One of the most practical sections of the Generative AI Leader exam is the part on identifying Gen AI use cases. After spending time on the AI landscape, model types, and how generative AI relates to machine learning and deep learning, the exam shifts toward something more applied: how do you actually figure out where Gen AI makes sense to use in a real business context?
I want to walk through how I think about this section, because the framing Google uses on the exam is specific and worth internalizing. It is not just a list of use cases to memorize. There is a mental model behind it, and once you have the model, the questions on this topic become much easier to answer.
Before evaluating any specific use case, it helps to know the range of things Gen AI is commonly applied to. The Generative AI Leader curriculum groups these into a handful of categories.
Customer support is one of the most widespread applications. This includes automated chat responses, ticket categorization, and FAQ generation. These are high-volume, repetitive tasks where a model can handle the bulk of the work and escalate edge cases to humans.
Written content generation covers blog posts and articles, marketing copy, and email campaigns. The model does not replace a writer, but it can accelerate drafting and iteration significantly.
Data analysis includes creating charts and summaries, finding trends in sales data, and forecasting demand and revenue. Gen AI is particularly useful here for the natural language layer, turning raw data outputs into readable narratives.
Creative design covers concept brainstorming, marketing asset creation, and brand messaging. This is where Gen AI acts more as a collaborator than an automator, helping teams explore options faster.
Code generation covers boilerplate templates, bug fixes, and API integration. This is one of the highest-ROI applications for developers, since a large chunk of day-to-day coding involves patterns the model has seen many times before.
These categories are not exhaustive, but they cover the bulk of what shows up on the exam when use cases are referenced.
Knowing the categories is one thing. The harder skill is actually identifying which use cases are worth pursuing in a specific business context.
The exam defines this skill explicitly: identifying different use cases is the skill of analyzing business situations and determining the optimal level and type of AI assistance. It requires balancing competing trade-offs across different scenarios: speed vs. control, automation vs. oversight, and innovation vs. risk.
There is no universal answer. The right level of AI involvement depends on the constraints of the situation, and that is the framing the Generative AI Leader exam wants you to bring to any use-case question.
The most useful comparison the curriculum offers is the contrast between a No Constraint framing and a Clear Constraint framing. This is the part most likely to show up directly on the exam.
A No Constraint framing is vague, broad, and tool-first. The example prompts look like this:
These are starting points, not use cases. They do not define what problem is being solved or how success would be measured. If a question on the exam describes a team approaching Gen AI with this kind of framing, the answer is almost always that the use case has not actually been identified yet.
A Clear Constraint framing is specific, measurable, and problem-first. The example prompts look like this:
These are the kinds of questions that lead to use cases you can actually evaluate and implement. The takeaway for the exam is that well-defined use cases are specific and problem-first. Vague, tool-first framing is a signal that the use case has not been properly identified yet.
The Generative AI Leader curriculum includes a concrete scenario that illustrates this framing in practice, and a similar setup is the kind of thing that can show up on the exam.
Imagine a team building a mobile app with a code-heavy workload. They are up against a deadline to demo to investors. Most of the sprint is complete, but there is a gap to close. The question they are asking is: how do you use AI to speed up prototype development without sacrificing overall quality?
That is a well-formed question. It names a constraint (the deadline), it names a goal (speed up prototype development), and it names a boundary condition (do not sacrifice quality). Compare that to a vague version like "we want to use AI on this project," and the difference is immediately clear.
What the team should do is use AI on targeted code snippets and functions. The example given is a fetch function that follows a predictable pattern:
func fetchUserProfile(userId: String) async throws -> UserProfile {
let url = URL(string: "https://api.example.com/users/\(userId)")!
let (data, _) = try await URLSession.shared.data(from: url)
return try JSONDecoder().decode(UserProfile.self, from: data)
}This is exactly the kind of task where Gen AI performs well. It is repetitive, well-defined, and low-risk to review. The team is not asking the model to design the architecture. They are asking it to write the boilerplate so the engineers can focus on the parts that actually require judgment.
The phrase to internalize from this section of the Generative AI Leader curriculum is this: use AI as a targeted accelerator, not a replacement for human judgment.
That framing captures the right mental model for evaluating almost any Gen AI use case. When an exam question presents a scenario, ask yourself two things. First, is the framing problem-first or tool-first? Second, is AI being positioned as a targeted accelerator on a specific bottleneck, or is it being positioned as a wholesale replacement for human work? Answers that match the targeted-accelerator framing on a clear, measurable constraint are almost always the correct ones.
If you can apply that lens consistently, the use-case questions on the Generative AI Leader exam stop looking like a memorization exercise and start looking like pattern recognition.
My Generative AI Leader course covers identifying Gen AI use cases, the clear-constraint framework, and the targeted-accelerator framing in depth, alongside the rest of the foundational material you need to pass the exam.
