Understand the technology
From creative potential to a justified decision.
For May and November 2027. This lesson is for SL and HL students. Bring together technical and ethical ideas through a worked pilot, decision matrix and balanced studio recommendation.
01 | THE QUESTION IS HOW TO USE THE OPPORTUNITY
Visionary Studios is considering generative AI for advertising, concept art and digital media. Producing an attractive image is only part of the decision. The studio also needs a dependable way to deliver work that fits the brief and respects the people affected.
Balancing opportunities and responsibilities means evaluating useful outcomes alongside costs, limitations and obligations. It does not mean that enough speed automatically cancels a serious unresolved concern.
This lesson brings together the earlier technical and ethical ideas. The campaign and numerical examples are fictional classroom material, not findings reported in the case-study booklet.
02 | OPPORTUNITY: EXPLORE MORE IDEAS
At the beginning of a project, a designer may want to compare several possible moods, settings or compositions. Generative tools can help produce visual starting points without completing every concept manually.
This can make it easier to discuss alternatives with a client. A rough generated scene may reveal that a proposed colour palette does not suit the campaign before the team commits to detailed production.
The value is the quality of the decisions enabled, not simply the number of images generated. Hundreds of similar candidates can add review work without adding useful variety.
03 | OPPORTUNITY: ITERATE AND ADAPT
A client may ask for a different background, lighting style or composition. A generation workflow can support experiments with these changes and help designers compare directions.
However, changing one requested feature can also alter details the client wanted to preserve. A character’s clothing or a product feature may drift between versions.
For the studio, useful iteration therefore means producing relevant alternatives while retaining required identity and brand details. Test that ability rather than assume every edit will be controlled.
04 | OPPORTUNITY: SUPPORT PEOPLE WITH DIFFERENT SKILLS
A person who can describe an idea but cannot draw it confidently may use generated images to communicate an initial concept. This can widen participation in creative discussions.
That does not remove the value of visual judgement. Someone must still identify misleading details, choose appropriate compositions and develop the work into a suitable final asset.
Accessibility also depends on tool design, cost and training. A workflow that helps one colleague may be difficult for another, so evaluate who benefits and what support they need.
05 | RESPONSIBILITY: DELIVER THE ACTUAL BRIEF
A beautiful image can still fail the task. If the brief requires the same character in six settings, six individually impressive but inconsistent characters are not a successful result.
Turn requirements into checks: identity features, style, resolution, product accuracy and suitability for the intended audience. Decide which are essential before comparing speed.
This prevents a striking demonstration from standing in for evidence that the studio can deliver repeatable client work.
06 | RESPONSIBILITY: RESPECT SOURCES AND PEOPLE
Dataset provenance, relevant permissions, fair portrayal and clear disclosure remain part of production even when a tool is convenient. They address different questions and need appropriate evidence.
For example, a varied dataset does not establish permission to use its images. A disclosure does not correct a stereotyped portrayal. An attractive output does not settle either issue.
Reuse the checks from the previous lessons, connecting each one to the studio’s intended activity and affected stakeholders.
07 | WHO GAINS AND WHO CARRIES THE COST?
| Stakeholder | Potential opportunity | Concern or responsibility |
|---|---|---|
| Designers | More concept exploration | Training, control and changing work roles |
| Clients | More options and potentially faster delivery | Accuracy, approval and clear information |
| Creators and contributors | New production and collaboration opportunities | Recognition, permissions and treatment of their work |
| People represented | Broader relevant portrayal | Stereotypes, exclusion and misleading depictions |
| Studio managers | New services and workflows | Total cost, reliability and accountability |
These are possible effects to examine, not guaranteed outcomes. A benefit for one stakeholder can create extra work or risk for another.
08 | MEASURE THE WHOLE WORKFLOW
Generation time is only one part of production. Include briefing, setup, rejected candidates, review, editing, approval and any corrections.
A tool producing an image in seconds may still be slow to use if most candidates are unsuitable. Conversely, a longer generation step may be worthwhile if the outputs need less editing.
Compare accepted assets meeting the same requirements. Counting every generated file as a finished deliverable exaggerates productivity.
09 | WORKED EXAMPLE: TIME PER ACCEPTED ASSET
| Workflow | Total production time | Accepted assets | Time per accepted asset |
|---|---|---|---|
| Existing workflow | 240 minutes | 8 | 30 minutes |
| AI-assisted workflow | 180 minutes | 8 | 22.5 minutes |
The AI-assisted time is 180 ÷ 8 = 22.5 minutes per accepted asset. The saving is 30 − 22.5 = 7.5 minutes per asset, or 7.5 ÷ 30 × 100 = 25% .
This is a saving within the measured pilot. Investigate whether the figures include setup, training and all review work before generalising. Acceptance should include the same essential checks in both workflows.
10 | RESOURCE USE AND ENVIRONMENTAL RESPONSIBILITY
Training and generation require computing resources. The relevant demand depends on the model, hardware, number of runs and workflow, so image count alone is not a complete environmental measure.
The studio can reduce avoidable computation by planning experiments, limiting unproductive retries and reusing suitable approved assets. It should seek relevant evidence rather than assume a local or hosted option is always better.
Compare resources used to achieve useful outputs where data are available. State gaps in energy or infrastructure information instead of inventing precise environmental claims.
11 | AUTOMATION CHANGES WORK RATHER THAN REMOVING EVERY TASK
Some production tasks may become quicker while selection, checking and editing become more important. Staff may need to learn new tools and new ways to document decisions.
Involve the people doing the work when designing a pilot. They can identify hidden effort, such as correcting small product details or keeping versions consistent.
A responsible plan considers training and role changes alongside productivity. Do not assume either that every job disappears or that nobody’s work will change.
12 | GATES AND TRADE-OFFS ARE DIFFERENT
A trade-off is a balance between competing desirable outcomes, such as faster delivery and more editing control. A decision gate is a requirement that must be satisfied before proceeding.
For example, the studio might require client approval and resolution of a particular source concern before publication. A high speed score should not automatically override a failed gate.
Agree essential conditions first. Then compare the benefits and costs of the options that remain suitable.
13 | USE A DECISION MATRIX CAREFULLY
| Criterion | Weight | Workflow A | Workflow B |
|---|---|---|---|
| Quality | 3 | 4 × 3 = 12 | 5 × 3 = 15 |
| Speed | 2 | 5 × 2 = 10 | 3 × 2 = 6 |
| Editing control | 1 | 3 × 1 = 3 | 4 × 1 = 4 |
| Total | 25 | 25 |
A weighted score multiplies a criterion score by its importance weight. Here the totals are equal, although the strengths differ. The matrix structures discussion; it does not make the decision objective by itself.
Explain the evidence behind scores and the choice of weights. If editing control becomes more important, the result may change. Check how dependent the recommendation is on those assumptions.
14 | MATCH THE TOOL TO THE TASK
One workflow need not serve every project. AI generation may be useful for early concepts while commissioned artwork or photography remains appropriate for assets needing precise control or documentary accuracy.
A combined workflow can use generation for exploration and human production for the final result. It may retain creative benefits while meeting requirements that the generated candidates cannot reliably satisfy.
Evaluate the added handover and editing effort. Combining methods is a proposal to test, not an automatic solution to every limitation.
15 | RUN A LIMITED PILOT BEFORE EXPANDING
Choose representative briefs, an appropriate comparison workflow and clear acceptance conditions. Set a review period and identify who can stop the trial if an important concern emerges.
Record both successes and failures. A selected showcase of the best image from each attempt can hide the effort required to obtain them.
Expansion should follow evidence of useful, repeatable results and workable responsibilities. A successful trial on one campaign does not establish suitability for every future client.
16 | MAKE RESPONSIBILITY SPECIFIC
“The team will be careful” does not identify who checks what. Assign responsibility for source records, visual review, client communication and final approval.
Provide a route for unresolved concerns and preserve the decision record. If the workflow changes, revisit the checks rather than assume an old approval still covers the new use.
Accountability is practical: named roles, clear decisions and the ability to investigate and correct a problem.
17 | WRITE A BALANCED EVALUATION
Start with a relevant benefit, explain why it matters to the studio, then examine a limitation and the evidence needed. Finish with a judgement linked to the brief.
Balance does not mean giving every issue equal importance. Explain which requirements matter most and why, then state the conditions and limits of your recommendation.
18 | RECOMMEND A RESPONSIBLE OPPORTUNITY
Visionary Studios wants more concept options without losing client confidence. Propose a limited AI-assisted workflow and justify when it should or should not be expanded.
Reveal a possible response
Use a representative concept-art pilot with agreed quality and consistency criteria. Record total production time, rejected candidates and editing. Assign source, fairness and disclosure checks and obtain client approval for the proposed use.
Expand only if the measured benefits remain after review costs and the essential conditions are satisfied. Retain human-led alternatives for unsuitable tasks and review the workflow after tool updates or new concerns. This supports creative exploration without treating convenience as sufficient evidence of responsible use.
Use these independent teaching notes alongside the official 2027 case-study booklet. The worked pilot and decision matrix are classroom examples.
CHECK YOUR UNDERSTANDING | CLASSIFY THE DECISION ELEMENT