Conclusion

To put it succinctly, the cost of implementing AI has decreased. However, the “cost of trusting AI” has skyrocketed.

ChatGPT costs about 20,000 yen per month. Image-generating AI is free. It can write code, compose text, and translate. For small and medium-sized enterprises, the “cost of using AI” has approached nearly zero.

Yet, another cost has suddenly emerged: the cost of verifying whether AI outputs can be trusted.

Chatbots have nearly caused fatalities among hikers. Google searches have been inundated with AI-generated garbage. Public comments from government agencies have been polluted with fake opinions generated by AI. All of these incidents actually occurred in 2024.

Who will ultimately bear this “trust cost”? Is it a threat or an opportunity for small and medium-sized enterprises?

—

A Hiker Nearly Died on a Trail Declared ‘Safe’ by a Chatbot

There have been multiple reports of hikers who trusted trail information provided by AI chatbots and nearly faced disaster in the vicinity of national parks in the United States.

The structure of the problem is simple. Chatbots generate “plausible answers” but do not guarantee their accuracy. Real-time information such as seasonal closures, landslides, and the availability of water sources is often not included in their training data.

Despite this, AI confidently states, “This route is safe and recommended for beginners.” Users are deceived by this “confidence.”

Consider this: What would happen if your company used an AI chatbot for customer service?

What if a restaurant reservation bot incorrectly stated, “We can accommodate allergies”? What if a real estate inquiry bot lied, saying, “This property allows pets”? What if an e-commerce site for health foods had AI claim, “This product is effective for diabetes”?

The cost of AI lying is borne by the companies that use it. This is the reality.

—

Google Search is Broken—SEO for Small and Medium-Sized Enterprises is Blown Away by ‘AI Contamination’

Another pressing issue for small and medium-sized enterprises is the collapse of Google search quality.

Low-quality content generated in bulk by AI has flooded the web, occupying the top spots in search results. In 2024, Google implemented a core update aimed at eliminating AI-generated content. However, the collateral damage fell on small and medium-sized enterprises and individual sites that were genuinely creating content.

There are concrete numbers. In a niche personal project, impressions in Google Search Console dropped from 500-700 per day to nearly zero after the update. Visibility for keywords directly linked to sales vanished.

This is not just a story about personal sites. Local construction companies, town tax offices, and community-focused restaurants—small and medium-sized enterprises that spent 50,000 to 100,000 yen per month on SEO to update their blogs and generate inquiries are losing search traffic due to AI spam.

An investment of 100,000 yen in SEO can suddenly yield zero returns overnight. This is the real damage of “AI contamination.”

Conversely, there is an opportunity here. Information that AI cannot generate—on-site photos, before-and-after construction examples, customer testimonial videos, and reports on local events—will become overwhelmingly strong in future SEO. The content produced by small and medium-sized enterprises from the field will be what Google truly values, as opposed to articles mass-produced by large corporations using AI.

—

Public Comments Polluted by AI—The Breakdown of ‘Democracy by Numbers’

Let’s touch on the third issue mentioned in the title. In the United States, there have been cases where thousands of AI-generated fake opinions were submitted to public comments (public opinion solicitations) regarding government regulations.

For instance, there was a case where thousands of “opposing opinions” concerning environmental regulations were submitted, generated by AI. At first glance, they appear to be individually written opinions, but analyzing the writing style and logical structure reveals characteristics of AI generation.

While this pertains to government, it is not unrelated to small and medium-sized enterprises. Reviews on rating sites, reputations on social media, and survey responses—a world where it becomes impossible to distinguish between “genuine human voices” and “AI-generated fakes” is already upon us.

When Google Maps reviews, Tabelog reviews, and company evaluations on job sites are contaminated with “AI-generated fake reviews,” it is the earnest small and medium-sized enterprises that will suffer the most.

—

Estimating the ‘Trust Cost’

So, if small and medium-sized enterprises integrate AI into their services, how much will the “trust cost” be? Here’s a rough estimate.

Case 1: Implementing a Customer Service Chatbot

• Cost of AI chatbot implementation: 10,000 to 30,000 yen per month (SaaS usage)
• Trust Cost: 20 hours per month for verifying answer accuracy × 2,000 yen/hour = 40,000 yen per month
• Risk of complaints due to incorrect answers: 100,000 to 500,000 yen for a significant complaint once a year
• Total: The cost of ensuring trust exceeds the cost of using AI.

Case 2: Using AI-Generated Content for SEO

• Cost of AI article generation: 100 articles per month × nearly 0 yen = 0 yen
• Trust Cost: 40 hours per month for fact-checking and editing by experts × 3,000 yen/hour = 120,000 yen per month
• Risk of Google evaluation: Being classified as AI-generated and excluded from indexing → Zero search traffic = Damage impossible to quantify
• Total: The notion that “articles can be created for free” is a fantasy. When quality assurance is included, it is comparable to traditional outsourcing costs or even higher when risks are factored in.

Case 3: Using AI for Internal Operations

• AI usage fee: 20,000 yen per month (e.g., ChatGPT Team)
• Trust Cost: 30 minutes per day for output verification and correction × 20 business days × 2,500 yen/hour = 25,000 yen per month
• Total: The AI usage fee and trust cost are nearly equal. However, if the efficiency gains are significant, it can pay off.

Conclusion: A “trust cost” of 1 to 4 times the cost of using AI will arise. If AI is implemented without factoring this in, it will not only fail to reduce costs but will only increase risks.

—

So, What Should Small and Medium-Sized Enterprises Do?

I have three recommendations.

1. Treat AI outputs as “drafts” rather than “finished products.”

Responses from chatbots, AI-generated articles, and AI-created estimates are all “drafts.” They only become “finished products” once verified by a human. Skipping this step will result in trust costs being billed later in the form of complaints or accidents.

2. Invest in content that AI cannot create.

On-site photos. Before-and-after construction images. Customer interview videos. Reports on local events. These are things that AI cannot produce. Moreover, this is the information that Google values and that customers trust. As AI-generated content floods the market, the value of authentic primary information will increase. Small and medium-sized enterprises are in the strongest position as “producers of primary information.”

3. Incorporate the “trust cost” into your business plan.

Stop writing only “20,000 yen per month” in the approval documents for AI implementation. Write, “20,000 yen per month + 40,000 yen trust assurance cost + 10,000 yen risk reserve = 70,000 yen per month.” If it still pays off, then implement it. If the trust cost is kept as an invisible cost, it will eventually become a fatal issue.

—

What Will Decrease and Increase in Value Due to AI

Finally, let’s discuss the structural aspects.

Things whose value will decrease due to AI:

• The ability to write “plausible text”
• The ability to produce large volumes of content
• Routine customer service
• The “quantity” of information

Things whose value will increase due to AI:

• The ability to guarantee that information is correct
• Primary information that only exists on-site
• Trust that comes from “If this person (or company) says it, I can believe it”
• The fact that it has been verified by a human

It’s ironic. The smarter AI becomes, the more valuable the stamp of “verified by a human” becomes.

For small and medium-sized enterprise owners, this is not necessarily bad news. As large corporations advance efficiency through AI, they distance themselves from the field. Small and medium-sized enterprises are present in the field. They can see the faces of their customers. They understand the local context. They possess a “trust” that AI cannot generate.

Understanding the trust cost correctly and leveraging it as a strength is the strategy that small and medium-sized enterprises should adopt in the AI era.