Watch: the module in 1:59
Read the transcript
Most organisations can tell you what they spent on AI last week. Very few can tell you what it produced last quarter. This course is about closing that gap.
The reason is simple. The two meters run at different speeds. Cost is metered to the token, in real time, by every provider. Value has no meter until someone builds one: a baseline, a quality threshold, an owner, and an outcome in a business system.
Where nobody builds that meter, the organisation governs a growing cost base with activity data alone. This site calls the difference the AI value gap. Attribution coverage is the share of spend that does have a baseline, an owner and evidence of change. The gap is everything else.
A token tells you something was consumed. It does not tell you the answer was right, that anyone used it, or that it was worth anything. And as tokens get cheaper, invoices tend to grow, because use grows faster than prices fall.
So follow the chain. Compute becomes tokens, tokens become outputs, outputs become accepted work, and accepted work becomes a business outcome. Each link has its own conversion rate, and each can fail while the token count stays the same.
The course places organisations on five levels of maturity. They measure how well AI is governed, not how much is in use. An organisation with forty pilots can still be at level one.
In the exercise, you will place a real-sounding organisation on that ladder and name its next move. Then the readings take you through each idea in depth.
Six modules, one exercise each, and a certificate at the end. Let's start.
By the end of this module you can
- Explain the AI value gap and the two-meter asymmetry behind it.
- Distinguish the token as a unit of consumption from the outcomes that carry value.
- Trace a workload through production, consumption, value and decision.
- Place an organisation on the five levels of AI economics maturity and name its next move.
Everyone. This module sets the vocabulary for the rest of the course.
The argument in brief
Most organisations can say what they spent on AI last week. Few can say what it produced last quarter. That difference is the subject of this course, and it has a name on this site: the AI value gap.
The gap exists because the two meters run at different speeds. Cost telemetry is metered to the token, in real time. Value telemetry has to be designed: a baseline, a quality threshold, an owner and an outcome that sits in a business system. Where nobody designs it, the organisation governs a growing cost base with activity data alone.
This module sets out the argument and the vocabulary the rest of the course uses. Read the value gap essay first, then the token piece, then the map, which shows the four jobs of AI economics: produce intelligence, consume it in work, measure the result and allocate the next pound.
Key ideas
- Attribution coverage is the share of AI spend linked to a baseline, an owner and evidence of change. The value gap is its complement.
- A token records consumption. It does not record that an answer was right, used or worth anything.
- Every link in the chain from compute to enterprise value has its own conversion rate, and each can fail without the token count changing.
- Maturity is about governance quality, not the volume of AI in use. An organisation with dozens of pilots can still be at level 1.
- Cheaper tokens tend to mean bigger invoices, because usage grows faster than unit prices fall.
Terms to know
- AI Value Gap
- The difference between what an organisation spends on AI and the share of that spend it can connect to a measured outcome: the complement of attribution coverage.
- Token
- The unit in which AI models read and write text, and the unit in which most AI usage is metered and priced. A measure of consumption, not of value.
- Token Productivity
- The business value generated per token consumed-a measure of how efficiently AI workflows convert token consumption into meaningful outcomes.
- Jevons Paradox
- The tendency for total consumption to rise when efficiency improvements make a resource cheaper or easier to use.
- AI Value Management
- The discipline of managing AI cost, usage, risk, adoption, quality, and realised business value across an enterprise portfolio.
Read
In this order. Each opens on the main site.
- 1The AI Value GapThe core argument, the evidence and how to compute attribution coverage.25 min
- 2The Token Is the Meter, Not the ValueWhy consumption is not value, and the conversion chain.10 min
- 3The AI Economics MapProduction, consumption, value and decision, with one workload traced through.12 min
- 4The Five Levels of AI EconomicsA maturity model built on governance capability.15 min
Worked example: one workload through the whole chain
Illustrative figures. A claims-handling team uses an AI assistant to draft first responses to routine insurance queries.
| Production | A mid-tier model through a cloud provider and a shared retrieval platform. Platform share and engineering: £9,000 a month. |
|---|---|
| Consumption | 60,000 drafts a month. Model and retrieval cost £4,800 a month, about 8p a draft. |
| Observe | Handling time per routine query fell from 11 to 7 minutes against a three-month baseline. |
| Net | 9% of drafts need substantial rework and complaint rates are unchanged. Net saving about 3.3 minutes a query, roughly 3,300 hours a month. |
| Convert and book | The team absorbs growing volume without two planned hires. Booked as avoided cost of about £7,500 a month, not as the hours multiplied by salary. |
| Decision | £7,500 of value against £13,800 of cost, with before-and-after evidence (E1). |
The result
Redesign, not stop. The case is negative mainly because of the shared platform charge. Test a cheaper model against the same quality floor, check the platform allocation, and add a holdout group to lift the evidence to E2 by the next review.
Without the full chain, this team would have reported 3,300 hours saved and asked to scale.
From The AI Economics Map, where the figures are explained in full.
Try it: place an organisation on the five levels
A logistics group has consolidated its AI inventory and reports £3.1m a year of AI spend across 11 tools, two clouds and four model providers. Spend is reported monthly by business unit. Nobody owns the AI cost envelope. Value is described as hours saved per employee from a survey. There is no baseline for any workload.
- Read the descriptions of levels 2, 3 and 4 on the Five Levels page.
- Decide which level the group is at, using the evidence in the brief rather than the size of its spend.
- Name the two actions that would move it to the next level, and the role that should own each.
Open the maturity assessment(opens in a new tab)
Show the worked answer
Level 2, Visible. Spend surfaces are known and reported, but nobody owns the envelope and value rests on self-reported time saved with no baseline.
Moves towards level 3: establish AI as a formal FinOps scope with a reporting cadence and cost per workflow for the top services (FinOps lead), and separate shared platform cost from local demand in the cost model (TBM lead). A named owner for the AI cost envelope, agreed by the CFO, is the governance step that makes both stick.
Reporting is not maturity. The test is whether someone can explain whether the spend is justified.
Check your understanding
Five practice questions. The final exam uses different questions in the same style.
Practice only. Nothing is recorded.
Progress is saved in this browser only.