AIYING FOR OUTCOMES
Unit 02 — Task ≠ Outcome
TIME: 15–20 minutes | LEVEL: Start | NEED: Any conversational AI | MODE: Standalone
1. OUTCOME FIRST
What will you accomplish?
By the end of this unit, you will be able to distinguish a task from its intended outcome and use that distinction to direct AI toward what actually matters.
The principle:
A task tells you what to do. An outcome tells you what you want to achieve.
2. PRACTICE FIRST
Do before you read
Don't ask AI yet.
Look at these requests:
Write an email to my team.
Prepare a presentation.
Compare three courses.
What do they have in common?
They describe tasks.
Now choose the first one:
Write an email to my team.
Predict three possible reasons someone might want that email.
Perhaps:
Get everyone to attend a meeting.
Persuade the team to adopt a new process.
Resolve confusion about responsibilities.
Same task:
Write an email.
Different outcomes.
Now take something you actually need to do.
MY TASK:
Without AI, complete this sentence:
I need to do this because I want __________________________ to happen.
Keep both statements. We'll return to them.
3. DISCOVER
The task is visible. The outcome may be hidden.
Generative AI makes completing tasks extraordinarily easy.
We can say:
Summarize this document.
Create a presentation.
Analyse this data.
Write an email.
Explain this concept.
And AI starts producing immediately.
But that speed creates a new problem:
AI can begin completing the task before we have clarified why the task needs to be done.
Consider:
“Summarize this 30-page report.”
AI can produce an excellent summary.
But why do you need it?
Perhaps you want to:
understand the report quickly;
identify risks before a meeting;
decide whether to approve a proposal;
prepare questions for the author;
explain the findings to somebody else.
Each outcome could require a different summary.
So we need to distinguish three things.
TASK → OUTPUT → OUTCOME
TASK
What needs to be done?
Prepare a presentation.
↓
OUTPUT
What gets produced?
A 10-slide deck.
↓
OUTCOME
What should become possible or change?
The leadership team understands the three priorities and agrees on next actions.
The task is a means.
The output is an artifact or response.
The outcome is the result we are pursuing.
Consider learning
Task: Study KNN.
AI Output: Explanation + examples + notes.
Outcome: Be able to explain how KNN works and correctly apply it to a simple classification problem.
AI might produce brilliant notes.
But if the learner still cannot explain or apply KNN independently:
Output achieved.
Outcome not achieved.
Consider work
Task: Analyse customer complaints.
AI Output: Categorized summary of complaints.
Outcome: Identify the recurring causes responsible for the most important customer problems and decide what to address first.
Again:
The analysis isn't necessarily the accomplishment.
The decision or improvement enabled by the analysis may be.
Consider everyday life
Task: Compare three laptops.
AI Output: Comparison table.
Outcome: Choose the laptop that best fits your actual requirements and constraints.
A 30-column comparison table might be an impressive AI output.
It might also make your decision harder.
The better question is:
Did the output help accomplish the intended outcome?
The hidden question behind every task
Whenever you encounter a task, try asking:
What is this task supposed to achieve?
Sometimes one Why? reveals the outcome.
Sometimes you need several.
For example:
Create a dashboard.
Why?
So management can see sales performance.
Why?
So managers can identify underperforming territories.
Why?
So they can intervene early.
The task started as:
Create a dashboard.
The outcome became:
Enable sales managers to identify underperforming territories early enough to take corrective action.
Now you can design a much better dashboard.
This is the Pattern AI relationship:
TASK → WHY? → PATTERN → OUTCOME
Look beyond the visible activity for the structure underneath it.
4. POWER PROMPT
Find the Outcome Behind My Task
I need to [DESCRIBE MY TASK].
Do not perform the task yet.
Help me identify the outcome behind it.
Ask me focused questions, one at a time, about:
why the task needs to be done,
who will use or be affected by the result,
what decision, action, understanding or change should follow,
what success would look like,
and how the task could be technically completed yet still fail in practice.
Then distinguish:
TASK — what needs to be done
OUTPUT — what will be produced
OUTCOME — what should change or become possibleState the proposed outcome in one sentence.
Do not proceed until I confirm that this is the outcome I actually want.
Notice that final instruction.
AI may help you discover and articulate an outcome.
You retain responsibility for choosing it.
5. PRACTICE EXERCISE
Find the outcome behind your task
Choose one route.
LEARN
Start with:
Task: Summarize this chapter.
Ask:
Why do I need the summary?
Perhaps:
Outcome: I should be able to explain the chapter's three central ideas without looking at the summary.
Use the Power Prompt and see whether your outcome changes.
WORK
Start with:
Task: Prepare a competitor analysis.
Ask:
What should this analysis enable?
Perhaps:
Outcome: Decide whether we should change our positioning before launching the product.
Now AI knows what the analysis is for.
LIFE
Start with:
Task: Plan a three-day holiday.
One person's outcome:
Have a relaxed family break with minimal travel.
Another person's:
Experience as much of the destination as possible in three days.
Same task. Different outcomes. Different plans.
Now return to your own example from Section 2.
Complete:
MY TASK
EXPECTED AI OUTPUT
MY ACTUAL OUTCOME
Now retrieve your original thinking without asking AI:
How has identifying the outcome changed what I would ask AI to do?
Write one sentence.
6. JUDGE
Don't let AI choose your outcome
Review the outcome AI helped you articulate.
Ask:
Is this genuinely what I want to accomplish?
Did AI confuse an output with an outcome?
What assumptions did AI make?
Whose outcome is this—mine, my teacher's, my customer's, my organization's, or somebody else's?
What important context does AI still lack?
Could the task be completed perfectly while the outcome still fails?
How would I know whether the outcome was actually achieved?
Now predict:
If I complete this task perfectly, what could still prevent the outcome from happening?
This helps expose the gap between completion and accomplishment.
Confidence check
How confident are you that you have identified the right outcome?
1 — Very unclear | 2 | 3 | 4 | 5 — Very clear
Below 4?
Clarify further before proceeding.
UNESCO's AI Competency Framework for Students explicitly places human agency, critical judgment, accountability, and examination of whether AI is fit for purpose within a human-centred approach to AI. (UNESCO)
The principle for AIYING is therefore:
AI can help clarify the outcome. Human judgment determines whether it is the right outcome.
7. ACCOMPLISH
Redesign the task around the outcome
Now give AI both your task and your confirmed outcome:
My intended outcome is [OUTCOME].
The task I originally planned is [TASK].
Before performing the task, assess whether this task is actually an effective way to achieve the outcome.
If there is a better approach, explain it.
Then help me proceed while keeping the outcome as the primary criterion.
AI might suggest that the original task should be:
changed,
simplified,
expanded,
or perhaps not done at all.
That's valuable.
Because Outcome-First thinking treats the task as a means—not an obligation.
When finished, evaluate three separate questions:
Was the task completed?
Yes / No
Was the output produced?
Yes / No
Was the outcome achieved?
Yes / No / Not yet known
If the third answer is No, don't hide behind the first two.
Ask:
Outcome → Current Result → Gap → Next Action
Then begin another cycle.
8. REFERENCES & EVIDENCE
What's behind this idea?
The sources below have been checked against the underlying publications or authoritative institutional sources.
Established evidence — Goal setting
Locke and Latham's 2002 American Psychologist paper summarizes 35 years of empirical research on goal-setting theory. Their work examines how goals affect task motivation and performance, including mechanisms, moderators and practical implications. This provides an established foundation for the importance of explicitly defining what one is trying to achieve. (PubMed)
Locke & Latham — PubMed record
Established educational principle — Outcomes and activities should align
Constructive alignment, associated with John Biggs, explicitly connects intended learning outcomes, learning activities and assessment. Monash University's explanation of the framework describes learning outcomes as foundational and stresses the intentional alignment of activities with those outcomes. (Monash University)
Constructive Alignment — Monash University
This supports the broader distinction between performing an activity and achieving the intended result of that activity.
Established evidence — Human-centred AI
UNESCO's 2024 AI Competency Framework for Students defines 12 competencies across four dimensions, including a human-centred mindset. The framework emphasizes human agency, accountability and critical judgment, including examining whether AI is fit for purpose. It also describes its core competencies as brand- and product-agnostic so they can transfer to future AI technologies. (UNESCO)
UNESCO AI Competency Framework for Students
Historical evidence — End-User Computing
Rockart and Flannery's 1983 paper, The Management of End User Computing, classified end users into six types and argued that different types required differentiated education, support and control. This supports the broader historical lesson that putting powerful computing capability directly into users' hands creates both opportunities and capability requirements. (Mendeley)
What is specifically the AIYING proposition?
The following formulations are proposed in this book, rather than claimed as established findings of the cited research:
TASK → OUTPUT → OUTCOME
and
TASK → WHY? → PATTERN → OUTCOME
as practical tools for End-User AI.
Likewise, Outcome-First AIYING is a proposed methodology.
The external sources provide related theoretical, educational and historical foundations. They do not constitute direct empirical validation of AIYING.
That distinction will be maintained throughout this book.
9. FURTHER STEPS
Where can you go from here?
TRY
Take five items from your current to-do list.
For each, write:
TASK → OUTPUT → OUTCOME
How many of your to-do items describe activity rather than accomplishment?
EXPLORE
Watch for Outcome Drift in your next AI conversation.
You begin with one purpose.
AI starts generating.
The conversation becomes increasingly elaborate.
Eventually you're improving the output rather than pursuing the original outcome.
When that happens, ask:
“Return to my intended outcome. Is what we're doing still helping me achieve it?”
READ
Explore Locke and Latham's goal-setting work, Biggs's constructive alignment, UNESCO's human-centred AI competency framework, and the history of End-User Computing.
They come from different fields, but each helps illuminate an important principle:
Tools, activities and processes should ultimately serve human purposes.
NEXT
Unit 03 — Define What Success Looks Like
Finding the outcome creates the next question:
How will you know when you've actually achieved it?
Unit 03 will turn a desired outcome into observable success criteria.
No comments:
Post a Comment