Practice Exam: Generative AI Leader

Question bank: 100 · timed mock exam: 90 min

Official exam page on the vendor's site (opens in a new tab) Official prep course (opens in a new tab)

Exam guide: format, domains and how to prepare

Compare: AIF-C01 vs Google Generative AI Leader

  • Google Cloud's gen AI offerings 36 questions
  • Fundamentals of gen AI 30 questions
  • Techniques to improve gen AI model output 20 questions
  • Business strategies for a successful gen AI solution 14 questions
Study mode (free) Exam mode (subscribers)

Free: 25 questions in study mode, all domains, no account.

Subscription: the 100 questions of this certification, the timed exam mode, and all other certifications too.

Subscribe · €9.99 / month

Secure card payment (Stripe) · No commitment · Cancel online

How to prepare for Generative AI Leader with InfraPrep

The 100 questions cover 4 domains: Google Cloud's gen AI offerings, Fundamentals of gen AI, Techniques to improve gen AI model output, Business strategies for a successful gen AI solution. 25 of them are free, without an account.

Study mode

Answer at your pace: each answer is corrected straight away, with an explanation and the link to the documentation it relies on.

Exam mode

A timed exam of 90 minutes, with a question map to flag and come back to questions before you submit.

Results by domain

Your score is broken down by domain, so you know what to review first.

Original questions

Every question is written from the public documentation, never copied from the exam. You can contest an answer from study mode.

10 free sample questions

Question 1 Fundamentals of gen AI

A retail chain has years of raw purchase records with no categories or tags attached. The marketing director wants to discover groups of shoppers who tend to buy related products, without deciding the groups in advance. Which machine learning approach fits this goal?

  1. Unsupervised learning, because the model finds patterns in data that carries no labels
  2. Reinforcement learning, because the model is rewarded each time a shopper buys a product
  3. Supervised learning, because the model predicts a known label for each shopper
  4. Supervised learning, because the model can train on a small sample of the records
Show the answer and the explanation

Correct answer: A

Unsupervised learning works on unlabeled data and is suited to discovering new patterns and groupings, such as buyer groups that purchase related products together (customer segmentation).

Source: official documentation

Question 2 Google Cloud's gen AI offerings

An executive asks what makes Google Cloud's AI-optimized infrastructure different. Which statement correctly describes Tensor Processing Units (TPUs)?

  1. Google's custom-built chips designed to accelerate machine learning
  2. Third-party graphics cards that Google resells under its own brand
  3. Standard CPUs that Google rents by the hour for any kind of workload
  4. A software library that speeds up models on existing hardware
Show the answer and the explanation

Correct answer: A

TPUs are Google's custom-developed application-specific integrated circuits (ASICs) used to accelerate machine learning workloads, and they are part of Google Cloud's AI infrastructure.

Source: official documentation

Question 3 Techniques to improve gen AI model output

A pilot chatbot sometimes cites web pages that do not exist and states wrong facts with confidence. What is the name of this limitation of gen AI models?

  1. Model drift, where the input data changes slowly after deployment
  2. Hallucination, where the model produces incorrect or misleading results
  3. Prompt injection, where a user slips hidden instructions into the prompt
  4. Overblocking, where safety filters reject harmless prompts by mistake
Show the answer and the explanation

Correct answer: B

AI hallucinations are incorrect or misleading results generated by AI models, and they can even extend to fabricated links to web pages that never existed.

Source: official documentation

Question 4 Business strategies for a successful gen AI solution

A team must decide whether a new candidate model should replace the current baseline model for drafting customer emails. Reviewers find it hard to write a precise scoring rubric, but they can easily say which of two drafts they prefer. Which evaluation method fits best?

  1. Counting the number of models deployed in the organization
  2. Precision and recall computed against a reference dataset of emails
  3. Pointwise metrics, scoring each draft alone against a fixed rubric
  4. Pairwise metrics, comparing both models' drafts to get a win rate
Show the answer and the explanation

Correct answer: D

Pairwise metrics compare the responses of two models and pick the better one to create a win rate; they are often used to compare a candidate model with a baseline and work well when a scoring rubric is hard to define and preference is enough.

Source: official documentation

Question 5 Fundamentals of gen AI

During an executive briefing, a manager describes a machine learning approach in which a model improves by trying actions and receiving positive or negative feedback on the results. Which approach is being described?

  1. Supervised learning
  2. Unsupervised learning
  3. Reinforcement learning
  4. Semi-supervised learning
Show the answer and the explanation

Correct answer: C

Reinforcement learning is described as "learn by doing": the model goes through trial and error and receives positive reinforcement when it performs well and negative reinforcement when it performs poorly.

Source: official documentation

Question 6 Google Cloud's gen AI offerings

A CTO wants one Google Cloud offering that brings together performance-optimized hardware, open software and flexible consumption models for large-scale AI training and serving, instead of assembling each piece separately. What should the CTO look at?

  1. Gemini Enterprise, the AI assistant for employees across the company
  2. AI Hypercomputer, a unified portfolio for AI infrastructure
  3. Model Garden, the catalog of foundation models on Agent Platform
  4. Agent Search, the service for search and recommendation apps
Show the answer and the explanation

Correct answer: B

AI Hypercomputer is an integrated supercomputing system for AI and ML workloads; it is the unified portfolio for Google Cloud's AI infrastructure and a single entry point to hardware, open software and flexible consumption models.

Source: official documentation

Question 7 Techniques to improve gen AI model output

A product owner learns that the Gemini model behind a customer app has a published retirement date. The team has no time to follow the full migration guide before that date. What quick action does the documentation recommend?

  1. Stop the application as soon as a replacement model is announced
  2. Retrain a custom model from scratch before the retirement date arrives
  3. Point the app to the recommended upgrade, test features, then deploy
  4. Keep the current model, since retired models remain active indefinitely
Show the answer and the explanation

Correct answer: C

When there is no time for the full guide, the documentation says to update the application to point to the recommended upgrades, test all mission-critical features, and deploy the updates as usual.

Source: official documentation

Question 8 Business strategies for a successful gen AI solution

A CISO hears about Google's Secure AI Framework (SAIF) and asks what it is for. Which description is correct?

  1. A pricing model that bills AI workloads by the number of prompts
  2. A family of Google foundation models trained for cybersecurity tasks
  3. Guidance that helps organizations secure AI systems by default
  4. A certification that AI engineers must pass before deploying models
Show the answer and the explanation

Correct answer: C

SAIF offers guidance to organizations looking to secure AI systems; it addresses concerns such as AI/ML model risk management, security and privacy, helping ensure that AI models are secure-by-default. Google also provides a SAIF risk self assessment to support its implementation.

Source: official documentation

Question 9 Fundamentals of gen AI

A board member uses the terms "foundation model" and "large language model" as if they were synonyms. How should a gen AI leader describe the relationship between them?

  1. Every foundation model is an LLM, since each one is trained on text
  2. LLMs are foundation models, but some foundation models are not LLMs
  3. Foundation models are small models that are later combined into an LLM
  4. The two terms are separate categories with no overlap between them
Show the answer and the explanation

Correct answer: B

LLMs are a major type of foundation model, but foundation models also include models trained on images, audio, video or several modalities. The relationship is parent-child.

Source: official documentation

Question 10 Google Cloud's gen AI offerings

A company has rolled out Gemini Enterprise. Business users in finance, who do not write code, want to create their own agents to automate repetitive steps of their monthly close. What should the gen AI leader recommend?

  1. Export the finance data to a spreadsheet and ask the assistant to edit it by hand
  2. Build the agents in workflow builder with its no-code, drag-and-drop tools
  3. Wait for a developer team to code each agent, since agents require programming
  4. Use the Google-made prebuilt agents, since custom agents are not supported
Show the answer and the explanation

Correct answer: B

Gemini Enterprise lets everyone build no-code custom agents for their own workflows in workflow builder, using simple drag-and-drop tools.

Source: official documentation

Frequently asked questions

Are there free questions, and do I need an account?

Yes. 25 questions of this certification are free in study mode, across all its domains, without a subscription and without registration. The account is created when you subscribe, at checkout.

How does the subscription work, and can I cancel?

One monthly subscription covers all certifications and the timed practice exams. There is no minimum period. You cancel online, at any time, with the "Cancel your contract" function in My account; cancellation takes effect at the end of the month already paid, and no further payment is taken.

Where do the questions come from?

Every question is original, written from the vendor's public documentation. Each answer comes with an explanation and a link to the documentation page it relies on. InfraPrep is not affiliated with the vendors, and certification names belong to their owners.

In which language are the questions?

The questions are written in English, also on the French version of the site.

Is the catalog updated?

Yes, the catalog changes over time: new certifications and new questions are added.