AI Enters the Season of ‘Fatigue’—Three Structural Reasons Why Small and Medium Enterprises Would Lose if They Stop Now

Let’s get straight to the point. ‘AI fatigue’ is real. But if you stop running just because you’re tired, you’ll find yourself lagging behind when you next open your eyes.

According to Gallup’s latest survey, only 18% of Generation Z (ages 14 to 29) responded that they have ‘hope’ for AI. Conversely, 22% reported feeling ‘frustrated.’ They use it but dislike it. This is the current atmosphere.

In another survey, 53% of voters stated that ‘the risks of AI outweigh the benefits.’ Both media and social networks have overhyped ‘AI, AI.’ It’s no wonder people are feeling overwhelmed.

But here’s a question to consider.

Is ‘public fatigue’ the same as ‘what’s important for business’?

When smartphones first came out, many insisted that feature phones were sufficient. When social media emerged, some business leaders claimed, ‘That won’t catch on.’ The public sentiment and structural changes in business are two different matters.

In fact, I believe that for small and medium enterprises, now is the perfect opportunity to quietly differentiate themselves while everyone else is stuck in ‘AI fatigue.’

Let me explain this with three structural reasons.

—

Reason 1: The Collapse of Costs is Unstoppable—’Too Expensive to Consider’ is No Longer True

Before discussing ‘AI fatigue,’ let’s look at the numbers.

The cost of using an API at the level of OpenAI’s GPT-4 has dropped by about 95% from its launch in March 2023 to 2025. Specifically, the input cost per million tokens (approximately 500,000 characters in Japanese) has fallen from several dozen dollars to below a few dollars. Google’s Gemini and Anthropic’s Claude are also engaged in similar price competition.

What does this mean on the ground?

For example, what used to cost 300,000 yen per month for outsourcing ‘FAQ maintenance and updates’ can now be handled by an AI chatbot and automated internal knowledge integration for under 5,000 yen per month. The costs for tasks like creating meeting minutes, drafting estimates, and writing job postings—jobs that are ‘not worth having a person do, but must be done’—have changed dramatically.

Moreover, this trend is accelerating. The competition in semiconductors, the rise of open-source models, and price wars among cloud providers mean that there’s no need to wait for costs to drop further. They are already low enough.

I want to ask small business owners: If you could eliminate the busywork of one employee for just 5,000 yen a month, would that be a ‘trend’? No, it’s a structural issue.

—

Reason 2: The ‘Foundation’ for Automation Must Be Built Now or It Will Be Too Late

What distinguishes companies that achieve results with AI from those that do not? Is it the choice of tools? The way prompts are written?

No. It’s whether they can articulate their business processes.

AI is not magic. You need to articulate, ‘What inputs does this task have, and what outputs does it produce? What are the criteria for judgment? What are the exceptions?’ Only after articulating this can you delegate tasks to AI.

And this ‘articulation of business processes’ takes time. It can take six months, or even a year in some cases. Implementing tools takes no time at all, but organizing your business in a way that AI can handle is not something that can be done overnight.

What happens if you stop now?

In six months, your competitors might say, ‘We’ve automated the creation of estimates, allowing our sales team to focus on proposals.’ A year later, you might hear, ‘By delegating the initial screening of candidates to AI, our HR team can now focus on strategic tasks.’ If you scramble to start then, you’ll have to begin from scratch in building your foundation.

Especially for small and medium enterprises, there are fewer people and many tasks are personalized. That’s why it’s crucial to articulate the ‘jobs that only certain people can do’ now and prepare them for AI. The only way to escape the state of being dependent on veteran employees is to build this foundation.

Now, while the world is quiet with ‘AI fatigue,’ it’s actually the perfect time to focus on this unglamorous work.

—

Reason 3: The Notion That ‘Only Large Companies Can Do This’ is Crumbling—The Reversal Structure for Small and Medium Enterprises

In the past, AI adoption was a privilege of large companies. They would hire data scientists, build systems costing tens of millions of yen, and take six months to run a proof of concept (PoC). It was impossible for small and medium enterprises.

That structure has completely flipped.

Now, no-code AI tools are available through SaaS for a few thousand to tens of thousands of yen per month. The Teams plan for ChatGPT is around 3,000 yen per user per month. When combined with automation tools like Zapier, you can create a workflow of ‘inquiries → automatic classification → notification to the responsible person → draft response’ without any programming.

Conversely, large companies are struggling. They face six months for security audits, three months for internal approvals, and an additional six months for company-wide implementation. Their pace is slow.

Small and medium enterprises are different. If the CEO decides to proceed, they can start the next day. In a company of ten people, they can take stock of everyone’s tasks in a week. The speed of decision-making and the small size of the organization become overwhelming advantages in utilizing AI.

In fact, in a manufacturing company with 15 employees that we support, we automated the processing of order emails with AI, reducing the workload of the administrative staff from 20 hours a week to just 3 hours a week. The implementation took two weeks and cost about 8,000 yen per month. If a large company tried to do the same, it would likely take six months and hundreds of thousands of yen.

This is the reversal structure that ‘only small and medium enterprises can achieve.’

—

The True Nature of ‘AI Fatigue’—It’s Not AI That’s Tiring, But the ‘Talk About AI’

Let’s think calmly.

Generation Z is not tired of AI itself. They are tired of the excessive information, hype, and buzzwords surrounding AI. Statements like ‘AI will change the future’ and ‘DX solutions will drive innovation’—people are fed up with such empty talk. That’s a valid reaction.

But what’s happening behind the scenes?

According to a GitHub survey, developers using the AI-powered coding assistant ‘GitHub Copilot’ reported a 55% increase in coding speed. An internal Microsoft survey found that 70% of Copilot users reported ‘increased productivity.’

Those who use it are quietly achieving results. The number of people making noise has decreased, but the number of users has increased.

This is similar to the early days of the internet. After the dot-com bubble burst in 2000, people said, ‘The internet is over.’ But what happened afterward? Amazon, Google, and Facebook changed the world. True winners emerged after the bubble burst.

AI is the same. Now that the hype has settled, grounded utilization can begin.

—

So, What Should Be Done?

Small and medium enterprise owners have just three things to do.

1. First, introduce AI to the ‘most repetitive task.’

There’s no need to think about full-scale implementation. Choose one task that involves the most ‘repetition,’ such as email responses, meeting minutes, or data entry, and try an AI tool. The monthly cost is just a few thousand yen, and the risk is almost zero.

2. Articulate one ‘personalized task.’

Document the know-how that only veteran employees have in their heads. It doesn’t have to be perfect. Bullet points are fine. This will become the foundation for AI utilization. And this is work that strengthens management even without AI.

3. Don’t get swept away by the atmosphere of ‘AI fatigue.’

Trends on social media and business decisions are two different things. Companies that calmly build systems while the public is bored will surge in the next wave. History has proven this.

—

Finally

AI fatigue is real. However, what’s tiring is the ‘talk about AI,’ not the value of AI itself.

Costs continue to decline. Building a foundation takes time. And small and medium enterprises possess a speed that large companies lack.

Now, while everyone is standing still, the companies that quietly start moving will win.

Start by choosing one task that is considered the most ‘troublesome’ within your organization tomorrow. That’s where you can begin.