Artificial intelligence (AI) is reshaping how work is organized, managed, and accessed. Over the past three years, we have worked alongside grantees, researchers, and other partners to better understand the emerging risks and opportunities associated with AI and what it will take to ensure low-income workers have the power to advance economically. This effort builds on Irvine’s long-standing focus on how technological change, particularly automation, affects workers paid low wages.  

What is different now is the speed and breadth of AI’s potential influence across industries and jobs. While public discourse often swings between hype and alarm, those narratives can distract from fundamental questions: Who has a say in how AI is built, used, and governed? Who benefits, and who bears the risks? And what will it take to ensure our workforce systems, safety-net programs, and broader economy work for everyone in the age of AI?  

Many of AI’s impacts are still unfolding, but one thing is clear: AI’s trajectory is not predetermined. The effects of AI will be shaped by the decisions made now by funders, policymakers, employers, training providers, and technologists. Several key insights have emerged that point to what will ensure equity and where attention is most needed.  

The workforce system must adapt to continuous change  

AI-related shifts in job tasks, skill requirements, and advancement pathways are challenging today’s workforce systems. Most training still happens through one-time reskilling programs, even as change has become more continuous. Sixty-six percent of workers say AI is already impacting their jobs, yet only one in three have the training needed to use it effectively. This highlights the need for a workforce approach that supports the continuous development of deeper, transferable skills and more consistently recognizes and values the skills that workers have.  

Models that allow workers to earn and learn at the same time, such as apprenticeships and incumbent worker training, are especially important because they help workers build skills while remaining employed. This distinction matters because research consistently shows that unemployment or underemployment can have long-term consequences, including reduced earnings, weaker workforce attachment, and greater instability for workers and their families.   

The risks are even greater for women and people of color, who are overrepresented in clerical, administrative, and low-wage service roles facing high AI exposure. Workforce responses must be grounded in their experiences, and we should not treat job displacement as inevitable or acceptable. As the nature of work changes, we must focus on supporting workers in accessing and transitioning into roles that offer stability, dignity, and opportunity. At Irvine, we will invest in stronger workforce infrastructure, support adaptive capacity among employers and training providers, and ensure workers have a meaningful role in shaping solutions.  

Policy and governance will shape how AI is used  

Many labor, civil rights, and privacy laws were not designed for algorithmic decision-making. As a result, policy is playing a critical role in shaping how AI is deployed and experienced, and California has taken important steps in this area. Recent updates to the California Consumer Privacy Act established new rules governing automated decision-making technologies, including stronger transparency and privacy protections in high-stakes areas like employment and housing. Updates to the Fair Employment and Housing Act regulations also further clarified that employers remain accountable for discriminatory outcomes, even when decisions are made by algorithms.  

Irvine grantees such as TechEquity Collaborative, the UC Berkeley Labor Center, Economic Security California, and others played key roles in shaping these safeguards. While these developments represent meaningful progress, their impact on workers’ lives will depend on how these policies are implemented, enforced, and complemented by other strategies like organizing and collective bargaining.  

Economic security systems must evolve alongside work  

Even with more adaptive workforce strategies, some workers will still experience job loss or reduced hours and wages, underscoring the importance of strong safety net programs. Today’s systems leave many workers out, offer limited support, and are not well equipped to handle the kind of disruption AI may accelerate. Irvine grantee Prosper California is making progress on this front by advocating for increased cash payments and other policies to create opportunities for Californians, including expanding the California Earned Income Tax Credit (CalEITC) to cover historically excluded families. To help workers navigate disruptions and maintain financial stability, we need income supports and benefits that are accessible, responsive, and sufficient.  

We need better data to understand what’s happening in real time  

Most research on AI and work focuses on which jobs are most likely to be affected. While useful, this captures only part of the story. We still lack timely, actionable data on how workers are experiencing these changes, including how people are moving between jobs, how wages and hours are shifting, and where displacement happens. Without this insight, it’s harder to create effective workforce development, policy, and employer solutions. Investing in stronger data infrastructure is critical to understanding current impacts and preparing for what’s next.   

AI’s impact on government and nonprofit services will depend on readiness and design  

Nonprofits are beginning to explore how AI could expand administrative capacity and improve program delivery, while public agencies are considering how it might enhance data use and streamline access to benefits and services. Technology can improve how systems operate, including making them easier to navigate. However, outcomes will still depend on how those systems are designed, organizations’ ability to implement them effectively and whether they’re built with the needs of communities in mind. As this work evolves, careful attention to organizational capacity, user-informed design and continuous improvement will be essential to ensure AI strengthens services rather than complicates them.  

AI is becoming a powerful force in our economy. It’s reshaping work and how decisions are made, while also creating new possibilities for collective action and more equitable change. Building on the progress California has made will require continued investment, coordination, and shared learning to make sure AI helps people move ahead economically instead of deepening inequality.   

In the next post in this series, I’m excited to share more of our grantees’ worker-driven approaches to AI. While our learning journey is far from complete, we hope this series serves as a resource for funders and organizations navigating AI’s growing role in our work and lives.   

Lul Tesfai is Director of Program Development at The James Irvine Foundation and leads the Foundation’s work on AI.  

This piece is part of our blog series on Irvine’s work to advance a worker-driven approach to AI. Read other posts in the series here.