SecondBrain
Ask the Brain
Index/Conceptupdated Sat Jun 27 2026 08:00:00 GMT+0800 (Philippine Standard Time)

Standardized vs Open Tasks

task-designworkflowproductivityorganizational-designapprenticeshipbank-of-korea

Standardized vs Open Tasks

The policy framework in BOK Issue Note 2026-12 for organising work so AI delivers actual productivity. The distinction is a sharper, more actionable version of the older economics "routine vs non-routine" split.

The distinction

Standardized tasks Open tasks
Output form Clear, evaluable against fixed criteria Not predetermined; varies with performer's judgment
Performer variation Low High
Examples Summarising reports, bookkeeping, regulation review, organising data New business planning, strategy, policy design, R&D, complex problem-solving
AI's role Can handle a substantial part of execution Augments human judgment; doesn't replace it
Productivity lever Workflow redesign + reallocation of saved time Human capacity to integrate AI-expanded options
What collapses if you ignore it Saved time turns to idle/leisure (no output gain) AI takes shortcuts past judgment (no quality)

Caveat the paper itself flags: "in practice most work exists in hybrid forms" — error costs, verification requirements, and degree of data standardization are the correction variables.

What each side requires

Standardized tasks — redesign workflows around AI

The point is not "use AI on these"; that just lowers per-task time. The point is to make AI central to the workflow:

  • Standardize inputs (data, formats, schemas).
  • Modularize the work into units AI can own.
  • Predefine verification + exception handling so the human role is goal-setting, result-verification, exception-handling, not direct execution.
  • Build reallocation mechanisms — explicit routing of saved time to high-value-added activities (new clients, additional analysis, new business). Without this last step, the time disappears (this is the AI Productivity Disconnect's "incentive misalignment" channel).

Open tasks — augment, don't substitute

In open tasks the same problem yields different results depending on judgment, experience, value-priors. AI's role:

  • Expand the option space — drafting alternatives, exploring framings, surfacing analogies.
  • Humans interpret, evaluate, decide. The leverage compounds with the human's domain depth, not with AI usage hours.
  • The more an org's work is weighted to open tasks, the more the constraint becomes growing judgment + problem-solving + domain knowledge alongside AI adoption, not just rolling AI out.

The apprenticeship problem

This is the part of the framework that travels furthest beyond the paper.

Historically, standardized tasks were the learning rung — data organising, basic analysis, standard report writing — that produced juniors who then graduated into open work. They served two purposes simultaneously: production AND skill formation.

If AI absorbs the standardized tasks, short-term productivity rises, but the long-run supply of people capable of open work erodes. The paper's countermeasures:

  1. Bring juniors into open tasks early via an observe → assist → lead progression — accumulate judgment and domain intuition without going through the standardized-task pipeline.
  2. Reinvest senior time saved by AI into mentoring, coaching, pair work — not just into more senior output.
  3. Even for AI-automated tasks, structure learning around "why these results emerge" — verification and exception-handling skills get accumulated from the junior stage.

This is the direct counterpart to the apprenticeship-pipeline concern in Designing IT Roles for an AI Era (Talent Strategy POV), and it's now backed by macroeconomic policy reasoning, not just AI-and-jobs commentary.

note The market may be funding the apprenticeship redesign, not blocking it AI Adoption and Headcount Growth (AI Companies Are Hiring More (AI Daily Brief), 2026-07-02) reports entry-level headcount growing ~12% at high-AI-adoption firms — adopters are hiring more juniors, selected for AI fluency. That's the demand that makes "observe → assist → lead on open tasks from day one" implementable rather than aspirational: the juniors are already coming in. The risk shifts from "will there be junior roles?" to "will those roles be designed as AI-leveraged apprenticeship or AI-bypassed solo work?" — which is exactly this framework's question.

Why the distinction is useful (and where it's still rough)

  • Useful: it tells you what kind of intervention closes the AI Productivity Disconnect in this domain — workflow redesign vs judgment-development. Most "AI strategy" decks conflate the two.
  • Rough: most real work is hybrid (one process has standardized sub-tasks and open sub-tasks). The paper acknowledges this. Treat the framework as a classifier for sub-tasks, not whole jobs.
  • Anti-pattern: introducing AI uniformly across both sides "without distinguishing" leads only to accumulated efficiency improvements at the individual task level, without fundamental changes in workflow — i.e. straight into the disconnect.

Connects to your work

For Manila IT and adjacent enterprise IT contexts:

  • Workflow redesign is owned by someone, or it isn't. The framework gives a clean criterion for which workflows justify a redesign owner (high standardized-task density + currently-rigid sequencing) vs which roles need an upskilling investment (open-task-heavy roles where AI is augmentation).
  • The apprenticeship issue lands hard in IT. Junior roles (data, basic analysis, standardized reports) are the rungs the framework names. The "observe→assist→lead on open tasks" prescription is implementable as a graduate-track curriculum.
  • Pairs with Designing IT Roles for an AI Era (Talent Strategy POV) — three-spine model + skill repricing was the axis; this gives the task-level operating rule underneath each spine.

Cross-references

Sources