---
id: "appraisal-xr-targeted-not-universal"
type: "appraisal"
source_timestamps: ["Enrichment Overlay"]
tags: ["counter-perspective", "learning-science", "ai-roi"]
related: ["framework-xr-modality-selection", "concept-forgetting-curve", "claim-ai-roi-failure"]
sources: ["reskilling"]
sourceVaultSlug: "hbr-seg-reskilling"
originDay: 10
articleStem: "hbr-edu-33-new-tools-workforce-training"
sourceUrl: "https://hbr.org/2025/12/the-new-tools-that-can-improve-workforce-training"
sourceTitle: "The New Tools That Can Improve Workforce Training"
---
# Appraisal: XR Is a Targeted Tool, Not a Universal Solution

## Appraisal: XR Is a Targeted Tool, Not a Universal Solution

The article's strong pro-XR stance benefits from three tempering counter-perspectives from the broader literature.

**1. XR wins are context-dependent.** Meta-analyses show VR is highly effective for **procedural and high-risk** skills, but its advantage over *well-designed* non-VR training is often **modest**. For knowledge-based content, **spaced repetition and retrieval practice** (Cepeda et al.; Roediger & Karpicke) and simple **job aids** can match or beat costly XR. Poorly designed XR can overwhelm learners or distract from the objective. → Use the [[framework-xr-modality-selection|selection matrix]] to decide *whether* XR fits at all, not just *which* XR.

**2. AI ROI failure is multi-causal.** The source pins missed AI returns on training gaps ([[claim-ai-roi-failure]]). Gartner and the productivity-paradox literature (Brynjolfsson) add **data quality, misaligned use cases, unrealistic expectations, governance, and integration** as co-equal causes. Upskilling is necessary but not sufficient.

**3. Situated-learning theory supports the *diagnosis*, not necessarily the *prescription*.** Lave & Wenger and Argyris & Schön validate that passive training fails — but the remedy is *practice in realistic contexts*, which XR is one (expensive) way to deliver, not the only way.

**Implication for a downstream agent:** endorse XR as a **targeted intervention** for the skill classes in the matrix; flag lower-tech alternatives for cognitive/knowledge content; and pair any XR recommendation with data/governance/change-management advice for AI initiatives.