For the 2027 IB Computer Science Paper 1 case study, useful notes must do more than summarize the booklet. They should help you retrieve precise definitions, connect technology to the Visionary Studios scenario, recall credible real-world evidence, and construct a balanced judgement under time pressure.
The most effective system has three layers: a terminology bank, a real-world evidence bank, and a separate page for each challenge question. This how-to guide explains how SL and HL students can build those layers without producing an unmanageable document.
Understand what the case study assesses
The 2027 case study is titled “Generative AI for image creation -- a diffuse vision.” It is required for the May and November 2027 examination sessions and focuses on Visionary Studios, a design company evaluating generative AI for advertising, concept art, and digital media.
Under the syllabus first assessed in 2027, the case study appears in Paper 1 for both SL and HL. The official IB Computer Science subject brief confirms that Paper 1 examines Theme A and includes questions related to the case study.
| Feature | SL | HL |
|---|---|---|
| Paper 1 duration | 1 hour 15 minutes | 2 hours |
| Paper 1 weighting | 35% | 40% |
| Case study marks | 12 marks | 24 marks |
| Challenge questions studied | 2 | 4 |
| Recommended case study teaching time | 15 hours | 30 hours |
The case study is normally released through the IB Programme Resource Centre 12 months before the May session and 18 months before the November session. This extended preparation period exists because students are expected to research beyond the booklet, not merely memorize its wording.
Your notes should therefore prepare you to:
- explain how the relevant systems and models work;
- apply Theme A concepts to Visionary Studios;
- compare possible technical approaches;
- discuss impacts and ethical issues;
- use independent research as evidence;
- evaluate strategies and reach justified conclusions.
These correspond broadly to AO1 knowledge, AO2 application, and AO3 analysis and evaluation. A page of copied definitions addresses only part of what the examination can assess.
Start with a structured reading of the booklet
Read the 2027 case study booklet three times, with a different purpose each time.
First reading: understand the scenario
Identify the organization, its objectives, its stakeholders, and the decision it must make. For Visionary Studios, the central question is not simply whether generative AI is useful. The company must select and operate suitable models while balancing output quality, computational efficiency, consistency, legal exposure, and ethical responsibility.
Write a scenario summary of no more than five sentences. If it becomes longer, you are probably reproducing the booklet rather than identifying its decision-making context.
Second reading: label examinable material
Use a simple annotation code on your working copy:
- D for a term requiring a definition;
- P for a process you must explain;
- C for a stated challenge;
- E for ethical or legal considerations;
- S for a syllabus connection;
- ? for an issue requiring independent research.
Do not use six different highlighter colours without labels. Colour alone does not tell you what to do with the information later.
Third reading: convert information into questions
Turn headings and claims into retrieval prompts. “Diffusion models use iterative denoising” should become questions such as:
- What is iterative denoising?
- What happens during forward and reverse diffusion?
- Why is a neural denoiser required?
- How can repeated denoising affect computational efficiency?
- Why does this matter to Visionary Studios?
Questions expose gaps that passive highlighting conceals.
Build a three-layer note system
Layer 1: create an exam-ready terminology bank
Definitions should be brief enough to recall but detailed enough to distinguish related concepts. Use a table rather than disconnected paragraphs.
| Term | Precise definition | Process or distinction | Scenario link |
|---|---|---|---|
| Diffusion model | A generative model that learns to reverse a gradual noise-adding process | It begins generation from noise and repeatedly predicts and removes noise | Can produce high-quality images, but repeated steps may require substantial computation |
| Neural denoiser | A neural network trained to estimate or remove noise at a particular diffusion step | Often uses convolutional structures to reconstruct meaningful visual features | Its accuracy affects the quality and coherence of Visionary Studios’ outputs |
| Conditional generation | Generation guided by an input such as text, a label, or another image | The condition constrains the output | Helps designers produce images that match a client brief |
| Bias | Systematic distortion or unfair representation arising from data, design, or deployment | It can be learned from unrepresentative training data | Outputs could reinforce stereotypes or misrepresent groups in advertising |
Avoid circular definitions such as “a diffusion model is a model that uses diffusion.” Define the concept, explain its mechanism, and distinguish it from nearby terms.
Turn the definition column into active-recall cards. RevisionDojo’s Computer Science flashcards can support spaced retrieval, while the new-syllabus Computer Science notes can help you repair gaps in Theme A knowledge.
Layer 2: build a real-world evidence bank
An example is useful only if you can explain what it demonstrates. Record each source using a consistent evidence card:
| Field | What to record |
|---|---|
| Example | Model, company, research paper, legal dispute, or documented incident |
| Claim supported | The exact argument the example helps establish |
| Mechanism | Why the result occurred technically or socially |
| Limitation | Why the example may not transfer perfectly to Visionary Studios |
| Recall tag | A short cue such as “bias -- training data” |
| Source details | Organization or author, title, year, and URL |
For example, evidence that diffusion models can generate detailed images does not automatically prove that they are the best option for Visionary Studios. You must also consider inference time, hardware, controllability, consistency, intellectual property, and the company’s project requirements.
Prioritize sources in this order:
- peer-reviewed research and original technical papers;
- government, university, or standards-body publications;
- official documentation from model developers;
- reputable reporting for legal disputes and recent incidents;
- explanatory sources used only to clarify difficult ideas.
Keep the full link in your research notes, but create a compact memory cue for revision. The aim is to understand and recall the evidence, not to memorize a bibliography word for word.
Layer 3: make one dossier for each challenge
The shared SL and HL challenges concern:
- managing iterative denoising and the computational demands of photorealistic image generation;
- navigating intellectual property and bias when using large training datasets.
HL students must additionally prepare:
- balancing the generator and discriminator in a GAN;
- evaluating how hybrid models could combine the strengths and weaknesses of VAEs, GANs, flow-based models, and diffusion models.
Use the same template for every challenge:
| Section | Notes to include |
|---|---|
| Core problem | One precise sentence defining the challenge |
| Technical causes | Three to five causal points |
| Consequences | Effects on quality, cost, reliability, stakeholders, or legality |
| Strategies | Possible technical and organizational responses |
| Trade-offs | Costs, risks, and limitations of each strategy |
| Evidence | Two or three well-understood real-world examples |
| Scenario application | Why the issue matters specifically to Visionary Studios |
| Judgement | Conditions under which one response is preferable |
This structure mirrors the thinking required for an evaluative answer: identify the problem, explain causes, compare responses, apply evidence, and make a qualified judgement.
Separate SL depth from HL depth
SL students should not treat the HL-only technologies as compulsory content. Your priority is mastering diffusion-based image generation, computational demands, intellectual property, bias, and their application to the two shared challenges.
HL students require broader comparative notes. For GANs, do not stop at defining the generator and discriminator. Explain adversarial training, unstable convergence, mode collapse, and how an imbalance between the networks can damage output quality. For hybrid models, organize notes by what each architecture contributes and what new complexity the combination creates.
A useful HL comparison table might use these headings:
| Model family | Main strength | Main limitation | Possible role in a hybrid | Relevance to Visionary Studios |
|---|---|---|---|---|
| Diffusion | High-quality, diverse output | Iterative generation can be computationally demanding | High-fidelity generation or refinement | Suitable where quality outweighs generation time |
| GAN | Potentially fast generation after training | Training instability and mode collapse | Sharp output or adversarial refinement | Useful when rapid production matters |
| VAE | Structured latent representation | Outputs may be less sharp | Controllable latent encoding | Supports variation and editing |
| Flow-based model | Exact likelihood and reversible mapping | High memory and architectural costs | Tractable transformation of representations | Relevant if interpretability of latent mapping is valuable |
Treat these as starting comparisons, not universal verdicts. Performance depends on implementation, data, hardware, training objectives, and the company’s success criteria.
Compress your notes into an exam-retrieval system
A large research document is useful during preparation but poor for final revision. Reduce it in stages:
- Master document: complete definitions, research, links, diagrams, and examples.
- Challenge dossiers: approximately two pages per challenge.
- One-page retrieval sheets: causes, impacts, strategies, evidence, and judgement cues.
- Flashcards: definitions, process steps, comparisons, and example prompts.
- Closed-book plans: outlines written from memory in five minutes.
Use consistent labels such as DEF, PROCESS, EVIDENCE, LIMITATION, and JUDGEMENT. Under pressure, visual consistency makes information easier to reconstruct.
Do not assume that extensive notes can substitute for recall. Students should expect to work from the examination materials provided under their school’s IB examination arrangements, not from personal research notes. Confirm practical booklet arrangements with your teacher or coordinator, then prepare on the assumption that independently written notes must be learned before the examination.
Turn notes into exam answers
Every week, select one challenge and answer a progression of prompts:
- Define: give the exact meaning of a term.
- Describe: state relevant characteristics or stages.
- Explain: show how or why something occurs through a causal chain.
- Compare: address similarities and differences using common criteria.
- Evaluate: weigh strengths and limitations, apply evidence, and reach a justified conclusion.
For an evaluation, plan around criteria rather than writing separate “advantages” and “disadvantages” lists. A recommendation for Visionary Studios might compare models using image quality, generation time, infrastructure cost, controllability, consistency, and legal risk.
Use timed practice to reveal which notes are not retrievable. The RevisionDojo Computer Science Questionbank and Computer Science predicted papers can help you practise applying concepts rather than rereading them. Jojo AI can also review a response for missing application, unsupported claims, or an unbalanced conclusion, although feedback should be checked against your teacher’s guidance and official materials.
Common case study note-taking mistakes
- Copying the booklet: this records information without explaining it.
- Collecting examples without claims: an example is not evidence until you state what it proves.
- Writing generic ethics notes: connect bias or copyright directly to training data, outputs, stakeholders, and Visionary Studios.
- Ignoring trade-offs: every plausible technical strategy has costs or limitations.
- Mixing SL and HL priorities: SL students can waste time on optional complexity, while HL students may omit required comparative depth.
- Leaving practice until the end: retrieval and argument construction improve through repeated use, not final-week rereading.
- Trusting AI-generated summaries without verification: technical errors become harder to detect once memorized.
Conclusion
Effective case study note-taking is a process of selection, organization, and retrieval. Build precise definitions, attach every real-world example to a clear claim, and structure each challenge around causes, consequences, responses, trade-offs, evidence, and judgement.
Keep a detailed research document, but progressively compress it into challenge sheets, flashcards, and closed-book answer plans. RevisionDojo’s Computer Science notes, Flashcards, Questionbank, Predicted Papers, and Jojo AI are most useful when they support this active process rather than replace your own verified research.
Sources and referenced URLs
- IB Computer Science subject brief, first assessment 2027
- IB Computer Science curriculum update
- 2027 case study booklet: Generative AI for image creation
- Computer Science case study guidance and lesson structure
- Computer Science 2027 syllabus changes overview
- RevisionDojo IB Computer Science first-exam-2027 resources
- RevisionDojo Computer Science revision notes
- RevisionDojo Computer Science flashcards
- RevisionDojo Computer Science predicted papers
- RevisionDojo Jojo AI
