Concept

Tools can help you make better work, as long as you stay the one who judges what the work is for, whether the reasoning holds and what the evidence shows.

Issue

A recommendation can sound right before you know whether its main reason holds up.

Solutions

Work out what your work needs, then check one task before relying on the result.

Say you’re writing a memo. It recommends a new reporting dashboard. You ask an assistant to improve the wording. The new version sounds clear and confident. But you still can’t explain why a dashboard would fix the delay you’re worried about.

The wording may be better. The missing explanation still matters. Maybe collecting the figures is what causes the delay. Maybe the problem sits somewhere else. Each answer points to a different choice.

You don’t have to finish all your thinking before you use a tool. Help can give you ideas, or better ways to say them. But before you rely on an answer, you need enough to judge whether it fits the job.

This module is for everyday knowledge work: the kind where you can check the result and change the decision.

What you’ll get from this module

In this module

By the end of this module, you’ll have a clearer way to work out what your work needs and to choose help that fits. We’ll look at research on writing help, on planning and explaining ideas, and on how sure people feel they understand something, plus one analysis of people checking machines. Then you’ll pick a situation from your own life and work through a short written exercise. You’ll finish with a task brief, a way to check the result, or one question to answer before you rely on it.

What makes help useful?

A writing experiment: help can improve the work.

In 2023, Shakked Noy and Whitney Zhang tested ChatGPT on professional writing tasks. They gave 453 college-educated professionals tasks from their own kind of work, with rewards for doing well. Half were picked at random to use the chatbot.

Average time fell by 40%. The quality of the work rose by 18%. The gap between stronger and weaker writers narrowed too. The help made the finished work better and faster.

That’s a reason to welcome useful help, even when you couldn’t produce the same quality alone. The results cover the professional writing tasks in the study. They measure better output, not whether people thought more clearly.

Why it matters: For your own work, the next question is what kind of help you need. Better writing and a sound reason for a recommendation are two different things. Check them separately.

Read the experiment on assistance with professional writing.

A writing experiment: an argument needs more than a list of ideas.

In 2018, Teresa Limpo and Rui A. Alves compared ways of planning in 63 undergraduates who were writing an argument. Before writing, each person either made an outline that showed how the argument fit together, listed their ideas, or did an unrelated task.

Only the outline improved the number of argument elements, how persuasive the text was and its overall quality. The researchers found no drop in the mental effort they measured.

A list tells you which ideas you have. An outline can show how those ideas support an argument. You can collect more notes and still be unable to explain why a reason supports your conclusion.

Why it matters: The study was about undergraduates writing arguments. The useful lesson here is the difference between collecting points and connecting them. You can give a draft that structure and still be willing to change the conclusion when you look at the evidence.

Read the study comparing writing plans and idea lists.

Which one connects the reason?

A learning review: explaining a connection does useful work.

In 2018, Kiran Bisra and colleagues brought together studies that asked people to explain the material to themselves while studying or solving problems. Self-explanation means working out the connections for yourself, such as why one step follows from another.

The review combined 69 effect sizes from 64 research reports. An effect size describes how much one group’s learning differed from another’s. Across the included studies, prompting people to explain had a positive average effect on learning. The tasks included learning from text, from worked examples and from problem solving.

The comparisons varied, and the studies didn’t always report equal study time. These are findings about learning, not about the quality of workplace memos. They give us a reason to work through the connection between ideas.

Why it matters: Restating a recommendation isn’t the same as explaining it. You can say the conclusion more smoothly without giving a reason for it. Explaining means showing why the conclusion follows from the evidence you have.

Read the meta-analysis of prompted self-explanation.

Does the reason hold up?

An explanation study: feeling you understand can change when you try to explain.

In 2002, Leonid Rozenblit and Frank Keil studied what they called an illusion of explanatory depth. In their first study, 16 Yale graduate students rated how well they understood 48 devices and phenomena. Then they explained four of the devices.

After trying to explain, people rated their understanding lower. The same people went through the whole sequence. There was no randomly chosen group that skipped the explaining.

The result is about how people rate their own understanding. Feeling that you know how something works is not the same as being able to explain it. When you’re making a recommendation, trying to explain the reason can show you a specific question to look into.

Why it matters: Finding it hard to explain isn’t a verdict on your ability. You may need information, a diagram, an example or help putting it into words. And a confident explanation still needs evidence behind it.

Read the study of explanatory depth.

A study of automation: a checking role has to be one a person can actually do.

In 1983, Lisanne Bainbridge wrote about the problems left for people when industrial processes are automated. In her section on monitoring, she asks whether a person can really check the decisions a computer makes.

Suppose a computer makes decisions faster, and from more factors, than a person can follow. Making that person responsible for checking every decision as it happens doesn’t give them the ability to do it.

This is an analysis of ideas about industrial control, with examples from aircraft flight decks. For ordinary work, it raises a useful question: do you have the information or the expertise you need to check the answer?

Why it matters: Being told to review a document isn’t enough if you can’t inspect its important evidence or judge its reasoning. You may need a source you can look at, a concrete test, or someone qualified to check that part.

Read the analysis of human monitoring in automated systems.

What this means for you

Use help where it makes the work better, and keep a way to judge the result. First, work out what’s missing: a clear purpose, information, reasoning, ideas or words.

Give the work a purpose and a standard

Before you ask for help with an important draft, write down who needs it, what they need to decide or do, and any limits the work must respect. If you don’t know an answer, put that question into the task.

Then describe what a useful result must include, and what would make it unacceptable. Keep the facts you’ve checked apart from your assumptions and from what’s missing. Now you have something concrete to compare the answer against.

Take the dashboard memo. “Make this persuasive” leaves the main question open. A useful memo would name the cause of the delay and explain how the dashboard would deal with it. If the cause is still an assumption, the recommendation needs more work. Better wording can’t supply that missing observation.

Your first brief can change. If the evidence points to a different problem, change the task instead of defending the original recommendation. A conversation or a tool can help you work out what’s unclear.

Connect the reason to the recommendation

Sketch the claim or option, the reasons and evidence behind it, the strongest objection or alternative, and the next action. If you’re still exploring, write down what you need to learn instead of picking a conclusion too soon.

Now explain one connection the recommendation depends on. Why does that reason favor this option? Why is the alternative a worse fit? What observation would change your choice?

For the dashboard memo, collecting figures would have to be part of what causes the delay. The dashboard would also have to reduce that work. Both connections need support before they can justify the recommendation.

If the delay happens somewhere else, a smoother explanation won’t make the original reason fit. Find out where the delay happens before you decide whether the dashboard would help.

That’s what an outline adds to a list. “Dashboard,” “delay” and “efficiency” are topics. Connecting them shows what must be true for the proposed solution to make sense.

Turn a hard explanation into a question you can look into

Try plain words, a diagram or a worked example. Look for the point where you need a fact or a connection you can’t yet supply.

In the made-up memo, “the dashboard will make reporting efficient” repeats the result you’re hoping for. “We need to find out which part of reporting creates the delay” names something to investigate.

If the trouble is the wording, use language help. If it’s a missing fact, find evidence. If judging that evidence takes expertise you don’t have, find someone who can help.

The aim is to name what’s unclear and find a way to work on it. A halting explanation doesn’t score your intelligence, and a polished one can still need correcting.

Match the help to the gap

Ask for help with the part that’s missing. Coming up with ideas, organizing them, improving the wording and doing the work are different jobs. You can use help without expecting yourself to produce the same result alone.

Compare the answer with the purpose and criteria you wrote down. Does it answer your question, or does it make a different answer look more attractive?

In the made-up memo, help with the wording fits when the evidence and reasoning are sound but the explanation is hard to follow. If you don’t know what causes the delay, you still need to find out.

You may find a better question while you’re using the tool. Update the brief when that happens. Keep it clear which points are checked facts, which are assumptions and which you don’t know yet.

Choose a check you can actually do

Trace the important factual claims and calculations back to evidence you can inspect. Ask what alternative, or what changed assumption, would weaken the conclusion. Where you can, try a small test you can undo before you take a big step.

For the dashboard, first check whether collecting figures causes the delay. If the evidence supports that link, a limited test can show whether the proposed change deals with it. If the link fails, think again about the recommendation.

Choose a check you have the knowledge to carry out. A source you can’t judge, or a test you can’t interpret, still leaves a gap. Ask someone qualified to review that part, or leave it open before you rely on the result.

Answering your own questions doesn’t prove the work is right. Use them to find evidence and possible failures. Keep the effort in proportion. Trivial tasks don’t need elaborate checking. High-stakes decisions, or ones you can’t reverse, may need a specialist’s review beyond this exercise.

What do you ask your tools for?

How about your daily life?

Where are you trying to turn an idea into something you can use? Perhaps you’re drafting a recommendation, explaining a plan, or improving a piece of writing with help.

Do you need better words, a clearer reason, missing evidence, or a way to check the answer?

List situations you’d like to understand or handle better. One good idea is enough. Add more if it helps you choose.

A situation in my daily lifeWhat I want to understand or improve
________________________________________
________________________________________

My chosen situation: ____________________

Choose everyday knowledge work where you can check the result and change the decision. Take it into the exercise below. You can change your choice if it turns out to need a different kind of review.

Check one task before relying on the result

Start with your chosen situation. If you came straight here, choose an ordinary writing, planning or explaining task: one with a result you can check and a decision you can change.

High-stakes decisions, or ones you can’t reverse, may need a specialist’s review beyond this exercise. Keep the work in proportion; a trivial task doesn’t need an elaborate process. You can use a conversation or a tool to help you make the task clearer.

  1. Define the job. Write down who the work is for, what it should help them decide or do, and any limits it must respect. If something is unknown, name it.
    Purpose and constraints: ____________________
  2. Describe a useful result. Write what the answer must include and what would make it unacceptable. Keep checked facts, assumptions and missing information apart.
    Criteria and what is known or missing: ____________________
  3. Sketch the reasoning. Set out the claim or option, the reasons and evidence, the strongest objection or alternative, and the next action. If you’re exploring, say what you need to learn instead of forcing a conclusion.
    Reasoning or question to investigate: ____________________
  4. Explain one connection that matters. Use plain words, a diagram or an example to show why the recommendation follows. Say what would change your choice. If the explanation breaks down, name the missing fact or connection.
    Use language help for the wording, evidence for missing information, or qualified help when the judgment is beyond your expertise. Finding it hard to explain is a reason to investigate, not a judgment of your ability.
    Connection and any gap to resolve: ____________________
  5. Choose the help and check the result. Ask for the ideas, structure, wording or doing that you need. Compare the result with your criteria and inspect the important claims and calculations. Use a small test you can undo where you can.
    If you can’t judge an important part, find the right help or leave it open before you rely on the result. If the evidence changes the task, revise the brief. Answering the prompts alone doesn’t prove the result is right.
    Assistance, actual check and next action or unresolved gap: ____________________

Ask Jing

Become a Jingius.

Check one task

Loading interactive tool…

Your next step

Finish with a short brief and a way to check the result that you can carry out yourself. If an important question is still open, name the source, test or person you need to help settle it.

A note on the evidence

These sources look at different things: how well professionals write, the quality of students’ arguments, learning, and how people rate their own understanding. The monitoring paper makes an argument about industrial control. Together, they help us tell useful help, explanation and checking apart. They don’t measure one shared quality called “better thinking.”

The learning evidence gives us a reason to work through the connections that matter. Time spent explaining isn’t a measure of success in itself. Focus on an important question you can look into.

How we prepared this module

This module brings together writing experiments, a review of learning studies, an explanation study and an analysis of people checking machines. The made-up dashboard memo shows how a clearer sentence can still leave an important reason unsettled. The exercise helps you look at one task and find a way to check its result.

Sources