Skip to content
The Goodwill Ledger
← All frameworks A framework

Accountable Intelligence

The unit of management for the human-AI organisation.

First proposed ·

It is the proposed replacement for span of control as the unit by which management is understood and measured in the human-AI organisation. What was implicit before must now become explicit.

From Span of Control to Accountable Intelligence

What was implicit before must now become explicit

Old model

Span of Control

Management is measured mainly by direct reports

Risk, checking and judgement existed, but often stayed implicit

New model

Accountable Intelligence

Management is measured by what a human can stand behind

Risk, checking and judgement must be explicit, adaptive and scalable

What it replaces

Span of control was the silent accounting system of management for the better part of a century. It measured how many people reported to a manager, how much budget passed through their authority, how much activity sat under them. It worked when work was a chain of human acts that a manager could observe, sample and verify by presence.

That picture no longer holds. Inside most organisations today, work is already produced by combinations of people, copilots, automations, AI assistants and embedded vendor systems. The headcount under a manager has stopped capturing what they are actually responsible for.

Accountable intelligence is the replacement unit. It moves the central question of management from how many people do you control to how much work, done by people and AI, can you safely stand behind.

How to recognise it

Four questions, taken together, make accountable intelligence visible. They are deliberately simple. The aim is to make explicit what experienced managers used to handle implicitly.

The Four Questions of Accountable Intelligence

Together they make the implicit explicit

  1. 1

    Scope


    How much work, done by whom and what?


    People, copilots, automations and agents, considered together

  2. 2

    Consequence


    What happens if the work goes wrong?


    Activity, data, decision and the harm that follows from each

  3. 3

    Verification


    Can the output be checked?


    Sampling, escalation, challenge and incident detection by design

  4. 4

    Ownership


    Can the human stand behind it?


    Enough understanding, authority and willingness to take responsibility

Strength in one question cannot substitute for absence in another

Strength in one question cannot substitute for absence in another. A manager with wide scope and no verification capacity does not have accountable intelligence over the work. They have a label.

What it implies for management

The framework changes what managers actually do. Five practical shifts follow.

What the Manager Tracks

From counting acts to seeing the work

Old measures

Span of Control

  • Headcount

    Number of people reporting

  • Budget

    Money under authority

  • Activity

    Meetings, reports, throughput

  • Deliverables

    Outputs produced by the team

Counts what is supervised

The measure of activity

New measures

Accountable Intelligence

  • Work in scope

    People and systems together

  • Risk profile

    Activity, data, decision, consequence

  • Verification design

    Sampling, escalation, challenge

  • Stand-behind capacity

    Understanding, authority, willingness

Counts what is owned

The measure of accountability

Framing of work moves from who is doing this to what is being produced, by what combination of people and tools, with what risk and what verification.

Delegation moves from human-to-human transfer to designed delegation across people and systems, with explicit limits, escalation rules and consequence handling. A manager who delegates to an AI system the way they delegate to a person is making a category error.

Verification moves from informal sampling to designed verification. Sampling regimes, paired reviews, challenge sessions and adversarial checking become part of the operating model, not occasional management hygiene.

Risk literacy moves from familiar categories to new ones. Hallucination, prompt manipulation, model drift and training-data leakage become part of the working vocabulary. The accountable person who does not understand how their system can fail is, in the limit, not the accountable person.

Performance measurement moves from activity and throughput to verified output, risk events, adoption and quality. Output can be increased by quietly outsourcing more of the work to systems whose risk is not visible; performance measurement has to track what the throughput is hiding.

Where it came from

The Accountable Intelligence framework was first set out in the essay Accountable Intelligence: What Replaces Span of Control. The essay carries the full argument. This page carries the framework itself.

The four stages

  1. 01

    How much work, done by whom and what

    Scope

    People, copilots, automations and agents, considered together. The honest answer to the question of scope is no longer a single number. It is a composition of the human and machine work that sits inside a role.

  2. 02

    What happens if the work goes wrong

    Consequence

    The activity, the data, the decision and the harm that follows. A team using AI to draft meeting notes is not in the same position as a team using AI to support a regulator-facing communication, even if the tool is the same.

  3. 03

    Can the output be checked

    Verification

    Sampling, escalation, challenge and incident detection designed into the work itself. Without verification capacity, accountability is nominal rather than real. Verification is the difference between a label and a job.

  4. 04

    Can the human stand behind it

    Ownership

    Enough understanding, authority and willingness to take responsibility. Without it, accountable intelligence collapses into responsibility-shopping, in which each layer of the organisation insists that the layer below, or the vendor, or the tool itself, owns the outcome.