Does that title strike a chord?
I’d be pleased if your reaction was, “That’s a surprisingly good way to put it.” But I imagine some of you are thinking, “What on earth are you talking about?”
In Hans Christian Andersen’s The Emperor’s New Clothes, two swindlers claim to make clothes that are invisible to anyone who is stupid or unfit for their position. Nobody wants to admit they cannot see the clothes—not even the emperor.
That is why I think the story is such a good metaphor for AI today. Using AI has somehow become a test of executive competence. A business leader may not really understand what is happening inside the AI black box, or whether the investment is actually paying off. But they do not want investors, other executives, or their own employees to think they are incompetent. So they announce, “We’ve adopted AI, and we’re seeing results.”
AI becomes the magical outfit that only competent people can see. Investors and employees do not want to look incompetent either, so they applaud the leader who put it on.
Meanwhile, has AI actually made their own work more efficient? My impression is that, in many cases, people are doing little more than drafting emails with ChatGPT or checking for typos. The benefits come nowhere close to justifying the investment.
I should be clear about my own position: I work on the vendor side of this arrangement. And what are AI vendors doing? We are the tailors, flattering executives and the people responsible for digital transformation as we dress them in AI. That is how I see the situation, anyway. I suppose I have to make my peace with it—it is the job, after all.
How Did AI Become an Outfit Only the Competent Can See?
I think a major reason we ended up here is that the benefits of adopting AI are so difficult to measure.
Compared with many other IT systems, AI makes even the basic question of whether an output is correct difficult to answer. Evaluation often comes down to something as vague and subjective as, “Do you feel this is useful in your work?”
Seriously measuring the impact of AI is hard work. In practice, assessments tend to stop at usage rates or satisfaction surveys—indicators that leave plenty of room for a favorable interpretation. Organizations that can actually quantify the benefits in money or time are, in my experience, few and far between.
The spread of generative AI has made this tendency even stronger. Putting a business value on its open-ended outputs is genuinely difficult.
That gives the executives who adopt AI plenty of room to exaggerate their achievements.
Meanwhile, “results from AI” has become something of a password for attracting investors. Announce those results loudly enough, and the market responds. The share price goes up. That is the mechanism we seem to have created.
Failed AI investments are easy to conceal, too. And if somebody does have to take responsibility, executives and digital transformation departments can blame the vendor that supplied the inscrutable black box. The vendor, in turn, can blame users for failing to make proper use of it.
Within this distorted arrangement, vendors sell AI as a magical outfit. Executives put it on and become naked emperors. Investors and employees gather around to praise them, even though nobody can actually see what the investment has achieved.
Executives at competing companies watch from the sidelines. They do not quite understand what happened, but they believe the company next door has achieved something with AI. They start to panic: “We need to adopt AI, too.” Vendors catch the scent of that anxiety and arrive with another magical outfit to sell.
Now, I make my living tailoring and selling AI, but I do not think the clothes are entirely imaginary. As an IT technology, AI is reasonably—well, perhaps quite—powerful. I just think the gap between public expectations and what AI actually delivers is enormous.
Big Claims About AI Tend to Be Short on Specifics
Creating business value with AI requires carefully choosing problems that AI can actually help solve. The use case and the data have to fit. Getting to that point takes thorough preparation.
To stay with the clothing metaphor, an outfit called AI can sometimes keep out the cold or protect you from injury. But it is no dazzling royal robe that elegantly solves every problem. It is workwear: a practical tool for dealing, patiently and painstakingly, with problems on the ground.
When someone starts presenting it as a luxury brand and showing off exaggerated achievements, that is when I start to see a naked emperor.
In fact, people talking about the achievements and potential of AI—especially generative AI—often offer accounts so abstract that there is barely a concrete example to be found.
AI itself is, after all, one technical component of a system. In that sense, it belongs alongside questions such as which cloud platform to use, how to implement security, or which programming language to choose. These are concrete matters for the people designing and building systems and applications. You could argue that it is odd for executives and investors to fixate on that particular component in the first place.
The Deeper Problem: Nothing Else Looks Like a Promising Investment
I suspect the deeper problem is that executives have run out of ideas for addressing company-wide challenges. Or perhaps the remaining options are painful enough that they would rather avoid them. They still need to announce something, though, so they cling to AI and its glittering promise.
You cannot exactly stand in front of investors and employees and say, “We’ve run out of ideas.” And talking about AI’s dreams and possibilities is much more pleasant—for speaker and audience alike—than discussing layoffs or restructuring.
Investors, meanwhile, are searching the world for somewhere to put their money. My view is that they do not see another opportunity with the promise that the internet or cloud computing once offered. They have money, but nowhere sufficiently attractive to invest it. So they end up pouring it into AI.
Even within IT, we have seen technologies such as 5G, VR, and Web3 attract enormous attention, only to fade from the spotlight without delivering the results people expected. Is there really anything other than AI that investors currently find as compelling?
Looking at the AI bubble, I get the impression that whether AI creates meaningful value has become almost beside the point. “AI” is the rallying cry for concentrating capital, driving up share prices, and getting the money out before things start to look bad. It feels like AI is being used as the vehicle for a speculative game.
Wouldn’t that mean enormous losses when the bubble bursts? Perhaps. But I do not think this lack of attractive destinations for capital is limited to IT. Looking at where humanity as a whole might commit its resources, there seems to be a shortage of convincing alternatives. So the AI sector remains awash with money.
I do think the bubble will burst eventually. But one reason the money stays put is that it has nowhere else to go.
Perhaps we should think a little harder before buying into the dream of AI.
One Last Thought About the Emperor
On a slightly different note, I have always thought The Emperor’s New Clothes is a rather strange story.
Suppose the tailors had been telling the truth, and clever people really could see a beautiful outfit. Surely not every person in the kingdom would be clever. Some would inevitably be foolish. Was the emperor really happy to parade his flabby, thoroughly overindulged body in front of those subjects?
That seems odd, doesn’t it?
Maybe it would be one thing if he had a gleaming, muscular physique. But the emperors in the picture books always seem to be a little pudgy and not especially easy on the eye.
Of course, the whole thing is a metaphor. I imagine there were figures in the politics or markets of Andersen’s time who had become naked emperors, and ideas that served as their magical clothes—much as AI does today.
If so, perhaps this AI boom is simply history repeating itself: another expression of human foolishness.
I do not see many people making this argument. Am I wrong? Or does nobody want to admit they cannot see the magical outfit, for fear of being called incompetent?
I have considered myself incompetent from the start, so that part does not bother me.
Until next time.
This article is an English translation, adapted for an international audience, of 「AIが裸の王様になった経営者の魔法の服である理由」, originally published on January 22, 2026, on the author’s personal blog, pickerlab.net.
Comments