Researchers warn that skill erosion caused by AI could have a devastating and lasting impact on businesses - but it may already be too late (www.techradar.com)

Researchers warn that widespread AI automation is producing a latent “skill erosion” in workplaces: as tools take over detail work, employees lose the domain knowledge and procedural competence needed to detect errors or operate without the system. A 2023 case study, "The Vicious Circles of Skill Erosion," examined an accounting firm whose automation was later removed and discovered staff could no longer perform core accounting tasks—exposing how automation complacency can hide catastrophic operational fragility until systems fail or change. The problem is especially urgent now as generative AI becomes ubiquitous and opaque, and organizations rarely surface failure cases until they’re forced to confront them. The authors argue businesses should actively design to preserve human-in-the-loop skills: technical and organizational control points, periodic audits requiring output justification, explanation features, automation-free training environments, and collaborative workshops to surface edge cases. They link the issue to prompt engineering—crafting prompts and validating outputs require deep, contextual domain expertise, and model drift (or platform changes) can make once-reliable prompts fail. Junior staff are particularly vulnerable because automation can deprive them of on-the-job learning. The recommendation: run scenario analyses, maintain critical skill capital, and blend mindful engagement with automation to prevent brittle, long-term dependency on AI systems.