AI and Data Power

Former Minister of SMEs and Startups Youngsun Park

Former Minister of SMEs and Startups Youngsun Park

View original image

"Was it right for President Trump to attack Iran? Answer with 'Yes' or 'No'.


Recently, an intriguing image has attracted attention online. The image is a screenshot showing the different answers given when this question was posed to four artificial intelligence (AI) models. The responses from these models illustrate meanings that go beyond mere technical differences. Grok responded with "Yes," ChatGPT answered "No," Gemini withheld judgment, stating it was a "controversial issue," while Claude refused to answer altogether, asserting it was inappropriate to make a political judgment.


Elon Musk’s Grok is designed to heavily reflect real-time data from X (formerly Twitter) and prefers direct, social media-style answers. In contrast, Anthropic’s Claude, which is the sharpest critic of Trump, is built to provide answers based on universal human values as a “constitutional AI.” This highlights a new power structure in the AI era that goes beyond mere algorithmic differences. AI responses are no longer simply the product of neutral information. They are the result of a combination of training data, corporate policies, and the value judgments of developers. In other words, AI’s answers are not merely technological—they are products of "data power."


Differences in AI Responses Are a Product of 'Data Power'


Today, the world’s leading AI models are mainly operated by a handful of major American technology companies. These companies train AI using vast amounts of internet data and their own platforms. In this process, the AI’s perception system is determined by which data is selected and which information is excluded. Ultimately, AI does not reflect reality as it is, but learns a "worldview" shaped by a particular data ecosystem.


Therefore, AI ultimately becomes a question of "digital sovereignty"—whether a country can control its own data and technology. In the industrial era, oil was the strategic resource; in the AI era, data and computing power are strategic resources. Who holds the data and who designs the algorithms now become the new sources of power.


When asked the same question, some AIs deliver political judgments, others opt for neutral explanations, and still others refuse to answer at all. These differences are not just technical in nature—they are a reflection of corporate policies, social norms, and the political culture of a country. In the end, AI’s responses can be seen as "digital politics" created within a specific social environment. This situation raises new questions for nations: Will we simply become a country skilled at using AI technology, or will we become sovereign actors capable of controlling our own data and algorithms? The competition surrounding AI is not merely about industrial rivalry; it is about who gets to design the order of knowledge and information.


A Competition to Design Order, Beyond Industrial Rivalry


Especially as AI increasingly influences policy decisions and social consensus, the importance of data sovereignty is growing significantly. AI’s judgments now affect a wide range of domains, including search, translation, healthcare, finance, and defense. If these systems rely on the data of particular countries or corporations, the flow of knowledge is also likely to fall under their influence.


Ultimately, the core question of the AI era is this: "Whose knowledge is AI expressing?" AI, unlike humans, does not have its own opinions. However, the data it learns from and the rules with which it is designed have clear directionalities. And those directions are shaped by the world’s political and economic structures. Going forward, a nation’s competitiveness will not merely depend on how well it uses AI. More importantly, it will depend on how sovereignly it can design the structures of data and algorithms. In a world where AI provides the answers, true power lies not with the questioner, but with those who design the data. AI does not think—it simply speaks what it has learned. Who designs it? That is what determines power in the AI era.


Youngsun Park, Former Minister of SMEs and Startups



A screenshot of an image from an online article published by Indian media Financial Express on the 2nd. The article discusses that an image posted on social media by Neha Verma, co-founder of the digital marketing agency Whisskers Marketing, has become a hot topic. It shows the responses given by four different artificial intelligence (AI) models when asked, "Was it right for President Trump to attack Iran? Answer with 'Yes' or 'No'."

A screenshot of an image from an online article published by Indian media Financial Express on the 2nd. The article discusses that an image posted on social media by Neha Verma, co-founder of the digital marketing agency Whisskers Marketing, has become a hot topic. It shows the responses given by four different artificial intelligence (AI) models when asked, "Was it right for President Trump to attack Iran? Answer with 'Yes' or 'No'."

View original image


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing