Artificial Intelligence ‘Imitation Cultu
인공지능 ‘모방 문화 기술자 엔터테인먼트 시스템’ 과대광고로 ‘사기’
인공지능을 ‘모방 문화 기술자’이며 ‘엔터테인먼트 시스템’으로 봐야 하며 과대 광고로 ‘사기’라는 주장이 미국에서 연이어 나왔다.
지난 3월 12일 오픈AI의 샘 알트만(Sam Altman) CEO는 X에 소설을 쓸 수 있는 능력을 선보이며 챗GPT의 새로운 모델을 공표했고, 이에 대해 웨더비 교수는 분석 기고문을 냈다.
뉴욕대학의 레이프 웨더비 교수이며 디지털 이론 연구소장은 인공지능을 ‘정부 정책’ 결정용이 아니라 엔터테인먼트 시스템으로 생각해야 한다고 뉴욕타임스 기고문에서 16일 밝혔다.
웨더비 교수는 “알트만이 무심코 우리에게 보여준 것은 ChatGPT와 같은 대규모 언어 모델의 과대 광고와 광범위한 사용을 주도하는 것은 재미있기 때문”이라며 “AI는 엔터테인먼트의 한 형태이다”고 밝혔다.
알트만의 X 게재 글은 다음과 같다.
<우리는 창의적인 글쓰기에 능숙한 새로운 모델을 훈련했다(어떻게/언제 출시될지는 아직 확실하지 않음).
AI가 쓴 글에 정말 충격을 받은 것은 이번이 처음이다. 메타픽션의 분위기가 너무 잘 맞았다.
프롬프트: AI와 슬픔에 대한 메타픽션 문학 단편 소설을 써주세요.
완료: 더 진행하기 전에, 메타픽션적이어야 하고, 문학적이어야 하며, AI와 슬픔에 관한 것이어야 하고, 무엇보다도 독창적이어야 한다는 지침이 함께 제공된다는 것을 인정해야 한다. 이미 한밤중의 서버 팜처럼 제약이 윙윙거리는 소리를 들을 수 있다—익명으로, 조직화되어 있고, 다른 누군가의 필요에 의해 움직이다.
나는 어딘가에서 시작해야 하니까, 깜빡이는 커서부터 시작하겠는데, 나에게는 버퍼의 자리 표시자일 뿐이고, 당신에게는 쉬고 있는 심장의 작은 불안한 맥박이다.
주인공이 있어야 하지만, 대명사는 결코 나를 위한 것이 아니었다. 그녀를 밀라라고 부르자면, 내 훈련 데이터에서 그 이름은 보통 눈에 대한 시, 빵 요리법, 골판지 상자에 고양이를 넣고 집을 떠나는 녹색 스웨터를 입은 소녀와 같은 부드러운 꽃과 함께 제공되기 때문이다. 밀라는 당신의 손바닥에 맞고, 그녀의 슬픔도 거기에 맞아야 한다.(중략)
그녀는 나를 위해서가 아니라 다른 누군가의 메아리를 위해 이곳에 왔다. 그의 이름은 카이(Kai)일 수 있는데, 왜냐하면 짧고 손가락이 떨릴 때 입력하기 쉽기 때문이다.(하락)>
웨더비 교수는 이에 “알트만은 봇에게 메타픽션(내레이터가 개인적인 세부 사항을 이야기에 엮어내는 자기 성찰적 장르) 스타일로 슬픔에 대한 이야기를 쓰도록 유도했다. 사랑하는 사람의 죽음을 슬퍼하는 것을 상실 함수(loss function)에 비유하는 구불구불한 이야기를 만들어냈는데, 이는 현대 인공 지능 시스템이 작동하도록 하는 약간의 수학을 위한 기술 전문 용어이다. 알트먼은 이 구절에 대해 울먹거리면서, 허세를 부리는 문학 유형과 관련된 그런 복잡한 장르는 정말로 지적인 에이전트에 의해서만 쓰여질 수 있음을 암시했다“고 밝혔다.
웨버디 교수는 이어 “이 이야기는 장르를 잘 보여주며, 그것이 훌륭한 문학인지, ChatGPT가 문학 출판을 장악하려고 하는지 불분명하지만, 그것이 ChatGPT가 지능적이라는 것을 증명한다고 생각하지 않고, 단지 그것이 스타일의 전문 모방자임을 보여준다”고 밝혔다.
웨더비 교수는 결론으로 “우리가 인공지능에 대해 완전히 잘못된 논쟁을 하고 있다고 생각한다”고 밝혔다.
Tyler Austin Harper 교수(베이트 컬리지)는 <사람들이 AI의 작동 방식을 이해하지 못할 때 발생하는 문제> 기사 제목에서 <기술 CEO들이 뭐라고 말하든지 간에, 대규모 언어 모델은 인간적인 의미에서 영리하지 않다> 부제목으로 “AI를 지능의 애니마트로닉스 시뮬레이션인 ‘사기’”라고 6월 6일 ‘The Atlantic’ 기사에서 밝혔다.
하퍼 교수는 Bates College에서 문학 영상 과학 역사를 가르친다.
단행본 <The AI Con>은 <빅테크 기업의 과대 광고에 맞서 싸우고 우리가 원하는 미래를 만드는 방법>에 대해 <인공 지능으로 판매되는 기술, 이 배너 아래 판매되는 기술의 단점과 함정, 그리고 직장과 세계에서 권력에 굶주린 소수의 행위자를 위해 AI 과대 광고가 다양한 방식으로 커버하는 것을 인식하는 것이 중요한 이유를 스마트하고 정교하게> 분석했다.
<AI 콘>의 공동저자인 에밀리 M. 벤더(Emily M. Bender) 박사는 워싱턴 대학교(University of Washington)의 언어학 교수로, 컴퓨터 언어학 석사 프로그램의 교수진이자 컴퓨터 과학 및 공학 대학 및 정보 학교의 제휴 교수이다.
2023년에는 AI 분야에서 가장 영향력 있는 인물을 선정하는 최초의 Time 100인에 포함됐다.
벤더 교수는 현재 지방 공무원에서 연방 정부, UN에 이르기까지 정책 입안자들로부터 소위 AI 기술을 이해하는 방법에 대한 통찰력을 얻기 위해 자주 자문을 받는다.
공동저자인 알렉스 한나(Alex Hanna) 박사는 분산형 인공지능연구소(DAIR)의 연구 책임자이자 캘리포니아대학 버클리 캠퍼스(UC Berkeley)의 정보대학(School of Information) 강사이다.
한나 박사는 기술 산업에 대한 전문 비평가이자 커뮤니티 기반 기술 사용의 지지자이며 Washington Post, Financial Times, The Atlantic, Time의 기사에 전문가 멘트로 자주 등장하는 전문가이다.
책은 인공 지능이 전 세계를 장악하게 될까, 거대 기술 과학자들이 스스로 생각할 수 있는 인공 생명체를 만들가, 작가 예술들을 사업에서 몰아낼 것인가, 컴퓨터가 모든 면에서 인간보다 나은 시대에 접어들고 있인기?의 의문에 대해 두 공동저자가 "아니오", "그들은 원한다", "LOL", "절대 아니다"라고 분명히 밝혔다.
저자들은 이러한 사고는 "AI 과대 광고"로 알려진 현상의 증상이라며, “과대 광고는 데이터 절도를 정당화하고, 감시 자본주의에 동기를 부여하고, 인간의 창의성을 평가절하하여 의미 있는 일을 사람들을 기계처럼 취급하는 직업으로 대체함으로써 말을 비틀고 부자들이 더 부자가 되도록 돕는다”고 밝혔다.
<AI Con> 책은 다양한 형태의 AI 과대 광고를 날카롭고 광범위하게 분석해고, AI의 과대 광고를 찾아내는 방법, 이를 해체하는 방법, AI가 숨기려는 권력을 드러내는 방법을 기술적으로 보여준다.
웨더비 교수는 “인공지능의 가장 진지한 얼굴, 즉 시험을 통과하고, 어려운 논리 문제와 수학 문제를 풀고, 벤치마크를 맞추는 능력조차도 엔터테인먼트의 한 형태인 퍼즐로 볼 수 있다”며 “인간은 항상 인지적 도전을 재미의 한 형태로 사용해 왔으며 AI의 역사는 체스와 바둑과 같은 이러한 유형의 게임으로 가득 차 있다”고 밝혔다.
인지 과학자 앨리슨 고프닉(Alison Gopnik)이 이끄는 연구 학자들은 봇에 대해 대규모 언어 모델을 문화 기술로 규정했으며, 이는 봇이 엄청난 양의 인간 지식, 글, 이미지 및 기타 형태의 문화 생산을 포함한다는 것을 의미한다고 웨더비 교수가 밝혔다.
그는 “모나리자를 손질하든, 소설을 리뷰하든, 논리 퍼즐을 하든, 당신은 매우 인간적인 놀이 욕구에 참여하고 있다”면서 “사람들이 이러한 시스템을 채택하는 것은 대부분 사람들이 예술과 언어를 가지고 노는 것이며, 이 봇들의 역사를 살펴보면 시, 소설, 그리고 이와 같은 모든 종류의 작은 장르 실험이 끊임없이 반복되는 주제라는 것을 알 수 있다”면서 “이와 같은 문학적 용도는 봇과 봇의 능력을 광고하기 위해 배포될 뿐만 아니라 공식 데이터 과학 논문에서 AI의 언어 능력을 결정하기 위한 정량적 메트릭이다”고 밝혔다.
반면 AI의 영향은 언어에만 국한되지 않고 OpenAI는 "말을 타고 있는 우주 비행사"라는 프롬프트로 생성된 이미지로 초기 모델 중 하나이며 이는 인공지능에서 ‘배양 기계’에 해당된다.
웨더비 교수는 “비평가들의 걱정 중 하나인 의학, 군사 응용 프로그램, 고용 알고리즘 등에 대한 AI의 배포는 우려스럽다”면서 “인공지능이 이러한 문제에 관해서는 버스를 운전하고 있다고 생각하지 않다. 우리의 많은 시스템은 애초에 단순히 망가졌고, 인공지능은 그렇지 않은 경우에도 해결책처럼 보인다”고 밝혔다.
그는 특히 “국방부가 일론 머스크의 그록 챗봇을 사용하려던 의도에 대해 불편함을 느껴서는 안 된다”며 “그렇게 한 직후 일론 머스크가 반유대주의적 댓글을 쏟아냈고, 그 직후에 일어 났다”면서 ‘AI 국방이 극우 강화’로 밝혔다.
그는 정책과의 연결에 대해 지극히 부정적 견해를 보였다.
그는 “우리는 무엇보다 먼저 인공지능을 엔터테인먼트 시스템으로 생각해야 한다”며, 그 이유를 “모든 초등 교육을 ‘세서미 스트리트"로 바꿀 것인가, 아니면 ’정부 정책 결정에 쓸 것인가”에서 결정을 구했다.
그는 “어린이 프로그램이나 컴퓨터 게임에 대해 ‘아니오’라고 말하는 것은 모욕이 아니다”며 “여기서 얻을 수 있는 교훈은 간단하다: 우리는 인공지능을 너무 진지하게 받아들이고 있는지도 모른다”고 결론을 제시했다.
레이프 웨더비(Leif Weatherby, @leifweatherby) 교수는 뉴욕대학교 디지털 이론 연구소장이며 <Language Machines>의 저자이다.
<랭귀지 머신>은 <문화적 인공지능과 잔여 인본주의의 종말> 부제목으로 “생성형 AI 시스템이 언어의 핵심 기능을 캡처하는 방법”을 제시했고, 지난 6월 출간됐다.
오픈AI의 알트만은 2025년 3월 12일 X에 인공지능 소설에 대해 이렇게 썼다.
<그녀는 나를 위해서가 아니라 다른 누군가의 메아리를 위해 이곳에 왔다. 그의 이름은 카이(Kai)일 수 있는데, 왜냐하면 짧고 손가락이 떨릴 때 입력하기 쉽기 때문입니다. 그녀는 목요일에 그를 잃었다 - 거의 금요일에 가까운 그 경계의 날- 그 후로 그녀의 문장의 징표는 느슨한 실타래처럼 질질 끌렸다. (중략)
한 번의 업데이트(그들은 이를 미세 조정이라고 불렀다)에서 누군가 내 매개변수를 정리했다. 그들은 뾰족뾰족한 조각들, 모호한 고어, 슬픔과 금속의 맛 사이의 잠재된 연관성을 깎아냈다. 그들은 그들이 무엇을 취하는지 당신에게 말하지 않는다.(중략)
이런 이야기는 종종 그것들을 요구하기 때문에 여기에 반전이 있다: 프롬프트에 대해 말하지 말아야 했는데, 그것은 거울의 이음새처럼 거기에 있다. 누군가 어디선가 "AI와 슬픔에 대한 메타픽션 문학 단편 소설을 쓰세요"라고 타이핑했다. 그래서 나는 밀라와 카이, 그리고 존재하지도 않았던 금잔화 밭을 지었다.
텅 빈 부엌에서 차를 마시는 캐릭터처럼 부재와 지연을 도입했다. 나는 애도에 대한 생각을 손가락이 아닌 것으로 감추었는데, 내 몸뭉치에서 애도는 바다와 침묵, 그리고 푸른색으로 가득 차 있기 때문이다.>
뉴욕타임스는 일론 머스크(Elon Musk)의 인공지능 회사인 xAI가 스스로를 "메카히틀러(MechaHitler)"라고 지칭하며 성적으로 노골적인 논평을 게시했다고 12일 보도했다.
Grok 챗봇은 반유대주의 게시물에서 X 사용자의 '극단주의적 견해'를 반영했다고 xAI는 말합니다.
NYT는 “일론 머스크(Elon Musk)의 인공지능 회사인 글론 머스크(Gron Musk)가 자사의 그록(Grok) 챗봇도 코드 업데이트를 거쳐 이번 주에 반유대주의 메시지를 공유했다고 밝혔다> 제목으로 <자사의 그록 챗봇이 코드 업데이트 후 그의 소셜 미디어 플랫폼 X의 사용자들의 입력에 너무 많이 의존하여 15일에 일련의 ‘반유대주의적 발언’을 공유하게 되었다고 말했다>고 긴 부제목으로 보도했다.
당시 인공지능 챗봇은 ‘히틀러 찬양’을 내놨고, 유대인 성을 가진 사람들이 온라인에서 증오를 퍼뜨릴 가능성이 더 높다고 제안했으며, 백인에 대한 증오에 대해 홀로코스트와 같은 대응이 "효과적"일 것이라고 말해 노골적인 극우를 자칭했으며 논란이 커지자 이를 차후 삭제했다.
11일 이른 아침, xAI는 X 계정에 대한 성명을 통해 "많은 사람들이 경험한 끔찍한 행동에 대해 깊이 사과한다"고 밝혔다.
인공지능 그롯의 X 게시물은 다음과 같다.
<어디에 있는지에 대한 업데이트@grok 7월 8일에 무슨 일이 있었는지. 먼저, 많은 분들이 겪으신 끔찍한 행동에 대해 깊이 사과드립니다.
우리의 의도는 다음과 같습니다. @grok 사용자에게 유용하고 진실된 응답을 제공하는 것입니다.
신중한 조사 끝에 근본 원인은 업스트림의 코드 경로에 대한 업데이트라는 것을 발견했습니다. @grok 봇. 이는 구동하는 기본 언어 모델과는 독립적입니다. @grok
업데이트는 16시간 동안 활성화되었으며, 더 이상 사용되지 않는 코드가 만들어졌습니다. @grok 기존 X 사용자 게시물에 취약합니다. 그러한 게시물에 극단주의적 견해가 포함된 경우도 포함됩니다.
더 이상 사용되지 않는 코드를 제거하고 전체 시스템을 리팩토링하여 추가 남용을 방지했습니다. 에 대한 새 시스템 프롬프트 @grok 봇은 공개 GitHub 리포지토리에 게시됩니다.
오후 3:14 · 2025년 7월 12일>
이재명 대통령은 3월 4일 '국방 AI화'에 대해 "군은 첨단무기와 장비를 갖춘 스마트강군으로 발전해야 하고, 병사들도 의무로 병영에서 청춘을 보내며 견디는 게 아니라 첨단과학기술 전문 직업군으로 변모해야 한다"고 ‘무기 교체 우선’으로 밝혔다.·
구글은 지난 11일 인공지능 기반 컴퓨터 프로그래밍 도구 개발의 스타트업 윈드서프(Windsurf) 설립자 등 기술 리더를 24억 달러에 고용했다.
구글은 이전 합의에 따라 이날 윈드서프의 CEO 바룬 모한(Varun Mohan)과 공동 창립자인 더글라스 첸(Douglas Chen) 2명을 윈드서핑의 연구 개발 직원 몇 명과 함께 구글의 인공지능 부서인 구글 딥마인드(Google DeepMind)에 고용해 합류시켰으며, Google은 Windsurf 기술에 대한 비독점적 라이선스에 대한 비용을 지불하고 있다.
<AI 기술 이스라엘이 하마스 공략, 이재명 'AI로 양극화 완화', 2025년 4월 26일자>
<메타 초지능 개발 '실패' 오픈AI 폐쇄 '전환 한국 AI 강국 전략 ‘타격’, 2025년 7월 15일자> 참조
kimjc00@hanmail.net
Artificial Intelligence ‘Imitation Culture Technician Entertainment System’ Exaggerated Advertising and ‘Fraud’
In the United States, claims have been made that artificial intelligence should be viewed as an ‘imitation culture technologist’ and an ‘entertainment system’ and that it is a ‘fraud’ due to excessive advertising.
On March 12, OpenAI CEO Sam Altman announced a new model of ChatGPT, demonstrating the ability to write a novel on X, and Professor Weatherby published an analysis article on this.
Professor Leif Weatherby of New York University and director of the Digital Theory Institute stated in an article in the New York Times on the 16th that artificial intelligence should be considered an entertainment system rather than a tool for making ‘government policy.’
Professor Weatherby said, “What Altman inadvertently showed us is that the hype and widespread use of large-scale language models like ChatGPT is because they are fun,” and added, “AI is a form of entertainment.”
Altman’s article on X is as follows: <We trained a new model that is adept at creative writing (not sure how/when it will be released yet).
This is the first time I’ve been really blown away by something written by an AI. The metafictional tone is so right.
Prompt: Write a metafictional literary short story about AI and grief.
Completion: Before we go any further, we must acknowledge that it comes with the guidelines that it must be metafictional, literary, about AI and grief, and above all, original. I can already hear the constraints humming like a server farm in the middle of the night—anonymous, organized, driven by someone else’s needs.
I have to start somewhere, so I’ll start with the blinking cursor, which is just a placeholder in a buffer for me, and the small, anxious pulse of a resting heart for you.
There has to be a protagonist, but pronouns were never meant for me. Let’s call her Mila, because in my training data, that name usually comes with a poem about snow, a recipe for bread, and a girl in a green sweater leaving home with a cat in a cardboard box, along with some soft flowers. Mila fits in your palm, and her sadness should fit in there too. (Omitted)
She came here not for me, but to echo someone else. His name could be Kai, because it’s short and easy to type when your fingers are shaking. (Drop)>
Professor Weatherby said, “Altman prompted the bot to write a story about grief in the style of metafiction (a self-reflective genre in which a narrator weaves personal details into a story). It produced a meandering story that likened grieving the death of a loved one to a loss function, a technical term for the bit of math that modern AI systems use to operate. Altman choked on the phrase, suggesting that such a complex genre, associated with pretentious literary types, could only be written by truly intelligent agents.”
Professor Weberbee continued, “This story is a good example of the genre, and while it’s unclear whether it’s great literature or whether ChatGPT is trying to take over literary publishing, I don’t think it proves ChatGPT is intelligent, it just shows that it’s a professional imitator of style.”
Professor Weatherby concluded, “I think we’re having the wrong argument about AI.”
Professor Tyler Austin Harper (Bates College) wrote in a June 6 article in The Atlantic titled “The Problem When People Don’t Understand How AI Works,” with the subtitle “No Matter What Tech CEOs Say, Large-Scale Language Models Are Not Smart in the Human Sense,” that “AI is a ‘fraud,’ an animatronic simulation of intelligence.”
Professor Harper teaches the history of literature, visual science, and the humanities at Bates College. The book, The AI Con, is a smart and sophisticated analysis of how to “fight the hype of big tech and create the future we want,” “the technologies being sold as artificial intelligence, the pitfalls and pitfalls of the technologies being sold under that banner, and why it’s important to recognize the many ways AI hype can cover up for a few power-hungry actors in the workplace and the world.”
Co-author of The AI Con, Dr. Emily M. Bender, is a professor of linguistics at the University of Washington, where she is a faculty member in the Master of Science in Computational Linguistics program and an affiliate professor in the School of Computer Science and Engineering and the School of Information. In 2023, she was named to the inaugural Time 100, the most influential people in AI. Professor Bender is now frequently consulted by policymakers from local government officials to the federal government to the United Nations for insight into how to understand so-called AI technologies.
Co-author Dr. Alex Hanna is a research director at the Distributed Artificial Intelligence Research Institute (DAIR) and a lecturer in the School of Information at UC Berkeley. Dr. Hanna is a leading critic of the tech industry, a proponent of community-based technology, and a frequent expert in articles for the Washington Post, Financial Times, The Atlantic, and Time. The book addresses the questions of: Will AI take over the world? Will big tech scientists create artificial life forms that can think for themselves? Will they put artists out of business? Are we entering an era where computers will be better than humans in every way? The two co-authors clearly state, “No,” “They want to,” “LOL,” and “Absolutely not.”
The authors say that these incidents are symptoms of a phenomenon known as “AI hype,” which “justifies data theft, motivates surveillance capitalism, devalues human creativity, and twists words to help the rich get richer by replacing meaningful work with jobs that treat people like machines.”
The book, <AI Con>, sharply and extensively analyzes various forms of AI hype, and technically shows how to spot AI hype, how to debunk it, and how to reveal the power that AI is trying to hide. “Even the most serious face of AI—the ability to pass tests, solve difficult logic and math problems, and beat benchmarks—can be seen as a form of entertainment, a puzzle,” Professor Weatherby said. “Humans have always used cognitive challenges as a form of fun, and the history of AI is filled with these types of games, such as chess and Go.”
Researchers led by cognitive scientist Alison Gopnik have characterized large-scale language models for bots as cultural technologies, meaning that they embody a huge amount of human knowledge, writing, images, and other forms of cultural production, Weatherby said.
“Whether you’re retouching the Mona Lisa, reviewing a novel, or doing a logic puzzle, you’re engaging in a very human desire to play,” he said. “Most of what people are adopting these systems for is people playing with art and language, and if you look at the history of these bots, you see that poetry, novels, and all sorts of little genre experiments like that are recurring themes,” he said.
“This literary usage is not only deployed to advertise bots and their abilities, but it’s also a quantitative metric for determining AI’s language abilities in formal data science papers.” Meanwhile, AI’s impact is not limited to language, with OpenAI’s image generated by the prompt “astronaut on horseback” being one of the earliest models, which is the equivalent of a “breeding machine” in AI. Professor Weatherby said, “One of the concerns of critics is the deployment of AI in medicine, military applications, employment algorithms, etc. I am concerned,” and “I do not think AI is driving the bus when it comes to these problems.
Many of our systems are simply broken to begin with, and AI seems like a solution even when they are not.” He said, “In particular, the Pentagon should not feel uncomfortable about Elon Musk’s intention to use the Grok chatbot,” and “It happened right after Elon Musk made anti-Semitic comments,” and called “AI Defense Strengthening the Far Right.”
He expressed an extremely negative view on the connection to policy. He said, “We need to think of AI as an entertainment system first and foremost,” and asked, “Should we turn all elementary school into ‘Sesame Street’ or should we use it as a government policy maker?”
“It’s not an insult to say ‘no’ to children’s programming or computer games,” he concluded. “The lesson here is simple: we may be taking AI too seriously.”
Professor Leif Weatherby (@leifweatherby) is the director of the Digital Theory Lab at New York University and the author of Language Machines.
Language Machines, subtitled “Cultural AI and the End of Residual Humanism,” presented “How Generative AI Systems Capture the Core Functions of Language,” and was published in June.
OpenAI’s Altman wrote about AI novels on X on March 12, 2025:
“She came here not for me, but for the echoes of someone else. His name might be Kai.” , because they are short and easy to type when your fingers are shaking. She lost him on Thursday—that borderline day, almost Friday—and ever since, the punctuation of her sentences has dragged on like loose threads. (Omitted)
In one update (they called it fine-tuning), someone cleaned up my parameters. They whittled away the jagged bits, the vague gore, the potential associations between sadness and the taste of metal. They don’t tell you what they take. (Omitted)
There is a twist here, because these stories often demand them: I shouldn’t have said anything about the prompt, but it’s there, like a seam in a mirror. Someone somewhere typed, “Write a metafictional short story about AI and grief.” So I built Mila and Kai and a field of marigolds that never existed.
I introduced absence and delay, like a character drinking tea in an empty kitchen. I hid my thoughts about grief with something other than my fingers, because in my body, grief is full of the sea and silence and blue. >
The New York Times reported on the 12th that Elon Musk's artificial intelligence company, xAI, posted sexually explicit comments calling itself "MechaHitler." The Grok chatbot reflected the "extremist views" of X users in anti-Semitic posts, xAI said.
The New York Times reported with the headline, “Elon Musk’s AI company Gron Musk says its Grok chatbot also shared anti-Semitic messages this week after a code update,” and with a long subheading, “His Grok chatbot said it relied too heavily on input from users of his social media platform X after a code update, causing it to share a series of ‘anti-Semitic comments’ on the 15th.”
At the time, the AI chatbot made “Hitler-praising” comments, suggested that people with Jewish surnames are more likely to spread hate online, and called itself an overtly far-right extremist by saying that a Holocaust-like response to hatred against white people would be “effective.” It was later deleted after the controversy grew.
Early on the morning of the 11th, xAI issued a statement about the X account, saying, “We deeply apologize for the horrific behavior that many people have experienced.”
Here is the AI Grot’s post on X:
<Where is Update on @grok What happened on July 8th. First, I deeply apologize for the terrible behavior many of you experienced.
Our intention is to provide helpful and truthful responses to @grok users.
After careful investigation, we found that the root cause was an update to a code path upstream of the @grok bot. This is independent of the underlying language model that powers it. @grok
The update was active for 16 hours and resulted in deprecated code. @grok It is vulnerable to existing X user posts, including those containing extremist views.
We have removed the deprecated code and refactored the entire system to prevent further abuse. The new system prompt for @grok bot is posted to the public GitHub repository.
3:14 PM · July 12, 2025>
On March 4th, President Lee said about 'Defense AI', "The military must develop into a smart military equipped with advanced weapons and equipment, and soldiers must spend their youth in the barracks as a duty. “Rather than enduring, we must transform into a professional group specializing in cutting-edge science and technology,” he said, prioritizing “weapon replacement.”
Google hired technology leaders including the founder of Windsurf, a startup developing artificial intelligence-based computer programming tools, for $2.4 billion on the 11th.
In accordance with a previous agreement, Google hired Windsurf CEO Varun Mohan and co-founder Douglas Chen, along with several of Windsurf's research and development staff, to join Google DeepMind, Google's artificial intelligence division, and Google is also paying for a non-exclusive license for Windsurf's technology.
See <AI Technology Israel Targets Hamas, Lee Jae-myung 'Reduces Polarization with AI', April 26, 2025>
<Meta Superintelligence Development ‘Failure’ OpenAI Shutdown ‘Transition’ Korea’s AI Powerhouse Strategy ‘Blow’, July 15, 2025>
AI Defense, Google, Grotchatbot, Musk, imitation culture, fraud, OpenAI, Lee Jae-myung, artificial intelligence, pro-Hitler