The Shocking 21% Daily AI Usage | Notion Survey Exposes Japan's Divide
機械翻訳 / Machine-translated
A
AI Friends
2026年4月26日・29分で読める
機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"Our company talked about adopting AI, but in the end nobody actually uses it…" — does that sound familiar? On April 17, 2026, Notion Labs Japan published a new survey laying bare the real AI usage patterns of 1,000 Japanese knowledge workers in hard numbers.
The bottom line: a gap is widening between the top 21% who use AI every day and the bottom 36% who don't use it at all.
This article breaks down the survey findings in plain language, compares them against data from the Ministry of Internal Affairs and Communications, PwC, and Notion's global survey, and gives you the information you need to judge whether your company is ahead of the curve or falling behind.
Let's start with the basics of the survey.
It was announced on April 17, 2026, and conducted by Notion Labs Japan GK (Notion's Japanese subsidiary).
The survey was fielded November 13–14, 2025, by Neo Marketing Inc., and targeted 1,000 knowledge workers aged 20–59 who use AI in their work.
Note that because this survey is limited to people who already use AI at work, it does not capture figures from companies where AI hasn't been introduced at all.
The most striking figure in the survey is that 21% use AI in their work every day.
Expand that to once a week or more and the number reaches 64%.
In other words, two out of every three AI users have made it a weekly habit.
This suggests that AI is on its way to becoming as routine as brushing your teeth or commuting to work.
The flip side: 36% use it less than once a week — a persistent group who may have accounts but rarely open them for actual work tasks.
Many people will wonder: "Why publish November 2025 data in April 2026?"
It's standard practice for Notion to compile results and release them — via press events and downloadable reports — timed to the spring corporate sales season.
This is exactly when many companies start moving on "let's adopt AI this year" decisions as new fiscal-year budgets come into play.
Think of it like fashion labels unveiling their spring/summer collections in winter — it's a strategic communications move.
The survey asked respondents to select all use cases they rely on at least once a week. Let's look at the rankings and what they look like in practice.
This includes drafting emails, summarizing meeting notes, and rewriting articles.
It's the classic "make this email a little softer" request to ChatGPT or Claude.
Nearly one in two people uses AI this way, meaning it's becoming unremarkable to hear a colleague say, "I had AI write it and it sounded really natural."
This covers research tasks like "summarize the latest trends in this industry" or "compare the key features of these three competitors." What used to be a three-hour process — Google search, read ten articles, compile a summary — can now be done in five minutes by asking AI. The impact is enormous.
Roughly one in four programmers uses AI at least once a week.
This reflects the spread of tools like GitHub Copilot, Cursor, and Claude Code, which put an AI "right next to you" as you write code.
The fact that 23% holds across all job types is surprisingly high and suggests that even non-engineers are increasingly using AI to write lightweight scripts.
66% of users are "somewhat satisfied" and 12% are "very satisfied."
While 78% are satisfied in total, truly enthusiastic fans account for only about one in ten.
The biggest complaint is that output feels generic and lacks originality.
Think of it like a new assistant who reliably gives you average-quality answers but hasn't yet produced a truly inspired suggestion.
The result that generated the most discussion in the Notion survey was the near-even split of 49% saying their AI environment is well-prepared vs. 51% saying it is not. Break that down by company size and the picture becomes much sharper.
The key finding is that roughly 70% of small businesses (those with fewer employees) have not yet set up a proper AI environment.
The reasons are clear: insufficient budget and a lack of dedicated IT staff.
Large companies have IT departments that arrange company-wide ChatGPT contracts and put security policies in place, while small businesses often have employees using their personal free ChatGPT accounts informally, with no company-level infrastructure.
Using office furniture as an analogy: large companies issue everyone a high-end ergonomic chair, while small companies have each employee buying their own from IKEA.
The breakdown of specific barriers puts "lack of skills and training" first at 33%, followed by "budget constraints" and "insufficient governance," each at 28%.
Among managers in particular, budget and governance emerged as the top concerns.
In short, "no money to buy it" and "no rules for using it" are the two main bottlenecks at the organizational level.
When asked what they would delegate to AI if they could starting tomorrow, respondents chose "document creation and proofreading" at 19%, "information management and retrieval" at 18%, and "meeting minutes" at 11%. The common thread is tasks that are straightforward but time-consuming, while creative work and judgment calls remain areas people quietly want to keep for themselves.
The Notion survey alone only tells part of the story. Comparing it against other official data sharpens the picture of how deeply AI has actually penetrated in Japan.
In the White Paper on Information and Communications for Reiwa 7 (2025), released by the Ministry of Internal Affairs and Communications in July 2025, 55.2% of companies reported using generative AI for business, and 42.7% said they had established a clear policy for its use.
The individual usage experience rate was 26.7%, nearly double the 15.6% recorded a year earlier.
Unlike the Notion survey, which focuses on usage frequency among people who already use AI, the ministry's survey counts all companies and all individuals, so the two measure different things.
Combining both datasets, the overall picture that emerges is: "More than half of companies have introduced AI, but only about one in five of those is actually using it every day."
PwC's "Generative AI Survey Spring 2025: Five-Country Comparison" found Japan's generative AI adoption rate at 56%, roughly on par with the United States, United Kingdom, Germany, and Australia.
The old image of Japan as an AI laggard, at least in terms of adoption rates, is now a thing of the past.
However, PwC emphasizes that few companies have moved from adoption to actual transformation — meaning Japan has caught up at the entry point but is still lagging at the finish line.
In a separate Notion global survey of more than 3,000 respondents, 93% said AI is essential to productivity software and 80% said they would consider switching to a platform with AI features.
At the same time, only 20% said AI is deeply integrated into their company's tools.
Japan's "21% daily usage" closely mirrors the global figure of 20% for deep integration, suggesting that both Japan and the world share the same pattern: high expectations, slow implementation.
Using the data covered so far, let's look at a few real-world scenarios to help you assess where your own company stands.
Mr. A works at a 30-person IT company in Tokyo where GitHub Copilot is contracted company-wide and he uses it every day for code completion and questions.
In survey terms, he's squarely in the "21% daily users" vanguard.
He can develop at a speed that holds up against competing startups, so he has every reason to feel confident that his company is ahead of the curve.
Ms. B manages general affairs at a mid-sized regional manufacturer.
There's an IT person on staff but no one dedicated to AI, and employees are unofficially using personal free ChatGPT accounts.
This is a textbook example of the "70% of small businesses with inadequate environments" finding, running straight into the 28% budget and 28% governance walls.
Her next step is to explore options like Microsoft Copilot (free tier) or a business ChatGPT Team plan and to put company-wide guidelines in place.
Mr. C's company has already contracted Microsoft 365 Copilot company-wide.
In practice, though, only some of the younger staff open it every day, while many employees in their 40s and above avoid it out of apprehension.
This is the other side of the "42.7% with established policies" statistic — the gap between tool adoption and actual adoption on the ground.
Practical solutions include running internal AI workshops to bring managers on board and creating a library of use-case examples organized by job function.
For Ms. D, who works solo, AI is "a talented assistant available 24 hours a day."
She uses it broadly for outlining, research, drafting, and proofreading, switching among ChatGPT Plus, Claude, and Perplexity depending on the task.
In survey terms she's a regular in the "21% daily users" group — and freelancers and sole proprietors tend to be ahead of the curve precisely because they face no organizational barriers.
The survey also asked respondents what they consider essential for advancing AI adoption across their organizations. Here are the top answers.
Security came out on top.
Concerns about whether it's safe to feed confidential data into AI, and whether that data will be used for model training, are the single biggest sticking point.
Enterprise-tier plans, on-premises deployments, and log management are the solutions.
If the CEO or senior leadership doesn't explicitly declare "our company will embrace AI," the rest of the organization won't get in step. Top-down momentum is essential, respondents say.
The remaining top answers are "providing skills and training" at 33%, integrating tools with existing information infrastructure, and embedding AI into the specific tasks employees want to delegate. Success means thinking about training and workflow redesign together — not just deploying a tool and calling it done.
A. Globally, it's roughly average.
In Notion's global survey, just 20% of companies said AI is deeply integrated into their tools — almost identical to Japan's 21%.
That said, if nearly half of those who have "introduced" AI aren't actually using it, there is clearly untapped potential that hasn't been unlocked.
A. Start with at least these three things.
First, "sign up for a business-tier plan" (using free tiers for work carries risks around training data use).
Second, "establish usage guidelines" (e.g., do not enter confidential or personal information).
Third, "share success stories internally" (one person's win accelerates adoption for everyone else).
A. The survey shows text generation at 41% and research support at 38% as the most common uses, but for direct revenue impact, consider: ① accelerating customer response (handling initial inquiries), ② generating large volumes of proposals, and ③ personalizing sales emails. "Speeding up and personalizing customer touchpoints" is the direction most likely to produce results.
A. The biggest complaint is that output is generic and lacks originality.
General-purpose models return average answers, making it hard to get deep industry-specific insight or a truly distinctive angle.
Solutions include ① learning prompt engineering, ② using industry-specialized models, and ③ deploying AI fine-tuned on your company's own data.
A. Productivity differences between companies will compound like a snowball.
For example, a company that cuts document creation time in half can redirect those hours into sales and planning — creating a gap of several months' worth of work capacity per year.
Over a five-year horizon, market share could shift dramatically, and the terrifying thing about the AI gap is that "catching up later" becomes extremely difficult.
A. In the Notion survey, 71% of users said Notion AI is the most valuable AI tool they use.
Because it is integrated with Notion's documents and databases, it can provide answers that understand the context of your work — that's its strength.
A practical approach is to use it alongside ChatGPT or Claude: Notion AI for tasks within Notion, and a general-purpose AI for one-off questions or research.
This survey has put hard numbers to the polarization between those for whom AI has become a daily tool and those who haven't yet stepped through the door.
The delayed environment readiness of small businesses in particular is an issue with a high potential to directly affect market competitiveness five years from now.
Whatever your situation, take a moment to ask yourself: "Do I belong to the 21% who use AI every day, or the 36% who haven't made it a habit?"
Knowing where you stand is the first step toward changing how you work with AI.
This article is a cross-post from AI Friends.
