Advanced · Ethics and society · module 149 of 155
Ethical assessments with a step-by-step plan
Weigh up an AI application with a fixed step-by-step plan: who is affected, what are the benefits and risks, and which choice can you justify?
After this lesson
You work through an ethical assessment with stakeholders, benefits and possible harm, alternatives and a recorded decision.
In view · Step by step
An ethical assessment in steps
What does the application do, for whom, and who is affected without having a say?
Benefits and possible harm per group; can it be done with less risk?
Record the reason, take measures and plan a moment to look back.
Part 1 of 3 · Application and those involved
Your preference for less movement is on. Choose each part yourself.
From gut feeling to assessment
With a new AI application, people often have a strong feeling: enthusiasm or unease. A step-by-step plan helps to turn that feeling into an assessment you can explain. First describe the application and the goal concretely: what does it do, for whom, and which decision does it support? Then map out the stakeholders, including those who are not at the table: customers, citizens, staff or children (lessons 83 and 93).
Weigh benefits and possible harm
For each group, note the possible benefits and the possible harm: mistakes, unequal treatment, loss of privacy or of human contact, dependence on a supplier. Also look at who is most vulnerable. Look for alternatives: can it be done without AI, with less data, or with a person who decides? Think of measures that limit the harm, such as human oversight, transparency or a procedure for raising objections (lesson 84).
Decide, record and revisit
Make a decision and write down the reason, with the risks you accept and the measures you take. Who decided, and when will you evaluate? A recorded assessment makes it possible to learn later, even if things turn out differently than expected. Agree what you will do if complaints come in or if the consequences turn out differently. Ethics is not a box to tick beforehand, but a question you keep asking.
New practice example
From explanation to a usable result.
A step-by-step plan turns a feeling about AI into an assessment that you can explain and evaluate later.
Fictional practice situationFictional: a care home is considering cameras with AI that detect falls in the rooms and alert the staff.
View a possible solution
This is how it can be done
Stakeholders: residents, family and staff. Benefit: faster help after a fall. Possible harm: loss of privacy in one’s own room, false alarms and fewer personal rounds. Alternatives: a fall sensor without a camera, or a wearable alarm button. Measures: only with the resident’s consent, no storage of images, and the rounds continue. Decision: first a trial with sensors without cameras; evaluation after three months.
Why this works
The alternative without a camera keeps the benefit with less harm to privacy. The assessment makes that choice visible and explainable.
Make it slightly more difficult
Which question would you ask the residents themselves before you decide?
Extra practice, at your own pace. You do not need to submit anything. This does not count towards an automatic assessment.
Your workplace · learning by doing
Make something. See what changes.
Checking the AI connection… You can already view the written example.
Present your assignment to Ende.
Read the answer, change one thing, and compare.
Previously held
New version
You remain in control
What do you want to improve in your next attempt?
On this device only. AI answers remain only in this page memory. This is self-review, not a grade.
Why this approach works
The alternative without a camera keeps the benefit with less harm to privacy. The assessment makes that choice visible and explainable.
A step further
Which question would you ask the residents themselves before you decide?
Try & discover · no obligation
Put the steps in a workable order
Move the steps with the arrows. Check whether each step has its necessary input and control.
What does a step depend on?
- Describe the application and the goal must precede Map all the people involved
- Map all the people involved must precede Weigh benefits and possible harm for each group
- Weigh benefits and possible harm for each group must precede Look for alternatives and measures
- Look for alternatives and measures must precede Decide and record the reason
- Decide and record the reason must precede Evaluate after an agreed period
Move a step and then check the sequence. The arrows also work with your keyboard.
This model checks dependencies, not duration, capacity, or actual execution. With multiple agents, the human decision remains a separate step.
This trial works locally with fictitious data and does not count towards your lesson completion. It starts anew with a new lesson or reload.
A prompt to get you started
Help me make an ethical assessment of this fictional application: a municipality wants to use AI to sort grant requests by completeness, so that staff can work faster. Go through the steps: application and goal, stakeholders, benefits and possible harm per group, alternatives and measures, and the decision with an evaluation date. Do not make a decision yourself; set out the considerations.
Apply it
A school wants to use AI to predict which pupils are at risk of dropping out. The idea comes from the school management; pupils and parents have not yet been involved.
Which approach best fits this situation?
- Introduce it straight away; early intervention always helps.
- Introduce it without telling the pupils, so as not to worry them.
- Work through an ethical assessment with pupils, parents and teachers, look at alternatives and safeguards, and only then decide.
With JavaScript, you can check your approach and receive immediate feedback.
Test your insight
Which step is most often missing from an ethical assessment?
- Describe the purpose of the application.
- Map out the stakeholders who are not at the table, and plan a moment to evaluate the decision.
- Check which supplier is cheapest.
Extra practice
Choose an AI application in your own environment. Work through the step-by-step plan on one page and present it to someone who sees the application differently from you.
Application: sorting requests by completeness. Stakeholders: applicants, staff and people who struggle with forms. Benefit: faster processing. Possible harm: a complete request is seen as incomplete and is left pending. Measures: AI only sorts; a staff member looks at every request flagged as incomplete; applicants get a clear point of contact. Evaluation after three months, with a sample of flagged requests.
