A new study (2025) suggests that when it comes to training employees, less AI might actually lead to better outcomes. (Note here that “less AI” doesn’t mean “no AI.” In the world of work, forgoing AI will soon be impracticable.) The research, published in the International Journal of Human–Computer Interaction, explores how different levels of AI automation in training influence worker motivation, engagement, and skill-building — and the results may (or may not) surprise you, depending on how you’ve been using AI.
The Setup: Not All AI Is Equal
The study tested three training approaches using an AI-powered quality control task:
- One group trained with no AI assistance.
- A second group used a fully automated AI, which made all decisions.
- The third trained with partially automated AI, where the AI suggested decisions but humans made the final call.
All groups were later tested without any AI to see how well they performed under real-world conditions — simulating a system failure scenario.
The Results: Partial AI Hits the Sweet Spot
Participants who trained with partial AI automation outperformed the fully automated group in identifying errors — and matched the performance of those trained without AI. But the benefits didn’t stop at performance.
Workers in the partially automated group also reported:
- Higher motivation, especially in terms of finding meaning in their tasks.
- Greater feelings of autonomy, crucial for workplace satisfaction.
- Increased physical engagement, as measured through wearable sensors.
In contrast, those trained with full automation showed reduced motivation and struggled more when the AI was removed — a critical concern in workplaces where AI is prone to malfunction or isn’t always reliable.
Why It Matters for the Future of Work
In an age where AI is rapidly integrating into workplaces, this research offers a vital reminder: humans must stay in the loop. Training systems that rely entirely on automation risk creating passive employees, ones who follow instructions but lack the skills to adapt or respond when technology fails.
Bottom Line
If the goal is to build resilient, motivated, and skilled workforces, companies should think twice before handing the wheel entirely to AI, especially in training. As we saw here, a bit of friction and decision-making during training can go a long way in preparing workers for the real world.
Main reference:
Mario Passalacqua, Robert Pellerin, Esma Yahia, Florian Magnani, Frédéric Rosin, Laurent Joblot & Pierre-Majorique Léger (2025) Practice With Less AI Makes Perfect: Partially Automated AI During Training Leads to Better Worker Motivation, Engagement, and Skill Acquisition, International Journal of Human–Computer Interaction, 41:4, 2268-2288, DOI: 10.1080/10447318.2024.2319914
Loor Lemon 
Leave a comment