math starts favoring consultants and gig workers over staff positions.
For social impact specifically, this raises uncomfortable questions: How do we advocate for dignified work while our own sector increasingly relies on precarious employment? What happens to the pipeline of emerging leaders if entry-level positions become contract gigs without mentorship or growth pathways? How do we build organizational culture and institutional memory with a rotating cast of consultants?
The WEF's net positive job numbers may prove accurate (I am skeptical)—but if most of those 170 million new jobs are gig positions without stability or benefits, that's a very different story than the one the headlines tell.
There's another key challenge: who controls AI, and who benefits from it? Right now, the development and deployment of the most powerful AI systems is concentrated in a relatively small number of corporations and well-resourced institutions. If that concentration deepens—if the productivity gains from AI flow primarily to those who own the technology rather than those who work with it—we could see wealth concentration accelerating rather than distributing, developing regions locked out of AI benefits due to infrastructure gaps and licensing costs, decision-making power over critical systems held by a shrinking number of actors, and communities affected by AI-driven decisions with little voice in how those systems are designed.
I hope AI development becomes more distributed, more open, more accountable. But hope isn't a strategy—and the current trajectory gives reason for deep concern.
The picture is nuanced—and deeply unequal across regions, genders, and income levels. According to UN reporting on ILO-NASK research, generative AI is more likely to transform jobs than fully replace them. But that transformation isn't distributed evenly: high-income countries see ~34% of jobs facing significant AI exposure, while low-income countries face only ~11%.
This gap cuts two ways. On one hand, workers in advanced economies face more immediate disruption. On the other, workers in developing regions risk being locked out of the productivity gains that AI enables—potentially widening global inequality rather than narrowing it.
From a macro lens, the IMF's analysis on AI and the future of work estimates roughly 40% of global employment is exposed to AI, with advanced economies at ~60% exposure and low-income countries at ~26%. The implications for social impact, international development, humanitarian response, and global equity are significant—and largely unaddressed in current policy conversations.
The gender dimension deserves attention. In high-income countries, women's employment in the highest-risk category sits at ~9.6%, while men's employment in that category is ~3.5%. These percentages represent real people—often in administrative, clerical, and data-processing roles—facing genuine uncertainty about their futures. For organizations committed to gender equity, this isn't an abstract concern. It means the AI transition, left unmanaged, could disproportionately affect women's economic participation. It also means intentional job redesign and reskilling investments can either reinforce or counteract existing inequalities.
One fascinating counterpoint: An OECD study examining 19 countries found no indication that AI affected wage inequality between occupations (2014–2018), and some evidence it may actually be associated with lower inequality within occupations. Of course, past performance doesn't guarantee future results—and generative AI represents a qualitatively different challenge than the AI systems of a decade ago. But this finding is a useful corrective to narratives that assume automation inevitably creates winners and losers. The outcomes depend heavily on how institutions, employers, and workers respond.
The World Bank's research on automation in East Asia Pacific reinforces this point: labor market outcomes aren't technologically determined. They're shaped by education systems, labor protections, infrastructure investments, and the choices organizations make about how to deploy new tools.
In my experience, AI pressure hits hardest where roles are heavy on repeatable "information work"—drafting, summarizing, triaging, basic analysis—and lighter on lived context, relationship trust, or navigating ambiguity in complex environments. Here's how I see specific roles evolving, based on patterns I'm observing across regions:
Grant writer ? Fundraising strategist + AI editor
Drafting proposals gets dramatically faster with AI assistance. A skilled grant writer can now produce first drafts in a fraction of the time, freeing hours previously spent on boilerplate. But here's what AI can't do: build the relationship with the program officer that gets your proposal read carefully, understand the funder's unstated priorities, position your organization's work within a shifting funding landscape, or tell the story that makes a reviewer feel something. The differentiators become relationship-building, strategic positioning, and ethical storytelling. The writing itself matters less; knowing what to write and for whom matters more. Grant professionals who embrace this shift are becoming fundraising strategists who happen to use AI as a drafting tool—not writers who occasionally check ChatGPT.
M&E officer ? Learning systems designer
AI can help with synthesis and pattern recognition across large datasets, coding qualitative data, generating dashboards and visualizations, and surfacing anomalies and trends. These capabilities are genuinely useful—and they're automating tasks that consumed significant M&E staff time. But humans still need to set the right questions (at least for now, a surprisingly difficult skill), interpret what findings actually mean in specific contexts, catch when metrics are being gamed—a near-universal phenomenon when measurement creates incentives—and translate findings into organizational learning that actually changes practice. The job becomes more strategic, not less important. M&E professionals who position themselves as learning systems designers—people who help organizations ask better questions and act on what they discover—will find growing demand for their skills.
Program manager ? "Human + machine" operator
Program management has always been about coordinating complexity: multiple partners, competing timelines, limited resources, difficult tradeoffs. AI adds new tools to that toolkit—but also new coordination challenges. The work increasingly becomes orchestrating partners, communities, and AI tools together. The value is judgment in ambiguous situations, awareness of power dynamics, cultural fluency across contexts, and execution in messy real-world conditions. These are precisely the capabilities AI struggles with. Program managers who develop practical AI fluency—knowing when to use which tools, how to quality-check outputs, how to integrate automation without losing the human touch—will be well-positioned as organizations navigate this transition.
Policy analyst ? AI governance + accountability specialist
Expect significant growth in roles that translate between legal frameworks, community realities, and technical systems. This is particularly true in high-stakes domains: elections and democratic processes, social protection programs, migration and asylum systems, humanitarian response, and public health. As AI systems become embedded in decisions that affect vulnerable populations, the need for people who can audit those systems, advocate for accountability, and ensure affected communities have voice will only increase. Policy analysts who build expertise at the intersection of technology governance and social impact will find themselves in growing demand.
Emerging niche: Community data stewards
A genuinely new category of role is emerging: positions focused on protecting community consent, data sovereignty, and safe use—particularly where vulnerable communities risk extraction or surveillance. These roles combine technical understanding of data systems, community trust-building and engagement, and ethical reasoning and rights frameworks. They're appearing in Indigenous data sovereignty initiatives, refugee response contexts, community health programs, and anywhere sensitive population data intersects with powerful institutions. If you're looking for a growth area that doesn't yet have established pathways, this is worth watching.
If you're early or mid-career, the "safe" path isn't picking a job title that seems automation-proof—because we genuinely don't know which titles will exist in ten years (a subject for a future post or many posts). The resilient path is building a profile that's hard to replicate:
Domain credibility + Human skills + Practical AI fluency
Domain credibility means deep expertise in a specific problem area, region, or population. Human skills include relationship-building, communication, judgment under uncertainty, and the ability to navigate complex stakeholder dynamics. Practical AI fluency means knowing how to use these tools effectively and critically—not becoming a technologist, but becoming someone who can work alongside AI systems productively.
If the new jobs AI creates are predominantly gig positions without stability, benefits, or growth pathways, we need fundamentally different policy responses than if they're stable employment. This requires serious thinking about rethinking the social safety net (universal basic income, portable benefits, expanded social insurance), labor protections for algorithmic management (transparency requirements, anti-discrimination protections, rights to human review, protections against surveillance), distributing AI's benefits more broadly (public investment in open-source AI, digital infrastructure in developing regions, antitrust enforcement, requirements for community voice), and education and reskilling at scale (shorter-cycle credentials, on-the-job training subsidies, curriculum reform that treats AI literacy as foundational, support for mid-career transitions).
These aren't easy conversations—and they don't have consensus answers. But avoiding them means letting the transition happen to workers rather than with them.
Policymakers and government officials should recognize that the risk isn't only job loss—it's job quality erosion, wage pressure, and widening inequality unless institutions actively shape how AI is adopted. Fund rapid reskilling at scale, since traditional education timelines can't keep pace with the speed of change. Update labor protections for algorithmic management as AI systems increasingly make or influence decisions about hiring, task assignment, performance evaluation, and termination. Invest in digital infrastructure globally, or the Global South will be locked out of productivity gains. Take job quality seriously, not just job quantity—track job stability, benefits coverage, and income predictability, not just employment numbers.
Educators and training institutions should treat AI literacy as a baseline capability, not an elective—especially for public policy, public health, humanitarian studies, international development, and education programs. Teach critical AI use: students need to audit outputs, cite sources appropriately, detect hallucinations, understand model limitations, and recognize when AI-generated content requires human verification. Redesign assignments and assessments, since the old model doesn't prepare students for environments where AI can generate first drafts instantly. Build human-centered workflow design into curriculum—"AI user" is becoming as fundamental as "email user" was twenty years ago.
Funders and private sector employers shouldn't just buy AI tools—redesign jobs so AI removes drudge work without stripping roles of growth, agency, and meaning. Consider making "responsible AI adoption + workforce transition" an eligible budget line. Model transparent AI use in your own operations. Resist the temptation to gigify—before converting a staff position to a consultant role, ask whether you're genuinely gaining flexibility or just shifting risk onto workers.
If you're navigating this transition, the PCDN Career Campus offers resources on how impact hiring is changing, practical guidance on career positioning, and 350+ curated impact jobs every month across regions and sectors.
Key Resources:
This post was prepared with assistance from a great AI tool provided by every.to called Spiral and combined with lots and lots of human editing, prompting and more.
