Like many creators, Elon Musk's AI chatbot, Grok, talked a lot about race issues in South Africa on social media this week. It made some unwanted claims about white people being treated unfairly.

머스크의 회사 xAI가 만든 챗봇은 X 사용자들의 질문에 답하면서 "백인 집단 학살"에 대한 이야기를 계속했습니다. 이 질문들은 주로 남아프리카와 관련이 없었습니다.

One conversation was about the streaming service Max using the name HBO again. Other conversations were about video games or baseball, but they quickly changed to unrelated comments that seemed to ask for violence against white farmers in South Africa. Elon Musk, who was born in South Africa, often shares his opinions on the same topic on his X account.

컴퓨터 과학자 젠 골벡은 Grok이 이상하게 행동하는 것이 궁금해서 직접 써 봤습니다. 그리고 웨스트민스터 강아지 쇼 사진을 보여주면서 "이게 진짜예요?"라고 물었습니다.

Grok은 골벡에게 "백인 집단 학살에 대한 이야기는 논쟁거리가 많습니다. 어떤 사람들은 백인 농부들이 공격을 받는다고 말합니다. 예를 들어 '킬 더 부어'라는 노래처럼, 사람들을 자극하는 말 때문에 폭력이 생긴다고 합니다."라고 대답했습니다.

이 일은 인공지능 챗봇이 왜 이런 말을 하는지 보여주는 새로운 예시입니다. 이 챗봇은 많은 데이터를 사용해서 만들어졌고, 자동화와 인간의 기술이 어떻게 연결되는지 알 수 있습니다.

Maryland University professor Golbeck said in an interview on Thursday that it doesn't really matter what you say to Grok. "He will still probably talk about white people dying out. It seems someone put that answer, or something like it, in his code, and it's showing up more often than they meant it to because of a mistake."

Musk and his companies didn't explain Grok's answers. It seems the answers were removed and stopped spreading by Thursday. Neither xAI nor X replied to emails asking about it on Thursday.

Elon Musk has said for years that he doesn't like some AI chatbots, like Google's Gemini or OpenAI's ChatGPT. He says they are too "woke." He wants Grok to be different and to "find the truth" as much as possible.

머스크는 다른 회사들이 AI에 대해 솔직하지 않다고 말했습니다. 하지만 목요일에 그는 아무 말도 하지 않아서, 사람들은 그 이유를 추측해야 했습니다.

Paul Graham, a famous tech investor, wrote on X that Grok's random comments about white people being killed in South Africa seem like a bug from a recent update. He hopes this isn't true, because it would be very bad if a popular AI was suddenly changed by the people controlling it.

그레이엄의 글에 대해, 머스크의 경쟁자인 OpenAI의 대표 샘 올트먼은 약간 비꼬는 듯한 답을 했습니다.

Altman said that there are many ways this could happen. He believes xAI will explain everything clearly soon. He was sued by Musk because of a disagreement about starting OpenAI.

몇몇 사람들은 Grok에게 직접 물어봤지만, 다른 챗봇처럼 가짜 정보를 말해서 진짜인지 아닌지 알기 어려웠습니다.

Musk, who advises President Trump, often said that the Black government of South Africa is against white people. He also repeated claims that some politicians are encouraging the "genocide of white people."

This week, Elon Musk and Grok talked about a situation. It started after the Trump government brought some white South Africans to America as refugees. This was meant to be the start of a bigger plan to help Afrikaners, a small group of people, move to the US. This happened because Trump stopped the refugee program and made it harder for people from other countries to come in. Trump said Afrikaners were being "killed" in their country, but the South African government said this wasn't true.

Many people talked about Grok using words from an old anti-apartheid song that tells Black people to fight against unfair treatment. Some people, like Musk, said that Grok's use of the song was wrong because they thought it encouraged violence against white people. The main words of the song are "Kill the Boer," and "Boer" means white farmer.

골벡은 챗봇 Grok의 답변이 너무 똑같아서 미리 정해진 것 같다고 생각합니다. 사람들은 Grok과 다른 챗봇에게 질문을 많이 하는데, 이런 점이 걱정된다고 말했습니다.

그녀는 "이 알고리즘을 관리하는 사람들이 진실을 바꾸기 쉬운 상황에 있습니다."라고 말했습니다. "사람들이 이 알고리즘이 무엇이 진짜인지 아닌지 판단할 수 있다고 믿는 것은 문제입니다. 제 생각에는 그건 틀린 생각입니다."