The Candy Machine Trap

​

The Candy Machine Trap, the bottom-right Maintenance × Shallow quadrant of the AI Allocation Matrix where most senior leaders’ AI hours cluster — routine email, document polish, meeting summary, the default destination.

​

The Candy Machine is the analogy. Cheap, quick, a small hit of competence on demand. Drafting routine emails, summarising meetings, polishing the slide that did not need polishing. Friction-free, addictive, and the default destination of AI hours when nothing pulls them elsewhere.

​

Two failure modes flow from the Trap. Lopsided allocation, the tool of highest leverage applied to the work of lowest leverage, so the institution pays for AI capability that is spent on tasks that never compound. Invisible displacement, the Candy-Machine hour crowds out the hour that might have been spent in Compounding leverage, and because neither hour is tracked, the displacement is never seen.

​

Genesis

Cal Newport (2016, Deep Work) made the case for deep work before generative AI was widely available: cognitive engagement that develops judgement, produces distinctive contribution and creates value the market does not commoditise. Elizabeth Grace Saunders (2013, The 3 Secrets to Effective Time Investment) draws an adjacent line in time-allocation terms, splitting the working week into investment activities (which compound) and maintenance activities (which do not). The Candy Machine Trap stacks the two as orthogonal axes; the matrix that results is the visual signature of the framework.

​

Sir Thomas Gresham’s law of money (1558), bad money drives out good, applied to work, names the mechanism: easy or routine work drives out hard or non-routine work whenever a leader has discretion over how time is spent. The Candy Machine is Gresham’s law of work, accelerated by a frictionless tool.

​

Why it matters now

Pre-AI, the Gresham dynamic was modest. Routine work still carried a small frictional cost; even the easy task asked something of the leader. AI removes that cost almost entirely. The activation energy for Candy-Machine work falls to near-zero. The activation energy for Compounding-leverage work, which still requires a leader to think, frame and judge, stays the same. The result is a structural pull on the leader’s week: AI’s frictionlessness moves hours from where leverage lives to where it does not.

​

The leaders getting most from AI are not the ones using it most. They are the ones using it in the Compounding-leverage quadrant.

​

The moves

Three moves relocate AI hours from the Candy Machine to Compounding leverage.

See where the hours go. Most leaders cannot answer the question ‘of the AI hours you spent this week, what fraction went to deep-investment work?’ Begin by tracking each AI working session against the four quadrants. Without the picture, the trap is invisible.

​

Refuse the easy-first impulse. When AI is reached for on a maintenance task, ask first: ‘is there a deep-investment task I am avoiding by doing this?’ Often there is. The maintenance task can wait an hour.

​

Apply AI where the leverage is, not where it is easiest. The strategic framing work that takes a leader four hours of thinking is the work where AI’s reframings, alternative viewpoints and adversarial scrutiny are most valuable. The polishing of a routine email is where they are not.

​

How ChangeSchool applies it with executives

We run senior teams through the Allocation Audit in three-hour cohort sessions. Bring a calendar week’s worth of AI prompts (or a screenshot of recent chat history). Map each AI working session onto the 2×2, by intent, not by output. Notice which quadrant got the most hours; notice which got least; notice the gap between where AI was used and where it could most have helped.

​

The discipline

The Allocation Audit surfaces the trap once. Three habits embed the discipline.

​

A weekly allocation review, five minutes on a Friday looking at the week’s AI prompts and assigning each to a quadrant. The picture builds over months and the drift becomes legible.

​

A deep-investment booking, at least one calendar block per week where AI is brought to a strategic question, not a routine one. The block is held against the Candy Machine pull, and the compounding hour stops being optional.

​

A friction-asymmetry log, capturing the moments when the easy task was reached for because it was easy, and a deep-investment task was visibly avoided. The log surfaces the trap mechanic in the leader’s own behaviour, where it can be interrupted next time.

​

The discipline is not to use AI less. It is to make AI hours appear where they generate compound returns, and refuse the gravitational pull that puts them where they do not.

​

​

“AI gravitates to the work that matters least, away from the work that matters most.”

 

 

Viren Lall, Managing Director,

ChangeSchool LDN (2026).

virenlall.com/candy-machine-trap

​

​

​