What Per Scholas and AIR learned about AI adoption in workforce development 

A new thought leadership article from Per Scholas and the American Institutes for Research (AIR) explores the realities of integrating artificial intelligence into workforce training and the lessons organizations have learned along the way. In “What We Got Wrong (and Right) about Bringing AI into Workforce Training,” authors Samia Amin and Tamara Johnson reflect on five years of collaboration between AIR and Per Scholas to better understand how AI can support learners preparing for careers in technology.

The article centers on Per Scholas’ implementation of an AI-powered intelligent tutoring system designed to help learners prepare for industry certification exams. While the tool proved effective for learners who used it, nearly half of the participants never engaged with the platform. Rather than treating the rollout as a failed pilot, Per Scholas and AIR used the findings to rethink how AI tools are introduced, supported, and evaluated within workforce training environments.

The piece highlights several key lessons from the partnership. One major takeaway is the importance of starting with a clear theory of change before implementing AI solutions. The authors emphasize that organizations must define how a tool is expected to create value, the assumptions about learner behavior, and how success will be measured over time.

The article also underscores the importance of measuring early indicators of progress rather than waiting for long-term outcomes, such as job attainment. By tracking learner engagement, adoption rates, and usage patterns, organizations can more quickly determine whether an initiative is gaining traction and where adjustments may be needed.

Another central theme is that adoption itself must be treated as a design challenge. The authors, Amin and Johnson, note that learners and staff often gravitate toward familiar tools and systems, meaning successful AI implementation requires thoughtful onboarding, support strategies, and continuous iteration.

The article further explores how organizational culture influences successful AI adoption. At Per Scholas, a strong focus on economic mobility and learner outcomes has guided decisions on which AI initiatives to pursue. Meanwhile, AIR adapted its traditional research approaches to better align with the pace and complexity of AI-driven program changes.

Looking ahead, the authors explain that these lessons are now shaping future work around AI-powered career coaching and job placement tools. Ultimately, the article argues that the organizations best positioned for the future will not simply be the first to adopt new AI technologies, but those that build internal capacity to continuously learn, adapt, and improve over time.

Read the full article here: What We Got Wrong (and Right) about Bringing AI into Workforce Training.