Load-balancing request

Tell the AI your current working-memory limit and get an explanation cut to exactly that size, so you stop accepting firehoses on tired evenings and thimblefuls when you could handle more.

When to use this: Before you start studying
Learning impact: Unlocks better later work
Learning evidence: Credible with caveats
Where you are with the topic: Meeting the idea
How often to use it: Worth repeating often

When to use it

You are about to start a new unit and want a plan before you open the textbook.
You keep meaning to revise, but nothing on your calendar says when.

What to ask AI

Type something like: “My working-memory budget right now is … new moving parts / … minutes.”.

Tell it your current level and ask it to stay one step harder than comfortable.

Keep the hard part on your side: you retrieve, attempt, explain, or decide before you ask for help.

How it helps

Cognitive load is state-dependent.

The same topic that is tractable at 10am is impossible at 11pm after three lectures.

AI does not know your state unless you say so.

Load-balancing is the explicit statement: how many new moving parts can I afford right now? matched to an explanation with that many parts.

This sits beside Minimal-sufficient explanation (minimal-sufficient) and Multi-level explanation (multi-level).

Minimal-sufficient explanation minimises for correctness; A24 minimises for today's capacity; Multi-level explanation measures where you break across sophistication.

Use A24 when the constraint is fatigue, time, or anxiety, not when the constraint is conceptual opacity alone.

How to run it

Next: declare a budget: e.g. "three new ideas," "one mechanism," "five minutes of reading.".
Forbid overflow: "If it won't fit, say what to cut, don't sneak extras.".
Next: consume, then restate within the same budget.
Next: schedule the rest for a higher-capacity session rather than "just one more paragraph.".

Example exchange

You

My working-memory budget tonight is about twenty minutes and room for maybe three new moving parts. Teach within that. Don't dump a full chapter on PCR.

Then tonight's three parts: what denaturation does, what primers do, and why the temperature cycles. Which one do you already half-know?

You

I know heat separates DNA strands. Primers are fuzzy.

We spend the minutes on primers only. After that, stop, denaturation can wait for a later session if you're full.

Copyable prompt

My working-memory budget right now is [N] new moving parts / [N] minutes.
Explain [TOPIC] within that budget. If the true minimal account needs more,
tell me what won't fit and stop, do not smuggle it in. Then ask me to restate
within the same budget.

The Tell

Here is how you know this method has flipped: you "finished" a topic in a low-budget session and tomorrow cannot restate even the budgeted core.

In that moment the AI (or the schedule, or the story) did the thinking, and you only recognised a finished product.

Tighten the prompt for load-balancing request so you attempt, decide, or retrieve before anything is handed to you.

Principle evidence

Strength of the underlying learning idea, not a claim about AI products.

The underlying learning idea is rated moderate. Explicitly budgeting working-memory load is applied cognitive load theory (Sweller): match instructional density to current capacity. Self-stated limits are rough proxies. Principle moderate as load management; the request itself is an enabler, not a practice effect. AI compliance with “cut to this size” is speculative.

AI delivery evidence

Whether an AI tutor delivers this method well is a separate question.

The learning principle may be moderate, but that does not prove an AI session delivers it well. Treat AI delivery as speculative unless a study of tutoring with this method is named, and none is claimed here.