Does Passing IT Passport Make Japan's GenAI Passport Easier? I Compared All 191 Terms — the Overlap Was 22%

The short version

In other words, "what gets easier" and "what you start from scratch" are cleanly separated. Here is the detail.

How I counted

The method is simple.

1. Extract every "detailed keyword" from all five chapters of the official GUGA syllabus PDF (4th edition). 198 entries, or 191 unique terms.
2. Match them against the headwords, English labels and aliases of the 2,651-term IT Passport glossary, normalising parentheses, long vowel marks and full-width/half-width variants.
3. Review every machine match by hand.

Step three mattered more than I expected. Machine matching produced 46 hits, four of which were wrong. The syllabus lists "Operator", which is OpenAI's agent product, while the glossary's オペレータ means a systems operations staff member — different things entirely. Likewise "normalisation" means image preprocessing in the syllabus and database normalisation in the glossary. Removing those left the final figure of 42.

Chapter by chapter, the picture changes

The per-chapter breakdown is far more useful than the headline 22%.

Chapters 2 and 3 together come to 78 terms. Four in ten of the syllabus sits in those two chapters, and that is precisely where the overlap disappears. That was the clearest thing the count showed.

Where it overlaps — your existing study pays off

Chapter 4 (law and security) is the biggest

Eighteen terms, the largest single block. Japan's Act on the Protection of Personal Information and its associated vocabulary; intellectual property (copyright, patents, trademarks, designs), the Unfair Competition Prevention Act, trade secrets, limited-provision data; malware, ransomware, smishing.

Listed out, the pattern is obvious: this is not "AI material", it is law and security material. Generative AI raises its own questions, of course — the relationship between training data and copyright, above all — but the underlying legal vocabulary is essentially what IT Passport already covers. This transfers directly.

Chapter 1 (AI fundamentals) is also around a third

Twelve terms: machine learning, supervised and unsupervised learning, clustering, reinforcement learning, neural networks, weights, overfitting, regularisation, dropout, features, big data. The machine-learning basics in IT Passport's technology domain map straight onto Chapter 1.

Where it doesn't — this is the new material

Chapter 2: the history of generative AI is almost entirely new

Sixty-two terms, the largest chapter, and only seven overlapped. The seven that did (generative AI, CNN, RNN, attention mechanism, RLHF, fine-tuning, hallucination) are all terms you would meet in the news anyway.

Here is the other direction. These terms returned zero results when searched across all 2,651 glossary entries:

Not even ChatGPT — that one surprised me as I wrote it. IT Passport tests "what a working professional needs to know about using AI", so it never descends into model architecture or lineage. The numbers show that clearly. Boltzmann machines, latent vectors, positional encoding, MLM, RoBERTa, ALBERT — expect to learn these from zero.

Chapter 3: RAG, AI agents and MCP only arrived in October 2025

Two of sixteen overlapped (RAG and vector databases). Chunking, AI agents, MCP, GenSpark, Manus, Skywork AI and Veo 3 were added in the October 1, 2025 revision. Much of the existing exam-prep material has not caught up with this section yet.

And some AI terms exist only on the IT Passport side

So far this has been the view from the GenAI Passport. Reverse the direction and something else appears.

These AI-related terms are in the IT Passport glossary but absent from the GenAI Passport syllabus:

The difference in character is visible in the list. IT Passport's vocabulary grows toward "how do we govern AI and account for its decisions"; the GenAI Passport's grows toward "what is generative AI made of, and what can it do right now". Neither one fills in for the other.

Incidentally, under the new exam system starting in FY2027, IT Passport's subject areas are being reorganised into "Business", "Technology" and "Security & Ethics", with security and ethics for the AI era strengthened. The IT-Passport-only vocabulary above points in exactly that direction.

Which to take first

One way to read the numbers.

Taking IT Passport first is the more natural order. It lays down the foundation that Chapters 1 and 4 rest on — machine-learning basics, law, security — so that GenAI Passport study can concentrate on Chapters 2 and 3. In the other order you end up learning the legal material twice.

I published the study notes

The comparison data behind this article became a glossary index for the GenAI Passport, laid out chapter by chapter, covering all 191 terms of the official 4th-edition syllabus. It is in Japanese.

The 42 overlapping terms link straight through to the existing IT Passport glossary — there is no point maintaining two pages for the same word. Terms added in the October 2025 revision have their own explanatory pages.

Conclusion

The intuition that "they're both AI exams, so they must be similar" turned out to be half right. The foundations are shared; what sits at the centre of each exam is not. Framing it that way makes a study plan much easier to build.

This count reflects the 4th-edition syllabus as of August 2026. GUGA revises it roughly annually, so the numbers will move. I will re-run the analysis and update this article when the next revision appears.

This site is not affiliated with the Generative AI Utilization Promotion Association (GUGA) or the Information-technology Promotion Agency (IPA). It is unofficial study material. Please check the respective official sites for current information.