Accountability check-ins

Report against commitments on a schedule, so promises to yourself get the same seriousness as promises to a lab partner.

When to use this: At the end of a session
Learning impact: Unlocks better later work
Learning evidence: Credible with caveats
Where you are with the topic: Checking readiness
How often to use it: Build and keep as a habit

When to use it

You make commitments and need a scheduled report against them, not another private intention.
Check-ins always “went fine” with nothing to show.
You want a short cadence that asks for artefacts, not feelings.

What to ask AI

Set a check-in prompt: commitments, evidence of done/not done, one repair for misses.

Forbid “went fine” without artefacts (problems attempted, pages recalled, verified units).

Keep the ritual short enough to survive midterm week.

How it helps

Accountability works when reporting is scheduled and evidence-based. Private intentions evaporate.

AI can act as a check-in partner that refuses vibes and asks for artefacts.

The repair for a miss should be specific and soon, not guilt and a bigger promise.

How to run it

Write three commitments for the week with done-when criteria.
Schedule check-ins (e.g. two evenings) before the week starts.
At each check-in, report hit/miss with evidence; repair misses with a dated replacement action.
Do not add new commitments until misses are repaired or dropped deliberately.

Example exchange

You

Weekly check-in. My commitments were three closed-book mixed sets by Thursday, error-log analysis on Sunday, and phone in the kitchen for two evening blocks.

Take the first commitment only. Done or not, with an artefact, no mood summary.

You

Did two sets. Scores were 6/10 and 7/10.

Miss on count. Now the Sunday analysis, done or not?

You

Not done.

Schedule a twenty-minute Sunday block now, or drop it deliberately. Don't replace it with a bigger promise. What time?

You

Sunday at 4pm.

Noted. We'll check the phone blocks next check-in with the same evidence rule.

Copyable prompt

Accountability check-in.
My commitments were: [ ].
For each: done/not done + evidence artefact. Refuse "went fine."
For each miss: one dated repair action (or deliberate drop). Keep reply short.

The Tell

Check-ins that always “went fine” with nothing to show are social soothing, not accountability.

If misses only produce bigger future promises, the loop is punishing you without changing behaviour.

Principle evidence

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

Moderate: external or scheduled accountability supports follow-through in self-regulation research, especially when paired with specific goals. Evidence beats reassurance.

AI delivery evidence

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

AI check-ins are only as hard as you allow. Models often soothe. Instruct refusal of vibes. This delivery pattern is speculative as a product claim.