The conversation about what AI can do has become tiresome. The real issue is that the cost of proving what ‘humans have done’ is quietly but surely skyrocketing.

Sony’s table tennis robot has defeated professional players. There are suspicions that the Pope’s AI warning message was generated by AI. At first glance, these seem like unrelated news items, but they share a common root: ‘Did a human do this?’—the cost of answering this question is rising across all fields.

For small and medium-sized business owners, this is not just someone else’s problem.

—

The Reality Presented by Sony’s Table Tennis Robot ‘Ace’

The table tennis robot ‘Ace’, developed by Sony AI, has won against elite players under the official rules of the International Table Tennis Federation (ITTF). It became the first robot to defeat a top human player by winning three sets in a five-set match, adhering to official rules.

The key point is that it did not win by sheer physical power. Ace has arms comparable in size to a human’s and wins through millisecond-level judgment and precise ball placement, rather than brute strength. In other words, it has covered the most human-like aspect of table tennis—’thinking and hitting’.

There have been instances of robots defeating humans in sports before, such as chess (1997, Deep Blue) and Go (2016, AlphaGo). However, those were battles of ‘intellect’. Table tennis is different. A ball coming at you at 0.2 seconds requires recognition, judgment, and physical action to be completed in that time frame. In this fusion of intelligence and physicality, robots have surpassed humans. This is something new.

So, what happens next?

The very act of ‘human play’ gains value. If a robot can do the same thing, what is the point of having a human do it? The cost of explaining and proving that value will arise. In the world of sports, ‘doping tests’ serve this purpose, but we may soon enter an era where ‘proof of not receiving robotic assistance’ becomes necessary.

—

The Irony of the Pope’s AI Warning Being ‘AI-Generated’

Pope Francis has issued warnings about the dangers of AI. However, there are suggestions that the message itself may have been generated by AI.

The irony is striking.

What stands out in this matter is the Chrome extension released by Pangram Labs, which labels text and images on social media as ‘possibly AI-generated’. In a way, it serves as an automatic ‘ingredient list’ for content on the browser.

However, there is a structural problem here. The ‘tool for detecting AI generation’ itself operates on AI. AI distinguishes what AI has created. Humans trust that judgment. This nested structure fundamentally complicates the proof that something was ‘created by humans’.

Currently, the accuracy of detecting AI-generated text varies according to research, but reports indicate a false detection rate of about 10-30% for outputs from models like GPT-4. This means there is a risk that text written by humans could be misidentified as ‘AI-generated’. Conversely, it is often possible to avoid detection by making minor adjustments to text written by AI.

Detection costs continue to rise, while detection accuracy hits a ceiling. This is the reality.

—

‘High-Quality Chaos’—The State Named by the Security Industry

Recent cybersecurity reports have referred to the current AI environment as ‘High-Quality Chaos’.

What is happening? The quality of phishing emails, fake reviews, and impersonation images generated by AI has dramatically improved. What used to be distinguishable by ‘awkward Japanese’ or ‘unnatural images’ is now at a level where even professionals find it difficult to discern.

In numerical terms, research shows that the click-through rate for AI-generated phishing emails is up to 40% higher compared to traditional manually crafted emails. The creation cost is nearly zero. For less than a few yen per email, persuasive messages that could have been written by professional copywriters can be mass-produced.

On the other hand, what about the costs for those defending against these attacks? The annual security budgets for mid-sized companies range from several million to tens of millions of yen. Employee literacy training, email filtering, and the implementation of multi-factor authentication—all come at a cost.

The cost of attacks approaches near zero, while the cost of defense continues to rise. This asymmetry is the essence of ‘High-Quality Chaos’.

—

What This Means for Small and Medium-Sized Enterprises

Now, to the main point.

The structural change of rising costs to prove ‘human agency’ is both a threat and an opportunity for small and medium-sized enterprises.

Threat: The Cost of Proving Trust Will Burden Management

Let’s look at some specific examples.

• Reviews and Testimonials: With the flood of AI-generated fake reviews, the cost of proving that your company’s genuine reviews are ‘real’ will arise. Analysis suggests that in some categories on Amazon, AI-generated reviews already account for over 30%.
• Copyright of Creations: There is a risk that photos taken by your company or blog posts written in-house could be questioned with ‘Was this made by AI?’ In fact, there have been increasing cases where illustrators post hand-drawn works on social media only to be misidentified as ‘AI art’, leading to backlash.
• Hiring: The cost of determining whether applicants’ resumes or portfolios are AI-generated. A survey by a recruitment agency indicates that by 2024, about 45% of application documents will be suspected of AI involvement.

All of these will weigh on management as ‘proof costs’. Large companies can afford to have specialized teams for this. Small and medium-sized enterprises do not have that luxury.

Opportunity: The Value of ‘Visibility’ Will Soar

However, consider the opposite perspective.

In an era where AI can create anything, the value of being able to see ‘this person, in this place, doing it this way’ will only increase.

Local small and medium-sized enterprises have this advantage.

• The faces of factory artisans are visible
• The CEO communicates directly on social media
• There are long-standing relationships with local customers
• Manufacturing processes can be shared in videos

All of these are assets that lower the cost of ‘proving human agency’. A craftsman from a local factory filming work processes on a smartphone is far more trustworthy than a large corporation issuing certificates via blockchain. The cost is also significantly lower.

In an era where proof costs are rising, small and medium-sized enterprises that are already ‘visible’ will have an advantage. This is the structural reversal point.

—

So, What Should We Do?

Here are three actionable steps to take starting today.

1. Record and Share Processes

Capture the processes of product creation or service delivery on your smartphone and share them. Minimal editing is fine. In fact, the more ‘raw’ it is, the more trustworthy it becomes. The cost is almost zero, but the impact as proof of ‘human agency’ is immense.

2. Honestly Label AI-Generated Content

Clearly state ‘Created using AI’ for any content produced with AI. Being honest is more trustworthy than hiding it. In the future, many countries will move toward mandatory labeling of AI-generated content. If you do it first, you won’t be caught off guard when regulations come.

3. Redefine ‘Work That Is Worth Doing by Humans’

Tasks that are faster, cheaper, and more accurate when done by AI should be handed over to AI. Redirect the freed-up time to ‘work that has value because it is done by humans’. Customer service, consultation, and customized proposals—these types of jobs will become more valuable as AI evolves.

—

Conclusion: Turning Asymmetrical Costs to Your Advantage

Sony’s table tennis robot has expanded the range of what machines can do. The Pope’s AI warning incident has highlighted the reality that distinguishing between what is made by humans and what is not is becoming increasingly difficult.

Both point in the same direction. The cost of proving ‘human agency’ will accelerate upward from here.

Large corporations will attempt to solve this issue with technology and funds. Blockchain, digital watermarks, AI detection tools—all come at a high price.

Small and medium-sized enterprises can take a different approach. Show your face. Share your processes. Build trust locally. These are proofs based on relationships, not technology. The costs are low, and imitation is difficult.

The evolution of AI will not stop. Therefore, companies that establish methods to prove ‘being human’ now will come out on top.

You don’t need expensive systems. Just a smartphone and honesty will suffice.