Altman: "If AI defends Seoul in case of North Korea invasion? Many questions to consider"

Brookings Institute Assumes North Korea's Seoul Air Raid Scenario
Discussion on AI Entrusting Capital Defense and Enemy Killing Issues
Emphasis on Importance of AI-Based Infrastructure Such as Data Centers

Sam Altman, CEO of OpenAI, spoke about the use of artificial intelligence (AI) as a weapon in warfare. While acknowledging the high value of AI in war, he emphasized the need for a cautious approach that considers the reliability of AI’s judgment, potential human casualties, and ethical issues from multiple perspectives.

On the 7th (local time), Altman expressed these views during a discussion titled "Geopolitical Changes in the AI Era" hosted by the Brookings Institution, a think tank based in Washington.

Regarding whether South Korea should entrust AI with the defense of Seoul in the event of a sudden attack by North Korea, he responded, "There are many questions to consider, such as whether we can truly trust AI’s situational judgment and the expected scale of human casualties."

The moderator posed a hypothetical scenario where North Korea launches 100 military aircraft toward Seoul, and South Korea uses AI-controlled robotic forces to shoot down all the aircraft, resulting in the deaths of 100 North Korean pilots. The question was about under what circumstances it would be acceptable to delegate the decision to kill people to AI.

In response, Altman said, "AI might make the interception decision when aircraft are approaching South Korea and there is no time for human decision-making, but can we be really sure such an attack is happening? How certain do we need to be? What are the expected human casualties? There are really many questions (we need to consider)."

He also noted, "I have never seen anyone argue that AI should be given the authority to launch nuclear weapons, nor anyone say AI should not be used in split-second decisions like missile interception. There is clearly a gray area where AI should be involved in matters concerning human lives." He added, "This is not my area of expertise. I hope OpenAI will never have to make such difficult decisions."

Altman also shared his views on the AI industry and geopolitical competition. He identified AI compute, semiconductors, and data centers as the most important AI infrastructure for the future. AI compute refers to the computing resources required to operate AI systems. He expressed his hope that "AI technology benefits all of humanity," and stated his belief that "a fairly broad and inclusive coalition led by the United States should take the lead in expanding this AI infrastructure."

However, he said, "The idea that only the U.S. should build AI data centers will not work, and it would not feel good to me or to other countries around the world." He continued, "We will have many clashes with China regarding AI, but I hope everyone shares the goal of reducing the catastrophic risks of AI."

Altman also mentioned the potential misuse of AI in elections. He said, "Companies are intensely focused on preventing AI from being used to interfere with elections," and emphasized, "We will respond to this issue with paranoia until the elections are over." He added, "People are increasingly wary of AI-generated fake news, so it will not be easy for AI to influence elections in the same way as before, but new threats may emerge."