---
id: "concept-jagged-frontier"
type: "concept"
source_timestamps: ["§ Step 4. Deliver the output with an explanation of how you and AI arrived at it."]
tags: ["ai-capabilities", "evaluation"]
related: ["concept-reasoning-trail"]
definition: "The uneven boundary between what AI models can execute flawlessly and where they fail, which varies heavily by specific domain and task type."
sources: ["reskilling"]
sourceVaultSlug: "hbr-seg-reskilling"
originDay: 10
articleStem: "hbr-edu-32-help-employees-get-better-with-ai"
sourceUrl: "https://hbr.org/2026/06/help-employees-get-better-not-just-faster-with-ai"
sourceTitle: "Help Employees Get Better—Not Just Faster—with AI"
---
# The Jagged Frontier

The **jagged frontier** — a term coined by researchers — describes the *uneven, highly contextual boundary* between tasks AI handles exceptionally well and tasks where it completely falls short. The boundary is not smooth; it zig-zags by domain and task type.

In the [[framework-four-step-ai-development|four-step model]], professionals end their [[concept-reasoning-trail|reasoning trail]] with a one-sentence assessment of the jagged frontier for their specific domain — e.g., *'On this type of task, AI is good at X but struggles with Y.'* Accumulating these observations over time lets a professional calibrate exactly where to trust AI and where to scrutinize it heavily.

The enrichment overlay treats the jagged frontier as an established *adjacent framework* for the article's 'where AI is good vs. where it struggles' guidance. It is closely related to the error class described in [[concept-looks-right-but-isnt|looks right but isn't]].