Scaling WBL with AI: Balancing Efficiency and the Human Touch

Work‑based learning (WBL) succeeds when schools, employers, and intermediaries move beyond classroom theory and give students real work, real mentorship, and real pathways. But WBL is resource‑intensive. 

As demand for high‑quality placements grows, programs face a familiar tension: how do you scale without sacrificing quality?

This spring, GPS Education Partners partnered with Hennepin West Consortium and SchoolJoy to pilot a conversational AI tool for the one task that often creates the biggest capacity bottleneck: student intake interviews. 

The goal was straightforward: see whether AI could reduce administrative burden, surface reliable, unbiased data about student interests and barriers, and free practitioners to focus on higher‑value work with students.

The pilot produced results worth sharing, both encouraging and candid.

What worked

• Efficiency gains. The AI completed 106 student interviews and saved the team more than 25 hours of initial call time. In a condensed pilot window, that level of throughput is meaningful for programs juggling tight timelines and capacity limits.
• Richer raw data. In many cases, the AI captured extensive, structured responses and produced summaries that were more complete than a typical intake note. That consistency can be helpful when programs need comparable data across hundreds of students.
• Accessibility and freshness. The tool showed up “fresh” for every conversation, which helped surface perspectives from students that decision-makers might otherwise overlook.

Where the human touch mattered most

Even with those gains, the pilot also uncovered why practitioners remain central to quality WBL. Experienced staff bring tacit judgment (reading tone, prioritizing follow‑ups, and making nuanced placement decisions) that’s difficult to reduce to rules or scripts. 

In practice, the AI output alone did not provide the confidence needed to place students without human review.

Two practical consequences emerged:

1. Implementation gaps show up early, but only if you test in small batches. We would have identified key mismatches sooner by piloting smaller batches rather than a large, single rollout.
2. AI is best deployed as a force multiplier, not a replacement. Its strength is in consistent data gathering and reducing routine workload. Final placement decisions still benefit from human synthesis and context.

What this means for WBL leaders

If your goal is to grow access to quality WBL, the pilot points to a pragmatic path forward:

• Treat AI as a tool for triage and enrichment. It can surface trends and free time, but design workflows that fold in practitioner judgment before finalizing decisions.
• Start small and iterate quickly. Run limited tests, compare AI outputs to human interviews, and refine prompts, scripts, and handling of edge cases before scaling.
• Document implicit wisdom. The most experienced practitioners reason in ways that aren’t obvious on paper; capturing that knowledge early will make any AI integration far stronger.
• Prioritize student privacy, authentication, and thoughtful scheduling so technology expands access without compromising safety or trust.

Collaboration matters

This pilot wouldn’t have been possible without Hennepin West Consortium’s willingness to experiment and SchoolJoy’s design partnership. Their contributions helped us ask the right questions about fairness, equity, and practical workflows, not just tech capability.

Learn more about the benefits of work-based learning.