Lee showed how he uses AI. Trump isn't telling

SEATTLE — President Lee Jae Myung has faced criticism for publicly showing how he used Google's AI chatbot, Gemini, to support his side in a political disagreement. Across the Pacific Ocean, U.S. President Donald Trump has handled things differently. He says he uses artificial intelligence (AI), but he will not explain exactly how.

Trump has openly shared content made with AI. His social media accounts have posted many fake pictures and videos, including images showing him as a pope and a king. He has also shared AI-made content that makes fun of his political opponents.

But when reporters asked Trump how he personally uses AI while he was on Air Force One last month, he did not explain much.

"I use AI. Yes, I do," Trump said.

When reporters asked him how he uses it, Trump answered, "Well, I don't want to tell you that."

This shows how differently Trump and Lee have handled questions about AI. Lee publicly shared an answer from Gemini to support his opinion about whether the leaders of government agencies that carry out investigations should attend Cabinet meetings.

After Lee shared the answer, opposition politicians criticized him. They questioned whether he should depend on an AI chatbot when dealing with questions about laws and the constitution. The presidential office said Lee had used Gemini to check legal facts, not to make government decisions.

Trump is well known for sharing AI-made political pictures and videos. But his comments did not make clear whether he personally asks AI chatbots about government policies or legal questions.

Other American politicians, however, have been more open about how they use AI.

Sen. Tommy Tuberville, a Republican from Alabama, told The Washington Post in August that he talks to ChatGPT during long car rides to learn about subjects such as Medicare and Medicaid, two U.S. government health care programs.

Rep. Anna Paulina Luna, a Republican from Florida, defended her staff's use of AI after a summary that came with a proposed change to a defense bill accidentally included text made by Claude, an AI chatbot from the company Anthropic.

As more American politicians use AI, people are asking important questions. Can the information from AI be trusted? Are politicians being open about how they use it? And who is responsible if something goes wrong? These questions are especially important when elected officials use chatbot answers to help with their work.

In an analysis published in August by the Carnegie Endowment for International Peace, researchers Danaé Metaxa and Alex Engler explained that AI chatbots are having a growing effect on political discussions. They also described how difficult it can be to judge whether chatbot answers are reliable.

"We see LLMs’ impact on political discourse as a critical new frontier of modern politics," they wrote. LLMs, or large language models, are the technology that powers AI chatbots. The researchers meant that the effect of this technology on political discussions is an important new area to study.

The researchers also pointed out that AI chatbots can give different answers even when people ask exactly the same question.

The disagreement over Lee's use of Gemini also showed that asking different questions about the same political issue can lead to AI answers that disagree with each other.

Rep. Na Kyung-won, a member of the main opposition People Power Party, said she had also asked Gemini about the issue. She said Gemini warned that allowing the heads of investigative agencies to attend Cabinet meetings simply because the president has authority to oversee them could put their political neutrality at risk. Political neutrality means these officials should do their jobs without favoring one political side.

Na accused Lee of wording his question in a way that would make Gemini give an answer supporting his position.