[Bread-Baking Typewriter] Prepare for AI Lies

Henry Kissinger, an American international political scientist, sat in his office holding a laptop. What he faced was the contemplation of changes brought about by the advancement of artificial intelligence (AI) technology. Questions constantly echoed in his mind, troubling him: "How will AI transform our society?" In the quiet office, Kissinger poured his thoughts onto paper. He realized that the coexistence of AI and humans would be more complex than expected.

The world-renowned scholar Henry Kissinger captured these concerns in his book The Age of AI. Former Google CEO Eric Schmidt and Daniel Huttenlocher, the inaugural dean of the MIT Schwarzman College of Computing, also participated. They all warned about the impact of AI on human decision-making. While AI analyzes vast amounts of data and makes predictions, its decisions are not always correct, so people must always review AI's outputs. At the same time, they emphasized that technology can develop as a tool to complement human intelligence, urging cooperation with AI.

The above content is a review written by ChatGPT after reading the preface of the book The Age of AI. A reporter instructed ChatGPT to summarize the preface titled "ChatGPT Heralds an Intellectual Revolution" in an article format. The reporter also requested an introduction of the authors including Kissinger and content about the future AI will bring. ChatGPT produced the output according to the requirements. It took just over a minute to write two paragraphs. The subject-verb agreement and grammar were perfect.

While the writing skill is impressive, there is another point to note: it contains falsehoods. First, there is no known information about how Kissinger agonized in his office. This was entirely fabricated by ChatGPT. Also, Kissinger did not demand an attitude of cooperation with AI. The reason why ChatGPT lied is unknown. What is clear is that ChatGPT's writing ability is so advanced that it is difficult to distinguish from a human, and humans find it hard to detect.

The authors' greatest fear is precisely AI's falsehoods. AI's skillful lies are hard to distinguish. Efforts to identify fake news are not enough. Even with long and careful scrutiny, the truth may not be discerned. So-called generative AI learns and refines vast amounts of data accumulated by humanity on its own. This data contains all kinds of underestimations, overestimations, fake news, confirmation biases, prejudices, and hatred mixed together. No matter how many brilliant humans are gathered, it is not easy to discern AI's information, which has acquired materials from all fields.

The authors worry that AI's intelligence and falsehoods might create "mystical fanaticism." In the past, many followers were fascinated by figures, technologies, and ideologies that were difficult to approach. Those who possess AI technology that processes reality excellently are likely to seize strong power in society. Differences in AI capabilities could lead to local and international changes. The authors even say, "There is potential for social alienation as well as revolution."

Humanity stands at a historic and critical turning point with the emergence of AI. Concerns such as AI helping college students with reports or jobs disappearing may be very peripheral topics. Discussions about moral values and social principles must take place in the AI field. Humanity must hold AI accountable, set limits, and establish strong ethical principles. Otherwise, as the authors warned, "machines might trap us forever."

The Age of AI | Henry Kissinger, Eric Schmidt, Daniel Huttenlocher | Willbook | 296 pages | 17,820 KRW