Generative AI Leader Part 1: Exam Overview and Gen-AI Fundamentals

Generative AI Leader is the world’s first gen-AI certification aimed at non-engineers — people who plan and oversee AI projects rather than write code. This first part covers the exam’s shape and the front two of its four domains: the fundamentals of generative AI (~30%) and Google Cloud’s generative AI offerings (~35%). Those two alone are two-thirds of the score.

The shape of the Generative AI Leader exam — four domains and weights

The four domains carry roughly: fundamentals of generative AI (~30%), Google Cloud’s generative AI offerings (~35%, the largest), techniques to improve model output (~20%), and business strategy for a successful gen-AI solution (~15%). Part 1 covers the first two; Part 2 covers the rest.

Domain 1 — fundamentals of generative AI (~30%)

This domain asks you to place generative AI within the broader AI landscape and to speak its vocabulary precisely. Hold the stack as five layers — from infrastructure and data, through foundation models, to the tools and agents built on top, and finally the business applications people actually use. Google’s own foundation models, the Gemini family, sit at the model layer. The exam rewards clean definitions: what a foundation model is, what makes a model generative, and how prompting, grounding, and fine-tuning differ.

Domain 2 — Google Cloud’s generative AI offerings (~35%, the largest)

This is the heaviest-weighted domain and the part AWS veterans must study fresh, because it is about Google’s specific product names and where each one fits. Learn the Gemini product family by audience and purpose.

On the developer side, the foundation and agent layer is what you would reach for to build custom solutions: Model Garden for choosing models, the platform for grounding and retrieval, and tooling to assemble agents. If you know AWS, the fastest way to absorb this domain is by analogy.

Where the front two domains trip people up

The biggest trap is treating the Gemini products as interchangeable. The exam distinguishes by audience: the Gemini app for individuals, Gemini Enterprise for company-wide platforms, and Gemini for Workspace for in-document help. The second trap is vocabulary: grounding, retrieval-augmented generation, and fine-tuning are tested as distinct techniques, so do not collapse them into one idea.

Conclusion — capture two-thirds of the score up front

Domains 1 and 2 are worth roughly two-thirds of the exam, and both reward clear definitions plus product-by-audience recall rather than coding. Anchor the five-layer landscape and the Gemini family map, then move on to the output-improvement and strategy domains in Part 2.

References

Google Cloud official Generative AI Leader certification page and exam guide (domain weightings), and Google Cloud Skills Boost learning path.

ブラウザだけでできる本格的なAI画像生成【ConoHa AI Canvas】

swiftwand

X

AIを使って、毎日の生活をもっと快適にするアイデアや将来像を発信しています。 初心者にもわかりやすく、すぐに取り入れられる実践的な情報をお届けします。 Sharing ideas and visions for a better daily life with AI. Practical tips that anyone can start using right away.